| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2003;76:S2224-S2229
© 2003 The Society of Thoracic Surgeons
a Department of Surgery, Thomas Jefferson University, Philadelphia, Pennsylvania and Alfred I DuPont Hospital for Children, Wilmington, Delaware, USA
* Address reprint requests to Dr Wolfson, General Surgery, Alfred I. duPont Hospital for Children, 1600 Rockland Rd, Wilmington, DE 19899, USA
e-mail: p.wolfson{at}nemours.org
Presented at the symposium, "Gibbon & His Heart-Lung Machine: 50 Years & Beyond," Philadelphia, PA, May 2, 2003.
Abstract
Extracorporeal membrane oxygenation (ECMO) is the utilization of a modified heart-lung machine to provide temporary support for patients with severe respiratory or cardiac failure. In contrast to patients managed with traditional cardiopulmonary bypass, patients on ECMO undergo cannulation of relatively accessible blood vessels, are maintained at normal body temperature, and only require partial anticoagulation with heparin. Although first developed for use in adults, ECMO has been most successful in the treatment of newborn infants with life-threatening pulmonary failure. Since 1974, over 17,000 infants have received ECMO with a 78% survival rate. There is a 15%–20% incidence of neurodevelopmental disabilities among ECMO survivors.
In 1974 a desperately poor, illiterate peasant woman in Baja, Mexico, discovered that she was pregnant and made a fateful decision. Determined that her child would have a better life as a United States citizen she crossed the border and headed for Los Angeles. During the journey her membranes ruptured and she took the next exit off the freeway, finding herself at the Orange County Medical Center where her daughter was born. But something was wrong. During the delivery the child had aspirated a large quantity of meconium and developed a severe chemical pneumonia. Very soon the baby's lungs were so damaged that they were unable to provide her with enough oxygen even with the ventilator turned up to its highest settings. It became increasingly apparent that the infant would not survive [1].
Three years earlier in 1971, a San Francisco surgeon named Donald Hill had used a modified form of extracorporeal bypass called "extracorporeal membrane oxygenation" (ECMO) to treat an adult who was dying from acute respiratory insufficiency [2]. The patient survived. By the time the baby was born in Orange County some 150 adults with the most severe respiratory failure had been treated with this device. Although only 10% to 15% survived the successes were dramatic. But no one had ever thought of using ECMO for a newborn infant with respiratory failure.
When the baby's pO2 went down to 12 the situation was considered so hopeless that there was nothing to lose. Robert Bartlett, a thoracic surgeon who had been involved in developing the membrane lung, wheeled in a machine from the laboratory. It probably was not informed consent in a strict sense as they tried to explain what they were going to attempt, which had never before been done in an infant, to the bewildered mother through an interpreter, using a flashlight in order to not disturb other patients. The mother signed with an "X" and then took one long, last look at her daughter before disappearing, perhaps being equally scared of the baby's likely outcome and of her own arrest and deportation. After 3 days of bypass, Esperanza—which means "hope" and is the name the nurses gave to the baby—completely recovered [1, 2].
What is ECMO exactly? Extracorporeal membrane oxygenation is the use of a modified heart-lung machine to provide gas exchange for prolonged support of patients with severe but potentially reversible respiratory or cardiac failure. The term "extracorporeal life support" actually describes the process more accurately and is preferred [3] but "ECLS" must be more difficult to say and the acronym "ECMO" is still most often used. As in traditional cardiopulmonary bypass, flowing blood is exposed to gas with a high pO2 and a low pCO2, resulting in gas exchange across concentration gradients. But there are several key differences. First, the traditional heart-lung machine is itself a lethal instrument if used for more than a few hours as direct exposure of blood to the continuously renewed raw gas surfaces in the bubble oxygenator will denature plasma proteins, hemolyze blood cells, and force microbubbles into the blood. In contrast the membrane oxygenator interposes a semipermeable membrane of silicone rubber between the blood and the gas to avoid the direct blood-gas interface. Second, there is no large venous reservoir in the ECMO circuit, which does away with the most thrombogenic portion of the bypass machine. As a result ECMO patients require only partial anticoagulation therapy with heparin and that is much safer for the prolonged time they are on bypass. Third, cannulation is performed extrathoracically through relatively accessible vessels and is accomplished at the bedside. And fourth, the patient is maintained at normal body temperature.
Why does ECMO work? After all ECMO itself does nothing to heal the ailing lungs. One of the major problems in treating patients with severe respiratory failure is that the extremely high ventilatory settings required to keep these patients alive, from the toxic oxygen concentrations to the damaging pressures, will destroy the lungs over time. The key to ECMO's success lies in its ability to allow the lungs to rest while the patient's vital functions are safely supported. Once the patient is on ECMO the ventilator can be turned down to very low settings until the lungs recover. If any specific therapy is needed (namely antibiotics, pulmonary lavage, and so forth) it can be safely administered while on ECMO.
