Extracorporeal membrane oxygenation for neonatal respiratory failure

Extracorporeal membrane oxygenation for neonatal respiratory failure

J THORAC CARDIOVASC SURG 1989;97:706-14 Extracorporeal membrane oxygenation for neonatal respiratory failure A report of 50 cases From February 198...

822KB Sizes 0 Downloads 14 Views

J

THORAC CARDIOVASC SURG

1989;97:706-14

Extracorporeal membrane oxygenation for neonatal respiratory failure A report of 50 cases From February 1985 through June 1987, 50 newborn infants in whom maximal ventilator therapy failed (80% predicted mortality) were treated with extracorporeal membrane oxygenation (ECMO) according to the foUowing inclusion criteria: arterial oxygen tension less than 50 torr (alveolar-arterial oxygen gradient greater than 630 torr) for 2 hours or arterial oxygen t~ion less than 60 torr (alveolar-arterial oxygen gradient greater than 620 torr) for 8 hours. Criteria for excl~ion from ECMO therapy included birth weight less than 2000 gm, gestational age less than 35 weeks, presenceof intracraitial hemorrhage, presence of other major congenital anomalies including cyanotic heart disease, and high levels of ventilatory support for more than 7 days. Mean birth weight was 3.28 ± 0.56 kg, mean gestational age was 39.6 ± 1.7 weeks, and mean age at the start of ECMO was 48.6 ± 36.9 hours. Meconium aspiration, ~ually associated with persistent pulmonary hypertensien, was the most common cause of pulmonary failure (62 %). Mean pre-ECMO arterial oxygen tension during maximal ventilatory and pharmacologicsupport was 34.5 ± 14.5 torr. Mean ventilatorysupport immediately before the institution of ECMO was as foUows: peak inspiratory pressure 46.8 ± 9.9 cm H 20 , positive end-expiratory pressure 4.6 ± 1.6 cm H 20 , and intermittent mandatory ventilation rate 101.0 ± 22.7 breaths/min with aU patients receiving an inspired oxygen fraction of 1.0. Lung management to prevent pulmonaryatelectasis during ECMO consistedof moderate levels of positive end-expiratory pressure(mean10.3 ± 2.6 em H 20 , range 8 to 14 in 94 % of patients. Other mean ventilatorparameters during ECMO were as foUows: peak inspiratory pressure 22.8 ± 1.6 cm H 20 , intermittent mandatory ventilation rate 11.8 ± 2.9, and inspired oxygen fraction 0.21. The overalllong-term patient survival rate was 90 %. Mean valuesfor arterial blood gases and ventilator settings immediately after the discontinuation of ECMO were as foUows: oxygen tension 78.4 ± 22.1 torr, pH 7.39 ± 0.10, carbon dioxide tension 37.4 ± 10.7 torr, peak inspiratory pressure 25.2 ± 3.9 cm H 20 , positive end-expiratory pressure 5.6 ± 1.2 em H 20 , and intermittent mandatory ventilation rate 41.3 ± 12.6 with an inspiredoxygen fraction of 0.42 ± 0.17. Despite slightly higher levels of ventilator support (peak inspiratory pressure 46.8 versus 45.0 cm H 20 , not significant) mean pre-ECMO oxygen tension was significantly lower than that reported from the National ECMO Registry (34.5 versus 42.0 torr, p < 0.01). Mean positive end-expiratory pressure maintained during ECMO was significantly greater than that reported by the National ECMO Registry (10.3 versus 4.5 em H 20 , p < 0.01). Mean duration of ECMO support was significantly shorter than in the National ECMO Registry data base (84.3 versus119.0 hours, p < 0.01).Mean time to extubation and mean length of total

Michael G. Moront, MD (by invitation), Nevin M. Katz, MD (by invitation), Martin Keszler, MD (by invitation), Marc S. Visner, MD (by invitation), Gregory R. Hoy, MD (by invitation), John J. O'Connell, CCP (by invitation), Cynthia Cox, RN, NNP (by invitation), and Robert B. Wallace, MD, Washington, D.C.

From the Departments of Surgery and Pediatrics, Georgetown University School of Medicine, Washington, D.C.

Dr. Moront was supported by a Fellowship from The American Heart Association (National Capital Affiliate).

Read at the Sixty-eighth Annual Meeting of The American Association for Thoracic Surgery, Los Angeles, Calif., April 18-20, 1988.

Address for reprints: Nevin M. Katz, MD, Department of Surgery, Georgetown University School of Medicine, 3800 Reservoir Road, NW, Washington, D.C. 20007.

