More than 1,000 infants in respiratory failure have been treated successfully with extracorporeal membrane oxygenation (ECMO) since the first survivor in 1975. 1 ECMO therapy is used in term or near term infants who are failing conventional medical therapy and who have less than a 20% chance of survival without this new technology. The key factor ECMO supplies to these infants is lung rest. Ex tracorpo real cardiopulmonary support allows diseased lungs to heal without the iatrogenic effects of high oxygen and ventilator therapy.
Although ECMO is now being offered in more than 40 centers in the United States, its success has been relatively recent and the history of its development is important. Although the cardiopulmonary bypass concept was developed in the early 1950s, it did not become feasible until Clowes et al2 developed the first membrane oxygenator in 1956. The 1960s and 1970s were important years for ECMO with the development of the membrane lung and technical aspects of the procedure. Although the adult experience during this period was not positive, recent studies have shown that, with complete lung rest, the adult with acute respiratory disease may also be an appropriate candidate for ECMO therapy.3,4
In the late 1960s the technique was used in the premature infant with hyaline membrane disease under the theory that ECMO would act as the "artificial placenta." Unfortunately, the systemic hepahnization required for the procedure caused all infants to die from intracranial hemorrhage. It was not until Bartlett et al5 pioneered the treatment for term or near term infants in respiratory failure that ECMO entered its successful phase. These early data have been confirmed with further work by Bartlett and others, with present day survival rates over 80% in infants with a predicted survival of less than 20% (Table 1). It must be noted that infants treated today meet criteria developed from historical controls predicting an 80% to 100% mortality. The only randomized controlled study using a predicted 90% mortality entry criterion was published by Bartlctt et al6 at the University of Michigan in 1985. The randomized trial presently being conducted at The Children 's Hospital, Boston, is incomplete, with results not available at rhe time of rhis publication. Because of the mori' Hund nature of potential ECMO candidates, many centers have chosen to use historical controls to develop their criteria.
ECMO Central Registry Summary
PATIENT POPULATION AND CRITERIA FOR ECMO
Currently, rhe appropriate ECMO candidate is the term (ir near term infant (more than 34 weeks gestation) who tails maximal ventilator^ and medical support and who, by institutional criteria, has only a 20% chance or less of survival. The infant cannot have congenital heart disease, and since systemic heparinization is required, those with any major bleeding disorder (including a significant inrracranial hemorrhage) must be excluded. The infant's lung disease must be potentially reversible within a 10 to 14 day period. Severe chronic lung disease cannot be reversed wirhin the rime limits oí present-day ECMO therapy; therefore these infants should be excluded. Most infants who are candidates for ECMO have, as an underlying process, persistent pulmonary hypertension (PPHN), which results in right to left shunting through the foramen ovale and/or the ductus arteriosas.' This condition occurs in diseases such as meconium aspiration syndrome, sepsis, severe hyaline membrane disease, idiopathic PPHN, and congenital diaphragmatic hernia (see Table 1). Maximal medical therapy varies from institution to institution. It usually includes attempts at alkalization either hy venrilatory or metabolic means (sodium bicarbonate drip) and therapy with vasodilating drugs such as tolazoline. High-frequency ventilation may be appropriate before ECMO therapy is considered.
ECMO Entry Criteria
Figure. The accepted ECMO procedure is venoartenal bypass.
When all methods of conventional therapy have been exhausted, appropriate candidates for ECMO are those whose predicted survival is less than 20%. We have found the alveolar-arterial oxygen difference (AaDO7) over time to be the most accurate predictor for mortality in our institution. H The criteria used at Children's Hospital National Medical Center are lisred in Table 2. Others have found the Oxygen Indexg with time, or PO2 less than 50 mmHg over time, to he as predictive. Infants imist meet each institution's ECMO criteria because criteria assume the individual institution's methods of conventional therapy. Criteria should only be applied after maximal therapy and should be linked to time. A single AaDO2 does not predict death.
ECMO EQUIPMENT AND PROCEDURE
There is no standard "ECMO machine": the system must be designed starting from cardiopulmonary bypass equipment. The membrane lung used today is silicone (0.8 M2 silicone, Sci-Med Life Systems Inc, Minneapolis, MN ). The accepted ECMO procedure is venoarterial bypass (Figure). Veno-venous and single catheter techniques are in the experimental stages, but results are not available to date.
After assembly, the circuit is primed with an albumin/blood mixture. While this is being done the infant is anesthetized with fentanyl citrate (10 to 15 µ^g/kg); most infants are already paralyzed with pancuronium bromide. The catheters are placed in the right internal jugular vein and right common carotid artery, with the venous catheter advanced such that it rests in the right atri'im and the arterial catheter advanced to the entrance of the aortic arch. After completion of these two procedures, the catheters are connected to the bypass circuit, taking care that no air is introduced during this step. The infant is slowly placed on bypass by increasing the bypass flows over a 15- to 20-minute period until 60% to 70% of the cardiac output is going through the circuit. Ventilator settings are reduced as bypass is achieved, until room air at minimal ventilator pressure is reached. The infant is subsequently allowed to awaken and breathe independently. All vasoacrive drugs can be discontinued. The infant must be anticoagulated.
Oxygenation is achieved by circulating the infant's deoxygenated blood past the membrane lung, which removes CO2 and replaces the needed oxygen. Venoarterial ECMO can also supply cardiac support for the hypoxic myocardium which may have decreased function. For the first one to two days, 60% to 80% of the cardiac output must flow through the circuit in order to keep the patient's PO2S between 75 and 80 torr. As the infant's lungs improve, the ECMO flow can be decreased gradually. Once bypass flows reach 10% of the CO, the infant is ready to be taken off the circuit.
