In 1984 and 1985, we conducted a prospective randomized trial of extracorporeal life support (ECLS) versus conventional treatment in neonatal respiratory failure. This study was reported in Pediatrics in 1985.1 Two decades later, ECLS is the standard treatment for neonatal respiratory failure that is unresponsive to other methods of management. The editors of NeoReview have invited this commentary on the development and clinical implementation of extracorporeal life support in neonatal respiratory failure.
The heart-lung machine was invented by John Gibbon and first used clinically in 1954 to replace heart and lung function long enough to allow operations on the heart. The entire field of cardiac surgery resulted from this invention. However, the heart-lung machine itself caused blood damage and multiple-organ failure when used for more than an hour. The major cause of blood damage was direct exposure of the blood to oxygen gas. In the 1960s, some laboratories and medical device companies developed gas-exchange devices in which a silicone rubber membrane was interposed between the blood and the oxygen. This modification (and others) allowed the use of a heart-lung machine for days or weeks. In 1971, prolonged extracorporeal circulation was first successfully used to treat a young man with acute respiratory distress syndrome (ARDS) after trauma.2 Over the next few years, several other cases were reported, using prolonged ECLS to treat cardiac failure and respiratory failure from a variety of causes. Because the membrane type of artificial lung was the unique part of the device, the entire technology of prolonged ECLS came to be known by the acronym ECMO for extracorporeal membrane oxygenation. Of course, the technology involves much more than oxygenation, but that acronym has remained. In 1975, the National Institutes of Health (NIH) sponsored a prospective randomized multicenter trial of ECMO for ARDS in adult patients.3 Only 10% of both ECMO and control groups survived, effectively stopping research on ECMO for adult respiratory failure for the next 15 years.
Our work on prolonged extracorporeal circulation began in 1965 with efforts to improve the efficiency of membrane oxygenators. The primary motivation was the very high mortality following complex congenital heart operations. Low cardiac output, oliguria, acidosis, respiratory failure, and death occurred in 50% of children after complex cardiac operations. Encouraged by Dr. Robert Gross and supervised by Dr. Francis Moore, Phil Drinker (an engineer from MIT) and I designed, built, and tested membrane oxygenators in the laboratories of then Peter Bent Brigham Hospital and Harvard Medical School. We developed techniques to improve the efficiency of membrane oxygenators,4 but more importantly, we began to study the technology and physiology of prolonged extracorporeal circulation. In 1970, Dr. Allan Gazzaniga and I (two residents recently graduated from the Brigham Program) joined the small surgical faculty at the new medical school at the University of California, Irvine (Irvine). Dr. John Connolly was the chair of the department, and he recruited surgeons with training and certification in general, thoracic, vascular, and pediatric surgery. Al and I were responsible for the day-to-day surgical care at the Orange County Medical Center (later the UCI Medical Center). Dr. Connolly provided laboratory space, and we undertook the serious study of prolonged extracorporeal circulation, using sheep as the experimental model.5 Our research was first supported by the National Institutes of Health in 1971 and continues today. We instituted clinical trials with the ECMO apparatus, which we were using in the laboratory. Our first successful case was in a 2-year old patient with cardiopulmonary failure following a Mustard operation for transposition in 1972. In 1975, we used ECMO for support of a newborn infant with meconium aspiration and persistent fetal circulation. This child survived, and other successful newborn respiratory failure cases soon followed.6 The use of ECLS for neonatal respiratory failure had been attempted by John White7 and Bill Dorson8 in the past. Both used the umbilical vessels and had major problems with bleeding and device failure. Building on their experience and our own laboratory experiments, we used the jugular and carotid for vascular access and titrated anticoagulation to very low levels.
The American Society for Artificial Internal Organs is the primary venue for physicians, engineers, researchers, and practitioners who are studying the development and use of artificial organs. We reported our first clinical cases of ECMO for cardiopulmonary support in infancy at the meeting of ASAIO in April 1976. This report was published in the Transactions of ASAIO.9 Of the 13 infants, 9 were neonates with respiratory failure; 3 of these patients survived. By the next year, we had used ECMO for 16 infants with neonatal respiratory failure, 9 of whom improved and 6 of whom ultimately survived. This series was presented at the meeting of the American Association for Thoracic Surgery and published in the Journal of Thoracic and Cardiovascular Surgery.10 Although the artificial organs researchers and the thoracic surgeons received these reports with cautious optimism, the community of pediatrics and neonatologists was very skeptical. Robin Jefferies, a surgical resident, presented our results at the Society for Pediatric Research in 1977 and was met with skepticism and criticism.