706

Volume 97 Number 5

May 1989

ECMO for neonatal respiratory failure

707

hospital stay were 4.9 ± 9.7 days and 28.6 ± 17.8 days, respectively, and compare favorably with the National Registry experience. We conclude that ECMO is a remarkably effective modality to reverse severe neonatal failure and that excellent survival can be achieved, despite the critical condition of these patients. The use of moderate levels of positive end-expiratory pressure during neonatal ECMO minimized pulmonary atelectasis and may importantly decrease the necessary duration of ECMO support.

DesPite the progress that has been made in neonatal intensive care, pulmonary failure remains an important cause of morbidity and mortality. In the past few years, extracorporeal membrane oxygenation (ECMO) has become a therapeutic alternative for neonates with life-threatening pulmonary failure.' To evaluate the effectiveness of the ECMO program at Georgetown University Hospital, we have reviewed the results of our first 50 patients.

Table I. ECMO patient characteristics (N = 50)

Birth weight (kg) Gestational age (wk) Apgar score at I min

Patients and methods

Apgar score at 5 min

Study group. Georgetown University Hospital initiated its clinical ECMO program in February 1985. From this date through August 1987, a total of 50 newborn infants, 31 boys (62%) and 19 girls (38%), were treated with ECMO in the Neonatal Intensive Care Unit for refractory cardiopulmonary failure, according to the following criteria:

Age when placed on ECMO (hr) Pre-ECMO P02 (torr)

Criteria for inclusion (despite maximal ventilatory and pharmacologic support) I. Acute deterioration: Arterial oxygen tension (Po 2) less than 50 mm Hg (alveolar-arterial oxygen gradient greater than 630) for more than 2 hours Arterial pH less than 7.15 for more than 2 hours 2. Failure to respond: Arterial P0 2 less than 60 mm Hg (alveolar-arterial oxygen gradient greater than 620) for more than 8 hours Criteria for exclusion Weight less than 2000 gm Gestational age less than 35 weeks Presence of intracranial hemorrhage Presence of congenital heart disease as a cause for respiratory failure . Presence of other major congenital anomalies More than 7 days of high ventilator support Patient characteristics are presented in Table I with comparison to means from the National ECMO Registry data base.s' The diagnoses of our patients are presented in Table II. Primary respiratory failure was the indication for ECMO support in all patients. Meconium aspiration, usually complicated by persistent pulmonary hypertension, was the most common cause of refractory pulmonary failure, occurring in 31 patients (62%). All candidates for ECMO therapy were managed by

Georgetown mean ± SD (range)

National Registry mean ± SD*

3.28 ± 0.56 (2.1-4.7) 39.6 ± 1.7 (37-43) 4.5 ± 7.2 (0-9) 6.2 ± 2.5 (1-9) 48.6 ± 36.9 (11-144) 34.5 ± 14.5t (11-75)

3.26 ± 0.62 39.1 ± 2.3 4.8 ± 2.5 6.6 ± 2.2 59. ± 53 42.0 ± 27.0

'National ECMO Registry data base.i' tp:o; 0.01 compared to National Registry mean.

attending neonatologists. The decision to institute ECMO support was made only after all conventional therapies were exhausted. These usually included a variety of ventilator manipulations, trials of volume loading, inotropic agents (dopamine, dobutamine), and tolazoline before the decision to implement ECMO was made. Two patients underwent highfrequency jet ventilation before ECMO. During maximal ventilatory and pharmacologic support, mean pre-ECMO arterial blood gases were as follows: oxygen tension (Po 2) 34.5 ± 14.5 mm Hg (range 11 to 75), pH 7.38 ± 0.21 (range 6.63 to 7.64), and carbon dioxide tension (Pco.) 41.5 ± 19.6 mm Hg (range 12 to 103). Mean ventilatory support immediately before the institution of ECMO consisted of a peak inspiratory pressure of 46.8 ± 10.0 cm H 20 , a positive endexpiratory pressure (PEEP) of 4.5 ± 2.4 em H 20 , and an intermittent mandatory ventilation rate of 101 ± 23 breaths/ min with the lungs of all patients being ventilated with 100% oxygen. After echocardiography and cranial ultrasonography to exclude congenital heart disease and intraventricular hemorrhage, the final decision to institute ECMO support was made by the attending neonatologist in the intensive care nursery. Informed consent was obtained in compliance with the guidelines of the Georgetown University Medical Center institutional review board. Technique. The ECMO perfusion circuit (Fig. 1) consisted of a 50 ml collapsible venous reservoir, a 5-inch roller pump, a silicone membrane oxygenator, and a heat exchanger, all

708

The Journal of Thoracic and Cardiovascular

Moront et al.

Surgery

t RESERVOIR WITH MICROSWITCH

ROLJ:ER PUMP 0,

~

~

COMPRESSED AIR

Y:;l I====!I

BLENDER

GAS FLOWMETER

Fig. 1. Diagram showing a neonate undergoing ECMO therapy and the components of the circuit.