The average time on the circuit is five days. Lung compliance on ECMO is initially poor and then gradually improves as the baby is weaned off the ECMO circuit.10 Lung compliance can in fact be used to predict successful decannulation, with extubation usually possible within 48 hours of removal from the ECMO circuit. The decannulation procedure requires that the infant be briefly paralyzed and anesthetized. In Children's Hospital National Medical Center, supplemental oxygen has been discontinued an average of five to six days after decannulation.
Systemic heparinization accounts for most of the complications related to ECMO. Infants with congenital diaphragmatic hernia repairs, if placed on ECMO early (within 12 hours after surgery), can have significant bleeding requiring early removal from ECMO support. The most common cause of death in the ECMO patient is severe inrracranial hemorrhage (ICH). In our population, as with others, the incidence ot severe ICH is approximately 10% with an overall incidence of 20% to 25% if minor hemorrhages are included. 1,11 In the sett ing of pro- ECMO asphyxia and possible neurologic damage, systemic heparinization places these infants at risk for hemorrhage and thus heparin therapy must he monitored closely.
150 ECMO Patients at Children's Hospital National Medical Center
Inability to wean from the ECMO circuit can be caused by undiagmxsed cardiac disease (eg, total anomalous pulmonary venous return [TAPVR|); significant PDA; hypoplastic lungs; and mechanical factors such as loss of pump head occlusion. The most common cause is TAPVR. Lung compliance is an important diagnostic factor in these patients.10
FOLLOW-UP AND OUTCOME
The severity of the pre-ECMO illness mandates extensive tollow-up. Although in our series Iigation of the carotid artery has not appeared to be associated with early morbidity, there may he long term problems with the cerebral circulation as these children approach adulthood. Neither neuroimaging nor neurological findings indicate short term concerns for carotid Iigation, but the high incidence of posterior fossa hemorrhage (5/9 major hemorrhages) in infants less than 34 weeks gestation has given concern for venous Iigation or occlusion in this group. Infants should be given routine post-ECMO neurodevelopmental and medical evaluations. Our follow-up studies to date are extremely encouraging. With our first 150 patients treated we have an 84% survival rate (see Table 3). Our one-year follow-up to date includes 87 infants. Neurological exam is normal in 75%, with developmental studies using the Bay ley exam being normal in 63%. Of the 20% who are delayed, only one LS profoundly delayed (MDl and PDI < 50). The remaining infants are considered "suspect" and are being followed closely to rule out learning clitriculfies at school age.12 Infants in our population less than 34 weeks gestation have a significantly higher ICH rate (58%) than the rerni infanr (!chord, unpublished data). With these results, ECMO should not be extended to the infant less than 34 weeks until heparin-free circuits are available. Long term complications of EC-MO have not been determined.
The use ot ECMO has unquestionably been successful in a large number of term or near term infants in critical respiratory failure. The combined morbidity of ECMO and the disease processes appear to be acceptable. Although this is an exciting new therapy, it is labor-intensive and is at the stage ot development that ventilators were 15 years ago. The procedure requires a regional approach to insure appropriate subspecialty back-up and adequate patient numbers to maintain quality ot care. With improvements in the technique and potential development ot nonthrombogenic systems, ECMO may soon become available to a much larger population of infants who currently die of their respiratory diseases.
1. Toomasian JM, Snedecor SM, Gimell RG, et al: National experience with extracorporal membrane oxygenation (ECMO) for newborn respiratory failure: Data from 715 cases. Trans Am Soc Artif Organs. 1988; 34:140-147.
2. Clowes GHA Jr, Hopkins AL. Neville WE: An artificial lung dependent upon diffusion of oxygen and carbon dioxide through plastic membranes. J Thank Cardiovasc 1956; 32:630-637.
3. Zapol WM, Snider MT, Hill DJ, et al: Extracorporcal membrane oxygenation in severe acute respiratory failure, a randomized prospective study. JAMA 1979; 242:2193-2196.
4. Gattinoni L, Pesenti A, Masheroni D, et al: Low frequency positive-pressure ventilation with extracorpoteal CO2 removal in severe acute respiratory failure. JAMA 1986; 256:881-886.
5. Bartlett RH, Toomasian J, Roloff D, et al: Extracorporeal membrane oxygenation (ECMO) in neonatal respiratory failure (100 cases). Ann Surg 1986-240:236-244.
6. Bartlett RH, Roloff DW, Cornell RG, et al; Extracorporcal circulatory support m neonatal respiratory failure: A prospective randomized study. Pediatrics 1985; 76:479-487.
7. Getsony WM: Neonatal pulmonary hypertension: Pathophysiology, classification, and etiology. Clin Pennatol 1984; 11:517-524.
8. Beck R, Anderson K, Pearson GD, et al: Criteria for extracorporeal membrane oxygenation in a population of infants with PPHN. J Pediatr Surg 1986: 21:297-302.
9. Ortiz RM, Cilley RE, Bartlett RH: Entracorporeal membrane oxygenation in pediatric respiratory failure. Pediatr Clin North Am 1987; 43:39-46.
10. Lotze A. Taylor J. Short BL: The use of lung compliance as a parameter for improvement in luni; function in newboms with respiratory failure requiring ECMQ Cri CKIE Med 1987; 15:226-229.
11. Taylor GA, Fitz CR, Miller MK, et al: Intracranial abnormalities, in infants rteaied with extracorporea! membrane oxygenation: Imaging with US and CT. Radiology 1987; 165:675-678.
12. Glass P, Miller MK, Short BL: Morbidity in ECMO survivors: Neurudevelopmental outcome at one year of age. Fediamo, lo be published.
ECMO Central Registry Summary
ECMO Entry Criteria
150 ECMO Patients at Children's Hospital National Medical Center