As our clinical experience grew, pediatric and cardiac surgeons visited Irvine to learn the technique and review the patients. Neonatal ECMO programs were established at the Medical College of Virginia by Tom Krummel11 and at the University of Pittsburgh by Bob Hardesty and Bart Griffith.12
In 1980, I moved from Irvine to the University of Michigan, bringing with me the laboratory and research projects on extracorporeal circulation. In 1982, we reported 45 cases of neonatal respiratory failure with 55% survival.13 Other centers reported similar results. A few young neonatologists who were frustrated by the high mortality of conventional treatment for neonatal respiratory failure came to Ann Arbor to learn the technology and established the first ECMO programs. These neonatologists were publicly accused of “academic suicide” by one of the deans of neonatology, but they persisted on behalf of their patients. Many of these neonatologists have become prominent in academic neonatology and pediatrics. They included Larry Cook, Bill Kanto, Devn Cornish, Billie Short, Deiter Roloff, Bob Schumacher, Tom Green, Martin Keszler, Pearl O’Rourke, Ernesto Gangiatano, and others.
By 1984, we had improved the technology to the point that the survival rate was 90% for most cases of neonatal respiratory failure and 70% for hypoplasia associated with diaphragmatic hernia. We knew the technique well enough to propose a controlled randomized trial. We immediately encountered the logistical and ethical problems of conducting a randomized trial of a complex life support technique in which the end point is death. The NIH adult trial had addressed this issue by creating a category of “de jure” death, that is, patients in the control group defined as “will surely die” who could be crossed over to ECMO support. The criteria were arbitrary, and that was one of many flaws in the NIH adult trial. Therefore, we determined that death would be the end point, but we knew we could select patients with a 90% risk of dying with conventional treatment, and we knew that 90% of the ECMO patients would survive.
To attempt to minimize the ethical problem, we used a statistical method that had been described in the statistical literature but never used for a clinical trial. Dr. Richard Cornell, Chairman of Biostatistics at the University of Michigan, was our statistical collaborator. We used a method described by Wei and Durham14 in which the first patient is truly randomized and subsequent patients are randomized, but the treatment assignment is based on the outcome of all of the previous patients in the trial. If there is no difference between the two treatment arms, patients will accrue at the same rate in both arms. If there is a major difference between the two methods of treatment, the patients will accrue more rapidly in the more successful arm. This type of adaptive design was well suited for our study. The study design was called “randomized play the winner” by the original authors. We conducted two simultaneous studies, one in full-term infants and one in premature infants. Three patients were entered into the premature study and two died of intracranial bleeding. The premature study was stopped because of these adverse events. In the full-term infants, however, the first ECMO infant survived and, the next patient was a conventional treatment patient who died. The next ECMO patient survived, so the odds of being assigned to ECMO grew with each successive case. By the time 12 patients had entered into the full-term neonatal trial, Dr. Cornell called up to say that we had reached statistical significance with a p value of .00000001. There were 11 patients in the treatment group, all of whom survived, and one patient in the control group who died.
The report of this study was submitted to two journals that declined to publish it because of the unusual statistical method. It was accepted by Pediatrics. Several conversations with the editor, Jerry Luce, preceded acceptance for publication. This was to be the first publication of successful neonatal ECMO technology in the pediatric literature. Most neonatologists still considered ECMO a radical, unproven, and unnecessary technology. After all, the NIH adult ECMO trial had proven that the technique was unsuccessful. The thought of a prospective randomized trial proving that ECMO was better than conventional care for neonatal respiratory failure was worrisome to most neonatologists, including the reviewers for Pediatrics. The treatment of neonatal respiratory failure at the time emphasized hyperventilation with induced respiratory alkalosis to decrease pulmonary vascular resistance. This often entailed the use of high ventilator pressure. Jen Wung, neonatologist at Columbia Presbyterian Hospital, had been using low-pressure ventilation for neonatal respiratory failure and had prepared a report of his Phase I experience.15Pediatrics published our randomized trial adjacent to Dr. Wung’s paper, along with several invited commentaries all of which criticized the ECMO report. I responded to these criticisms with a letter to the editor which was never published.
The most articulate of the critics16 from the Boston Children’s Hospital sniffed that “a proper study has yet to be done.” Pearl O’Rourke of that institution did a second randomized trial, using a different type of an adaptive design, and demonstrated that survival was much better in the ECMO-treated patients.17 Meanwhile, many neonatal centers learned and implemented the technology. By 1990, every major children’s hospital had an ECMO team or a plan for triaging ECMO patients. In 1990, the NIH convened a workshop on diffusion of high-tech medicine from bench to clinic using neonatal ECMO as the prototype example.18 Stimulated by the neonatal experience, today ECMO is used for cardiac and pulmonary failure in all age groups.