Table II. ECMO patient diagnoses (N = 50) No.

Meconium aspiration Persistent pulmonary hypertension Respiratory distress syndrome Blood aspiration syndrome Congenital diaphragmatic hernia Congenital pneumonia/sepsis Pulmonary hypoplasia

31 4 4 4

Total

50

3 3 1

%

62 8 8 8 6 6 2

100

connected by \4-inch polyethylene tubing. Venous blood was drained by gravity to the venous reservoir (SciMed Life Systems Inc., Minneapolis, Minn). From the venous reservoir, blood was circulated by a 5-inch roller pump capable of accurate low flow control (Picker International Ossining, N.Y., or COBE Laboratories, Lakewood, Colo.) through a 0.8 m 2 membrane oxygenator (SciMed) to a pediatric heat exchanger (Dideco-Electromedics, Inc., Englewood, Colo). Blood exited from the heat exchanger and was returned to the patient via the arterial line (Bentley Labs Bypass 65, Baxter Healthcare Corporation, Bentley Laboratories, Inc., Irvine, Calif.). The blood pump was servo-controlled by a pressure switch against the venous reservoir, which stopped the pump if venous return was inadequate. In addition, an ultrasonic air

detector capable of sensing 0.3 mI of air at a blood flow of 300 ml/rnin was located around the arterial line distal to the membrane oxygenator. Both the venous servo-regulator and the ultrasonic air detector stopped the pump and sounded audiovisual alarms when activated. Ventilation gas to the membrane oxygenator was controlled by precision flowmeters (Sarns Inc., Ann Arbor, Mich.) and by a Sechrist air-oxygen gas blender (Sechrist Industries Inc., Anaheim, Calif.). ECMO cannulation was performed in the neonatal intensive care unit on a standard radiant warmer bed. Anesthesia and paralysis were achieved with local lidocaine (1%) and intravenous morphine sulfate (0.3 mg/kg) and pancuronium (0.2 rug/kg). The patient's head and neck were extended to the left, with the aid of a shoulder roll. The right internal jugular vein and right common carotid artery were exposed through an oblique incision centered on the anterior border of the sternocleidomastoid muscle. Meticulous hemostasis was maintained with an electrocautery and fine sutures. The patient was systemically heparinized (100 to 200 U/kg) while the cannulas were measured and prepared. Pediatric chest tubes, lOF, l2F, and 16F, were used for cannulation early in the series. In the last 40 patients, Ele-Cath cannulas (Electro-Catheter Corporation, Rahway, N.J.) were used. These cannulas were specifically designed for neonatal ECMO and appear to have better flow characteristics than pediatric chest tubes. Either 12F or 14F catheters were most commonly used for venous cannulation, although two neonates were cannulated with 16F venous cannulas. Arterial cannulation was accomplished with 8F, lOF, or 12F cannulas.

Volume 97 Number 5 May 1989

ECMO for neonatal respiratory failure 709

Table ill. Arterial blood gases before, during, and after ECMO support (N = 50) Po, (mm Hg) pH Pco, (mm Hg)

Before ECMO

1 hr on ECMO

6 hron ECMO

1 hr after ECMO

34.5 ± 14.5 7.38 ± 0.21 41.5 ± 19.6

180.4 ± 107.1 7.52 ± 0.11 26.5 ± 5.5

97.0 ± 35.1 7.52 ± 0.07 30.9 ± 4.76

78.4 ± 22.1 7.39 ± 0.10 37.4 ± 10.7

Data are expressed as mean ± standard deviation.

Table IV. Ventilator support before, during, and after ECMO (N = 50) Parameter PIP (em H,O) PEEP (em H,O) Ventilation rate (breaths/min)

Fio,

Before ECMO 46.8 4.6 101.0 1.0

± ± ± ±

9.9 1.6 22.7 0.0

During ECMO 22.8 10.3 11.8 0.21

± ± ± ±

1.6 2.62 2.9 0.0

1 hr after ECMO 25.2 5.7 41.3 0.42

± ± ± ±

3.9 1.2 12.6 0.17

Data are expressed as mean ± standard deviation. PIP, Peak inspiratory pressure; PEEP, positive end-expiratory pressure; Fio, inspired oxygen fraction.