The ECMO circuit
Figure 1 is a schematic of the venoarterial (VA) ECMO circuit. Venous blood is drained from the right atrium through a cannula placed into the right internal jugular vein. A roller occlusive pump drives the blood through the tubing and into the membrane lung where the gas exchange occurs. The now "arterialized" blood is rewarmed to body temperature in the heat exchanger and is then returned to the patient's aortic arch through a catheter in the right common carotid artery. The speed at which the pump is set determines what proportion of the patient's cardiac output will be diverted into the circuit and is adjusted according to how much support the patient requires.
|
|
|
How did ECMO develop? After anecdotal success treating adults in the early 1970s the National Institutes of Health sponsored a multi-institutional prospective, randomized, controlled study [4]. Patients critically ill with acute respiratory failure from a variety of causes (for example bacterial and viral pneumonia, pulmonary emboli, shock lung) were randomly assigned to treatment with conventional medical therapy or ECMO. The trial was halted prematurely after only 90 patients had been entered, with a mortality rate of 90% in both groups. Although ECMO could support most of these patients for the short term, most developed irreversible pulmonary fibrosis from their primary disease process or the deleterious effects of the high ventilatory settings they had been on before ECMO and could never be weaned off bypass. The authors concluded that there was no benefit of ECMO in terms of survival [4] and as a consequence the use of ECMO in adults all but ceased.
The experience of ECMO in newborn infants could not have been more different [5]. After Esperanza's recovery Bob Bartlett continued to treat infants with ECMO who were dying of respiratory insufficiency. In spite of having a 90% predicted mortality 75% of these babies recovered [6, 7]. Why should there have been such a contrast from the experience with adults [1]? (1) The causes of acute respiratory failure in newborns are very different. Instead of parenchymal lung disease most of these infants have abnormalities due to immaturity, problems with their airways, and disorders of the pulmonary circulation, which are more readily reversible. (2) The neonatal lung may be inherently more capable of repair and regeneration. (3) Finally these newborn patients were started on ECMO much earlier in the course of their illness than were the adults, before irreversible fibrosis could occur.
Despite the impressive numbers many remained skeptical about the efficacy of ECMO in infants without a randomized trial [6], considering how the adult study had turned out. Bartlett was faced with a huge dilemma. Could he ethically perform an objective, convincing protocol if it meant denying patients a treatment which he believed to be lifesaving? As a solution the University of Michigan chose a novel study design called the "randomized play the winner" technique [7–9]. The design is such that even though patients are randomly assigned, if one treatment proves to be superior, the chance of assignment to that group increases over time and the probability of patients being assigned to the inferior treatment group decreases. Here is how it works. Imagine an urn containing one ECMO ball and one conventional medical therapy ball. The first patient has a 50-50 chance of drawing either one. If that patient is randomly assigned to ECMO and survives, a second ECMO ball is added; if the patient dies, a second conventional medical therapy ball is added. Similarly if the first patient receives conventional medical therapy and survives, a second conventional medical therapy ball is added whereas if the patient dies a second ECMO ball is added. This process continues until a sufficient number of patients have been treated so that statistical significance is achieved. The investigators also chose to obtain consent for the study only from those families of babies who were randomly assigned to receive ECMO. It was argued that consent is only needed for the new therapy as the patients in the conventional medical therapy group were given the best standard treatment commonly used and were therefore not research subjects. The study designers were concerned that families would become distraught over not receiving the new treatment and especially if things were not going well there would be tremendous pressure from parents to have their dying children cross over to the ECMO group [8, 10]. The first baby was randomly assigned to ECMO and survived; the second was randomly assigned to conventional medical therapy and died. Following the protocol there was now a 3:1 chance that the next baby would receive ECMO. And that is then what happened for the next 10 infants! A total of 11 newborns were treated with ECMO and survived and one received conventional medical therapy and died. At this time the investigators concluded that their results had reached statistical significance.
A storm of controversy followed with many neonatologists not convinced that a well-conducted study had been performed [11]. Four years went by and in 1989 a unique opportunity for another randomized trial presented itself at Boston Children's Hospital [12]. Physicians in their multidisciplinary intensive care unit had been using ECMO on congenital diaphragmatic hernia infants with success whereas pediatricians in their neonatal intensive care unit were skeptical and refused to try ECMO without a convincing study. The two groups agreed to cooperate in designing a randomized protocol with patients in each group receiving treatment in different intensive care units so that the staff could continue to deliver the care that they believed to be superior [10]. They again chose an adaptive design as in Barlett's study, also in order to limit the number of deaths if one treatment should prove to be superior to the other. The study had two phases. Phase 1 involved strict randomization and would occur until there were four deaths in either the ECMO or the conventional medical therapy group. In phase 2 all patients would now only receive the other treatment until four deaths occurred with this therapy or a statistically significant difference between the groups was achieved. Consent was also obtained only for patients assigned to the ECMO group. Phase 1 ended when there were four deaths in the conventional medical therapy group with six conventional medical therapy patients surviving; nine patients had received ECMO and all survived. In phase 2 all patients therefore received ECMO. By the end of the study 19 of 20 (97%) ECMO patients survived compared with 6 of 10 (60%) conventional medical therapy patients. The authors concluded that their trial proved that ECMO is efficacious for neonates critically ill with respiratory failure [12].