The availability of ECMO as a rescue treatment facilitated the study of inhaled nitric oxide, high-frequency oscillation, and other methods of treatment. Today, many infants who would have been on ECMO in 1995 improve with simpler methods. However, using ECMO as “rescue” after other treatments fail results in unnecessary death and chronic lung and neurologic disease.19 Early use of ECMO (at 50% mortality risk) improves outcome without increasing cost.20 As technology improves, ECLS will be used much earlier in all patients, including newborns.21
One major import of this article was to stimulate discussion of the problems and methodology of randomized trials of life support systems. The UK neonatal ECMO trial resulted from this discussion.22 The methodology of that study defines a close-to-ideal design for randomized trials of life support in which the end point is death.
1. Bartlett RH, Roloff DW, Cornell RG, et al: Extracorporeal circulation in neonatal respiratory failure: A prospective randomized study. Pediatrics
4: 479–487, 1985.
2. Hill JD, O’Brien TG, Murray JJ et al: Extracorporeal oxygenation for acute post-traumatic respiratory failure (shock-lung syndrome): Use of the Bramson Membrane Lung. N Engl J Med
286: 629–634, 1972.
3. Zapol WM, Snider MT, Hill JD. et al: Extracorporeal membrane oxygenation in severe acute respiratory failure: A randomized prospective study. JAMA
242: 2193–2196, 1979.
4. Bartlett RH, Isherwood J, Moss RA, et al: A toroidal flow membrane oxygenator: Four day partial bypass in dogs. Surg Forum
20: 152–153, 1969.
5. Bartlett RH, Fong SW, Burns NE, Gazzaniga AB: Prolonged partial venoarterial bypass: Physiologic, biochemical and hematologic responses. Ann Surg
180: 850–856, 1974.
6. Bartlett RH, Gazzaniga AB, Jefferies R, et al: Extracorporeal membrane oxygenation (ECMO) cardiopulmonary support in infancy. Trans Am Soc Artif Intern Organ
22: 80–88, 1976.
7. White JJ, Andrews HG, Risemberg H, et al: Prolonged respiratory support in newborn infants with a membrane oxygenator. Surgery
70: 288–296, 1971.
8. Dorson WJ, Baker E, Cohen ML, et al: A perfusion system for infants. ASAIO Trans
15: 155, 1969.
9. Bartlett RH, Gazzaniga AB, Jefferies R, et al: Extracorporeal membrane oxygenation (ECMO) cardiopulmonary support in infancy. Trans Am Soc Artif Intern Organ
22: 80–88, 1976.
10. Bartlett RH, Gazzaniga AB, Huxtable RF, et al: Extracorporeal circulation (ECMO) in neonatal respiratory failure. J Thorac Cardiovasc Surg
74: 826–833, 1977.
11. Kirkpatrick BV, Krummel TM, Mueller DG, et al: Use of extracorporeal membrane oxygenation for respiratory failure in term infants. Pediatrics
72: 872–876, 1983.
12. Hardesty RL, Griffith BP, Debski RF, et al: Extracorporeal membrane oxygenation: Successful treatment of persistent fetal circulation following repair of congenital diaphragmatic hernia. J Thorac Cardiovasc Surg
81: 556–563, 1981.
13. Bartlett RH, Andrews AF, Toomasian JM, et al: Extracorporeal membrane oxygenation (ECMO) for newborn respiratory failure: 45 cases. Surgery
92: 425–433, 1982.
14. Wei LJ, Durham S: The randomized play-the-winner rule in medical trials. J. Am Stat Assoc
73: 840–843, 1978.
15. Wung JT, James LS, Kilchevsky E, et al: Management of infants with severe respiratory failure and persistence of the fetal circulation without hyperventilation. Pediatrics
76: 488–494, 1985.
16. Ware JH, Epstein MF: Extracorporeal circulation in respiratory failure [commentary]. Pediatrics
76: 849–851, 1985.
17. O’Rourke PP, Krone R, Vacanti J, et al: Extracorporeal membrane oxygenation and conventional medical therapy in neonates with persistent pulmonary hypertension of the newborn: A prospective randomized study. Pediatrics
84: 957–963, 1989.
18. Wright L (ed): Report of the Workshop on Diffusion of ECMO Technology.
NIH Publication No. 93-3399, 1993.
19. Campbell BT, Braun TM, Schumacher RE, et al: Impact of ECMO on neonatal mortality in Michigan (1980-1999). J Pediatr Surg
38: 290–295, 2003.
20. Schumacher RE, Roloff DW, Chapman R, et al: Extracorporeal membrane oxygenation in term newborns: A prospective cost-benefit analysis. ASAIO J
39: 873–879, 1993.
21. Bartlett RH: Extracorporeal circulation in 2050: A speculation. Perfusion
18: 207–209, 2003.
22. UK Collaborative ECMO Trial Group: UK collaborative randomized trial of neonatal extracorporeal membrane oxygenation. Lancet
348: 75–82, 1996.