Once an adequate activated clotting time of 300 to 500 seconds (normal 90 to 120) had been obtained, cannulation was undertaken. The distal common carotid artery was ligated, and after either a suture or vascular clamp had been applied for proximal control, an arteriotomy was made. The arterial cannula was gently advanced into the transverse aortic arch, usually 3 to 4 em, with the distance marked on the cannula by a suture. This distance was determined from the patient's chest film by measuring from the incision site to the transverse aortic knob. The right internal jugular vein was then ligated, and venous cannulation was performed in a similar fashion. The venous cannula was usually advanced 7 to 8 em so that the tip of the cannula was positioned in the body of the right atrium, I to 2 em superior to the inferior vena caval orifice. The distance was again marked on the cannula by a suture after being determined from the patient's chest x-ray film by measuring from the incision site to 1 em above the inferior vena cava-atrial junction. Once the cannulas were doubly secured in proper position, the wound was reinspected for hemostasis. In our last 32 patients, topical fibrin glue was placed in the cannulation site wound before closure to ensure hemostasis during the entire period of ECMO support.' The cannulas were then connected to the ECMO circuit, and after de-airing of the cannulas and their connections, ECMO support was initiated. Initial bypass flows needed to provide adequate oxygenation were usually 100 to 150 nil/kg/min with an inspired oxygen fraction (Fio.) through the membrane oxygenator of 0.6 An oxygenator Fio, of 0.6 was sufficient to maintain patient venous saturation in the 75% to 80% range and arterial saturation greater than 98%. Once adequate bypass flows were established, ventilatory support was adjusted so that peak inspiratory pressures were lowered to 22 to 24 em H 20 , intermittent mandatory ventilation rate decreased to 10 breaths/min, and Fio, reduced to 0.21. In patients free of an active air leak, PEEP was increased to 8 to 14 em H 20 to prevent the deterioration of lung function commonly seen with conventional ventilator settings during the early stages of ECMO support.v" Inotropic support was generally weaned over the first hour of ECMO therapy. Dopamine (3 to 5

JIg/kg/min) was often continued to optimize urine output. Throughout the period of ECMO support, mean systemic arterial pressure was maintained at 45 to 60 mm Hg, hematocrit value at 45%, and urine output at 2 ml/kg/hr by infusion of either plasma or packed red blood cells. From the time of initial heparinization, the activated clotting time was determined every 30 to 60 minutes. A heparin infusion was started when the activated clotting time fell to less than 300 seconds. Subsequently, the activated clotting time was maintained at 200 to 250 seconds. Platelet counts were monitored every 4 to 6 hours and platelet transfusions were given for counts less than 50,000 mm', If active hemorrhage was occurring, platelet levels were maintained greater than 100,000 mm ', Total parenteral nutrition was begun on the first or second day of ECMO support. Total parenteral nutrition was administered directly into the circuit and the patient's electrolytes were checked every 6 to 8 hours. All patients received ampicillin and gentamicin before ECMO, and either methicillin or vancomycin was added at the time of cannulation. Blood for culture was drawn from the ECMO circuit every 24 hours. Cranial ultrasonography and chest roentgenograms were performed daily. The decisions to reduce ECMO support were based on improvement in serial arterial blood gases, daily chest x-ray films, and in some cases measurement of dynamic lung compliance. ECMO pump flows were gradually decreased if oxygen levels could be maintained at 70 torr or greater. Once ECMO pump flows were less than approximately 60 to 80 rnl/kg/min and the roentgenographic appearance of the chest had improved, ventilator Fio, was increased from 0.21 to 0.30. An increase in arterial P0 2 on these low, stable ECMO flows was often indicative that ECMO support could be weaned over the next 8 to 12 hours. Ventilator peak pressures were maintained in the range of 22 to 26 em H 20 while Fio, and intermittent mandatory ventilation rate were adjusted to 0.40 and 35 breaths/min, respectively, as weaning progressed. PEEP levels were decreased to 5 to 8 ern H 20 as ECMO support was slowly replaced by conventional ventilatory support. Once ECMO bypass flow was 20 to 30 ml/kg/min, flow

710

The Journal of Thoracic and Cardiovascular Surgery

Moront et al.

Table V. ECMO patient survival by diagnoses (N = 50) No.

No. of survivors *

%

Meconium aspiration Persistent pulmonary hypertension Respiratory distress syndrome Blood aspiration syndrome Congenital diaphragmatic hernia Congenital pneumonia/sepsis Pulmonary hypoplasia

31 4

31

100

3 3

75 75 100 33

3

4 I 2

1

1

100

Total

50

45

90

4 4

3

67

'Defined according to the Guidelines of the National ECMO Registry as any patient who survives more than 24 hours after extubation.

Table VI. ECMO hospital course

Duration of ECMO support (hr) Time to extubation (hr) Total length of hospital stay (days)

Georgetown mean ± SD

National Registry mean ± SD*

84.4 ± 50.4t

119.0 ± 69.0

112.8 ± 242.4 28.9 ± 18.3

139.2 ± 222.0 32.0 ± 36.0

'National ECMO Registry data base.i' tp:5 om compared to National Registry mean.

to the patient was discontinued. The volume in the circuit was recirculated. During this trial period off ECMO support, both the patient and the ECMO circuit remained heparinized. This period lasted for 4 hours. After a stable trial period without ECMO support, morphine sulfate and pancuronium were administered for anesthesia and paralysis, and the cannulas were removed. Both the artery and the vein were ligated. The same membrane oxygenator was used for the entire period of ECMO support. Statistical analysis. Statistical analysis was performed with Student's t test for differences between means. The National ECMO Data Registry defines ECMO mortality as deaths occurring during ECMO or during the post-ECMO period up to 24 hours after extubation,"

Results A summary of the changes in arterial blood gases and ventilatory support before, during, and after ECMO are presented in Tables III and IV. Mean total duration of ECMO was 84.36 ± 50.4 hours (range 17 to 253). Patients remained heparinized and cannulated for an additional trial period off ECMO that averaged 5.6 ± 2.3 hours. In four cases, the patient's condition deteriorated during this trial period off ECMO and support was reinstituted. The mean value of peak sustained ECMO flow was 0.44 ± 0.09 L/min. Survival results according to diagnoses are presented in Table V. All deaths occurred while the patients were

still intubated for mechanical ventilation. Two of 46 patients died within 18 hours of decannulation and three patients died 2 to 9 days after ECMO support was discontinued. Of the two patients who died during or early after ECMO, one had intracranial and massive occult intraabdominal bleeding and subsequently died. The second ECMO death occurred when severe pulmonary hemorrhage developed concomitantly with an intracerebral hemorrhage after 71 hours of ECMO. The patient died of pulmonary failure 18 hours after ECMO support was discontinued. Of the three patients who died 2 to 9 days after ECMO support, one patient had sudden exsanguinating intraabdominal and intrathoracic bleeding related to dehiscence of a diaphragmatic hernia repair. The second patient died 48 hours after conclusion of a lO-day period of ECMO support for diaphragmatic hernia, and at postmortem examination the lungs were found to be severely hypoplastic. The third death occurred 9 days after an emergent decannulation for intracerebral hemorrhage. Postmortem examination revealed severe hyaline membrane disease, acute renal tubular necrosis, and a large right parietal cerebral hemorrhage. To date there have been no late deaths. Mean duration of ventilatory support after discontinuation of ECMO was 112.8 ± 242.4 hours. In four patients high-frequency jet ventilation was required after ECMO support was discontinued. Two of these patients (described earlier) eventually died. Mean length of total hospital stay was 28.9 ± 18.3 days (Table VI). One patient briefly required supplemental oxygen after discharge. The limited follow-up data available suggest that normal growth and development have occurred in approximately 80% of our survivors. Patient-related and technical ECMO complications are presented in Tables VII and VIII, respectively. Mechanical ECMO complications did not result in important patient morbidity or mortality. Although patent ductus arteriosus was detected in more than 75%

Volume 97 Number 5

ECMO for neonatal respiratory failure

May 1989

Table VII. ECMO patient complications (N = 50) Excessive neck wound drainage (10 ml/hr) Hypertension (MAP 75 mm Hg) Intracranial hemorrhage Cannulation site wound infection Seizures Renal failure Hemothorax Intraabdominal hemorrhage Intraparenchymal pulmonary hemorrhage Pneumothorax Vocal cord paralysis

No.

%

8

16

7

14

7

14 12

6 6

2

12 10 6 6 4

2 2

4 4

5 3 3

MAP. Mean arterial pressure.

of our patients, no patients required surgical closure of the ductus either during or after ECMO support.

Discussion Our data, as well as data from other institutions.t->" indicate that ECMO in neonates is a remarkably effective technique to reverse severe, near-terminal pulmonary failure. Bartlett and colleagues's have reported a prospective, randomized study in which they concluded that ECMO is superior to conventional therapy. The randomization technique of this study has been criticized.' Data from the published series," I I as well as from the National ECMO Registry.s ' indicate that the overall patient survival rate is between 54% and 90% for clinical states that have been associated in the past with an 80% to 90% mortality rate. Our results of 90% overall survival compare favorably with recent reports. In 1986, Weber and associates'? reported a 68% survival rate in 22 patients treated with ECMO for respiratory failure. Bartlett and associates," in the largest series of 100 patients treated with ECMO for respiratory failure, reported a 72% overall survival rate with a 54% survival rate in the "early period and a 90% survival rate in the later phases of the experience. Redmond and colleagues," in reporting the experience from the Oschner Clinic, where 48 neonates were treated with ECMO for respiratory failure, had 39 survivors, an 81% overall survival rate. Trento, Griffith, and Hardesty? reported a 54% survival rate in 33 neonates treated for near-terminal respiratory failure. The criteria for entry into our ECMO program were adhered to in all cases. In fact, our concept of maximal conventional support was rigorous, so that most of our

Table VID. ECMO technical complications (N Pump failure Accidental arterial decannulation Microswitch malfunction Header tubing rupture Accidental excessive heparinization

711

= 50)

No.

%

2

4 4

2 I I I

2

2 2

No mechanical complications resulted in patient deaths.

patients had a P0 2 less than 40 torr for more than 4 hours before ECMO, with a peak inspiratory pressure in excess of 46 em H 20 . As a result, our patients had a significantly lower mean pre-ECMO P0 2 than that of the National ECMO Registry population 2• 3 (34.5 versus 42.0 torr, p :::; 0.01), despite a slightly higher level of ventilator support (mean peak inspiratory pressure 46.8 versus 45.0, NS*). Despite the extreme severity of their illness, our patients, most of whom were treated with our high PEEP protocol," had a mean duration of ECMO therapy of 84.4 hours, compared with the National ECMO Registry':" mean of 119.0 hours (p -< 0.01). Mean time to extubation and total length of hospital stay compare favorably with the National Registry experience (Table VI). These facts, taken together with our 90% overall survival rate, suggest that excellent survival can be achieved, even when ECMO is reserved for patients with extreme pulmonary failure. We do not support more liberal use of ECMO so long as ECMO necessitates ligation of the common carotid artery. Early in our experience, even after maximal conventional support was exhausted and the decision to proceed with ECMO support was made, cannulation would be deferred occasionally when a patient's P0 2 would transiently improve. In almost every instance in which ECMO was postponed the patient eventually required ECMO support, often on an emergency basis as the clinical status rapidly deteriorated. Accordingly, our present policy is that, once our neonatalogists make the decision that ECMO support is necessary, ECMO is initiated unless a marked improvement in arterial blood gasses is observed and persists while ventilatory support is rapidly weaned to less injurious levels. Management of the lungs during ECMO has received little attention to date. It is clear that, given sufficient time, a neonate's lungs will heal in the vast majority of cases when the damaging effects of vigorous mechanical ventilation and high levels of inspired oxygen are eliminated. During the first 24 to 48 hours of *NS = Not significant.

7 12

The Journal of Thoracic and Cardiovascular Surgery

Moront et al.

ECMO support, despite conventional ventilatory rest settings of 18 to 20 em H 20 for peak inspiratory pressure, 3 to 5 em H 20 for PEEP, and a rate of 10 breaths/min, lung function and chest x-ray appearance deteriorate markedly and then gradually improve. 5, 6 Discouraged by this initial deterioration in pulmonary function very early in our experience, we" modified our approach to lung management during ECMO. We hypothesized that a higher level of PEEP would better maintain alveolar volume and therefore lung compliance. Further, we believed that higher PEEP levels would also counteract the tendency toward pulmonary edema, which may contribute to the "whiteout" appearance of the lungs commonly seen in patients maintained on low PEEP. Our experience with 46 patients suggests that the use of moderately high levels of PEEP during bypass, in the range of 8 to 14 em H 20, can prevent this initial deterioration in lung function and thus shorten the amount of time required for lung recovery." In addition, patient safety is increased in two ways: First, total patient dependence on ECMO for gas exchange may be avoided; second, the length of exposure to possible ECMO complications is reduced. Patient safety has guided our selection of perfusion equipment. Our current circuit evolved after unacceptable mechanical complications occurred early in our clinical experience (Table VII). One clinical case of tubing header rupture and several failures in the laboratory led us to reevaluate the polyethylene tubing used in the ECMO circuit. Bypass 65 tubing (Baxter) was selected after testing tubing of various diameters and manufacture. We routinely used Bypass 65 tubing for 5 days without header rotation. Tubing wear is minimal and there have been no tubing ruptures. Our current pump (COBE) was selected after several models had been tested. Our previous pump failed twice during bypass, once because of blood penetration of the motor controller and once because of power cord damage caused by excessive tension. The current COBE pump demonstrates excellent low-flow characteristics and precision occlusion and has operated without failure. We believe the addition of an ultrasonic arterial air detector (COBE) and use of an arterial heat exchanger with good air retention characteristics reduce the chance of arterial air embolism. Concerning the technique of cannulation, we have found that the cannulas can be most expediously positioned by using the chest x-ray film to estimate the intravascular length of the cannula (see Patients and methods). Diffuse bleeding at cannulation sites is a widely recognized complication of ECMO therapy. 15

Use of topical fibrin glue has almost completely eliminated this problem in our patients." In our initial experience we had the impression that cannula size was an important determinant of the amount of ECMO flow that could be obtained. As we gained more experience, we appreciated that the venous cannula position is at least as important a factor, in that satisfactory flows could be obtained with a moderate-sized drainage cannula if it was well positioned; that is, a 12F venous cannula is capable of 0.40 L/min flow, which is generally adequate to support a patient weighing 3.5 kg. Although many institutions successfully employ modified pediatric chest tubes for cannulas, we prefer the Ele-Cath ECMO cannulas. In the future, neonatal ECMO can be refined by advances in cannulation and circuit components. Venavenous ECMO may be as effective as vena-arterial ECMO in selected patients. Thus the permanent interruption of the proximal right common carotid artery and prolonged perfusion of the arterial system may be avoided. The effectiveness of neonatal vena-venous ECMO was demonstrated clinically by Andrews and colleagues" in 1983. Recently, in a short-term study in our laboratory, we": 18 have investigated hemodynamics and gas exchange during ECMO performed with a single-cannula, tidal flow, vena-venous ECMO circuit. Ultimately, an ECMO system that would not require heparin would clearly be desirable, but design limitations in the oxygenator have yet to be overcome. We appreciate the assistance with statistical analysis of Susan Ahmed, PhD, the secretarial assistance of Ms. Virginia Lewis, and the artwork of Mr. Peter Stone. REFERENCES 1. Bartlett RH, Gazzaniga B, Huxtable RF, Schippers HC,

O'Connor MJ, Jefferies MR. Extracorporeal circulation (ECMO) in neonatal respiratory failure. J THORAC CARDIOVASC SURG

1977;74:827-33.

2. National ECMO Registry. University of Michigan, Ann Arbor, Michigan. April, 1987. 3. Toomasian JM, SnedecorSM, CornellRG, et al. National experience with extracorporeal membrane oxygenation for newborn respiratory failure. Trans Am Soc Arlif Intern Organs 1988;34: 140-7. 4. Moront MG, Katz NK, O'ConnellJ, Hoy GR. The useof fibrin glue at neonatal ECMO cannulation sites. Surg GynecolObstet 1988;166:358-9. 5. Taylor GA, Short BL, Kriesmer P. Extracorporeal membrane oxygenation: radiographic appearanceof the neonatal chest. Am J Radiol 1986;146:1257-60. 6. Taylor GA, Lotze A, Kapur S, et al. Diffuse pulmonary opacification in infants undergoing extracorporeal mem-

Volume 97 Number 5 May 1989

brane oxygenation: clinical and pathological correlation. Radiology 1986;161 :347-50. 7. Ware JH, Epstein MF. Extracorporeal circulation in neonatal respiratory failure: a prospective randomized study. Pediatrics, 1985;76:849-51. 8. Bartlett RH, Toomasian J, Roloff 0, et al. Extracorporeal membrane oxygenation (ECMO) in neonatal respiratory failure: 100 cases. Ann Surg 1986;204:236-45. 9. Trento A, Griffith BP, Hardesty RL. Extracorporeal membrane oxygenation: experience at the University of Pittsburgh. Ann Thorac Surg 1986;42:56-9. 10. Weber TR, Pennington DG, Connors R, et al. Extracorporeal membrane oxygenation for newborn respiratory failure. Ann Thorac Surg 1986;442:529-35. 11. Short BL, Pearson GO. Neonatal extracorporeal membrane oxygenation: a review. Intensive Care Med 1986;1:47-54. 12. Bartlett RH, Roloff OW, Cornell RG, et al. Extracorporeal circulation in neonatal respiratory failure: a prospective randomized study. Pediatrics 1985;76:479-87. 13. 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. 14. Keszler M, SivaSubramanian KN, Smith VA, et al: Pulmonary management during extracorporeal membrane oxygenation. J Crit Care [In press). 15. Sell LL, Cullen ML, Whittlesey GC, et al. Hemorrhagic complications during extracorporeal membrane oxygenation: prevention and treatment. J Pediatr Surg 1986; 21:1087-91. 16. Andrews AF, Klein MK, Toomasian JM, Roloff OW, Bartlett RH. Venovenous extracorporeal membrane oxygenation in neonates with respiratory failure. J Pediatr Surg 1983;18:339-46. 17. Keszler M, Moront MG, Cox C, Milewski M, Visner MS. Oxygen delivery with a tidal flow veno-venous system for extracorporeal membrane oxygenation. Clin Res [In press). 18. Moront MG, Keszler M, Analoui A, Cox C, Milewski M, Visner M. The effect of variable preload on left ventricular function in a single cannula system for extracorporeal membrane oxygenation. Surg Forum 1988;39:291-3.

Discussion Dr. John C. Callaghan (Edmonton, Alberta, Canada). I would like to congratulate Dr. Moront and his group. Sometimes to appreciate the magnificence of an achievement in 1988, it is worthwhile to go back 30 years to see the realization of a dream. I got involved as a young resident, approximately the age of Dr. Moront, in the period before the era of the neonatologist. We were faced with the same problem-a great number of neonates dying of prematurity because of pulmonary disease. While attending the Surgical/Medical Research Institute at

ECMO for neonatal respiratory failure

7I3

the University of Alberta, a requirement for cardiac surgeons in those days, I attempted to create an artificial placenta, which was a plastic chamber. We transferred unborn lambs into this so-called "womb with a view" and supported these animals with a circuitry somewhat similar to what Dr. Moront described. However, it took us 4 hours just to prepare for the experiment by the laying of membranes and the development of a manifold to deliver the gases to make the membrane oxygenator. Now the physician can go to the shelf and choose a membrane oxygenator that gives magnificent results, like those Dr. Moront has achieved. I would like to give credit to the engineers and engineering expertise in industry that have provided the tremendous ability for membrane oxygenation in such a usable and, as Dr. Moront has shown, successful manner. To look at this over the perspective of 30 years is tremendously exciting for me. Dr. Phillip G. Ashmore (Vancouver, B.C., Canada). It is interesting to have John Callaghan remind us of some of the memorable experiencesin the early days of ECMO. I think we should name John the official "ECMO archivist," if we could get him to serve in that role. I would like to ask Dr. Moront three questions. First, I think the most important point in applying ECMO is appropriate patient selection. Our hospital in Vancouver has a large neonatal service and treats all of the sick newborn infants that come from about a 3 million population. Our neonatologists tell us that they are surprised at the number of patients with meconium aspiration syndrome reported in the literature as requiring treatment with ECMO. Their assessment is that they have no more than about one or two patients a year in their nursery that do not do perfectly well with conventional treatment for meconium aspiration syndrome. I'd like to ask Dr. Moront to comment on that group, because it is the largest group in his series. Second, Dr. Moront, have you used any different pump techniques, such as the centrifugal pump? This device is said to make it unnecessary to use the bubble traps and shut-off valves that are needed when a roller pump is used. Finally, I would like to ask if your group has made any progress in developing ECMO for use in the patients that you now exclude, namely, the low-birth-weight patients of early gestational age. This is a large group that might benefit from ECMO, but because of the possibility of intracranial hemorrhage and other complications they are usually not considered suitable for ECMO therapy. Has any progress been made toward treating this group? Dr. Moront (Closing). I would like to thank the discussants for their comments. Dr. Ashmore, with regard to your questions, there are a number of centers that successfullytreat babies for meconium aspiration without ECMO support. In certain ECMO centers it would appear that the technique is used primarily to treat patients with congenital diaphragmatic hernia and rarely to treat meconium aspiration. Despite the fact that our neonatologists treat our patients with meconium aspiration very aggressively, a number of these babies would either die or have significant lung disease in the form of bronchopulmonary dysplasia without ECMO therapy. With regard to the pump techniques, we have not used centrifugal pumps in either neonates or adults for ECMO. However, we are aware that a few centers have used the

714

Moront et al.

centrifugal pumps without problems. The concerns that we have with centrifugal pumps are as follows. The low flow rates of 300 to 500 ml/rnin, which are usually used during neonatal ECMO, are in our experience difficult to control accurately. Also, the inertia that exists within the centrifugal pumps prevents prompt cessation of flow. The ability to stop flow precisely is required by the venous reservoir microswitch and the ultrasonic air detection system to ensure perfusion safety. We have in the laboratory and hope to institute very soon a program of single-cannula tidal flow veno-venous ECMO in neonates.

The Journal of Thoracic and Cardiovascular Surgery

Finally, in regard to the question of expanded use of ECMO in low birth weight neonates, namely those less than 35 weeks or under 2000 gm, the greatest obstacle remains the fact that almost all these children, probably 80% to 90%, will have intracerebral hemorrhage when heparinized. The major limitations in the circuit that remain to be resolved are within the oxygenator itself. To my knowledge no progress has been made toward the creation of an oxygenator not necessitating heparinization

Bound volumes available to subscribers Bound volumes of THE JOURNAL OFTHORACIC AND CARDIOVASCULAR SURGERY are available to subscribers (only) for the 1989 issues from the Publisher, at a cost of $55.00 ($72.00 international) for Vol. 97 (January-June) and Vol. 98 (July-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copies are shipped within 60 days after publication of the last issue of the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact The C.V. Mosby Company, Circulation Department, 11830 Westline Industrial Drive, St. Louis, Missouri 63146-3318, USA; phone (800) 325-4177, ext. 351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.