This study was also attacked. The "adaptive design" was criticized from both sides with many arguing that it was not good science without the subjects all being randomly assigned [10] and others claiming that it was unethical to randomize at all [13]. (It has been noted that maybe these two extreme views indicate that the strategy of the study actually represented the best possible compromise [10]). The policy of obtaining consent only from those patients randomized to the experimental group was even more harshly criticized with the National Institutes of Health taking the unusual action of publicly criticizing Harvard's Institutional Review Board and categorically rejecting their claim that the control patients were not research subjects [3, 14].
A lengthy and sometimes heated debate followed these studies about the nature of randomized trials in general and the utility of ECMO in particular. One scathing commentary was even titled "Neonatal ECMO: how not to assess novel technolgies" [15]. The more thoughtful discussions centered on the many problems inherent in designing randomized controlled trials for evaluating potentially lifesaving treatments that are rapidly evolving [10]. It was pointed out that there is possibly an unresolvable conflict for clinicians in their role as healer in which their commitment is to the well-being of their patients, and as investigator in which their goal is to promote the acquisition of knowledge. The Declaration of Helsinki by the World Medical Association is less ambiguous, stating "Concern for the interests of the subject must always prevail over the interests of science and society" [16]. Randomization may be more ethical when the underlying condition is not life-threatening or if the physicians involved are completely uncertain about the best treatment option. The problem is that by the time a high technology treatment is finally ready for a trial, the physicians skilled enough to deliver it usually have a strong bias as to its likely benefit. The alternative approach of using historical, retrospective data for the control group may also be problematic as it may not reflect the current state-of-the-art results. By the time the Harvard trial ended after 6 and a half years tremendous advances in the management of neonatal respiratory failure had resulted in a decrease in mortality from 80% to 10% in infants who had met ECMO criteria at the beginning of the study without the need for ECMO [10, 17].
In 1996 a third randomized study evaluating neonatal ECMO was performed, this time in the United Kingdom [18]. British investigators agreed that the prior studies demonstrating reductions in mortality using ECMO were not conclusive. There were five ECMO centers in the United Kingdom and all clinicians agreed to limit ECMO within the context of a completely randomized trial. Babies would only be transferred to one of the five ECMO centers if they were randomized to receive ECMO; otherwise they would receive conventional medical therapy from the non-ECMO center in which they were already located. This arrangement avoided the predicament of having infants side by side in the same nursery and being managed with different lifesaving treatments. All treating hospitals were considered able to provide similar state-of-the-art therapy short of ECMO, an essential condition for the results to be valid. It had been anticipated that 300 infants would need to be enrolled to establish meaningful results but the trial was halted when a clear answer emerged after 185 babies were treated. Survival was 70% for the 93 ECMO infants versus 41% for the 92 conventional medical therapy patients (p = 0.0005). The authors concluded that not only did their study leave little doubt that ECMO can reduce mortality in this population but also demonstrated that large, collaborative, randomized trials for life-threatening conditions can be performed. There is currently a similar randomized trial being conducted in the United Kingdom involving ECMO in adults with results expected in 2003 [3].
Management
There are now approximately 100 centers worldwide that provide neonatal ECMO, most of which are in the United States. The selection criteria for placing an infant on bypass have been reasonably standardized and include (1) gestational age no less than 34 weeks; (2) birth weight more than 2,000 g; (3) no active bleeding or uncorrectable coagulopathy; (4) no intracranial bleed except grade 1; (5) a reversible lung process; (6) no lethal congenital abnormalities; and (7) failure of optimal medical management. This last criterion is the most subjective and has changed over the years as additional modes of therapy have been advanced that can often obviate the need for ECMO. The specific respiratory variables considered indicative of a low likelihood of survival and therefore warrant placing a baby on ECMO vary among centers but often include (1) an alveolar-arterial oxygen gradient more than 600 mm Hg for several hours; (2) arterial pO2 consistently less than 40 mm Hg; or (3) an oxygenation index (OI) more than 40 mm Hg (OI = MAP x FIO2 x 100/PaO2).
Once the baby is on ECMO the positive pressure ventilator is lowered to minimal "idle" settings in order to protect the lungs from further injury. Postural drainage, chest physiotherapy, and tracheal suctioning are performed frequently. Heparin is administered to maintain the activated clotting time between 180 to 220 seconds and platelets are transfused as needed. Parenteral nutrition and broad spectrum antibiotics are infused directly into the circuit. An ultrasonogram of the head is obtained at least every other day to monitor for intracranial bleeding. The ECMO pump is initially set to divert approximately half the baby's cardiac output to the circuit with the rate then adjusted to achieve acceptable arterial blood gases. As the lungs improve the pump settings can be decreased and when lower than 10% of cardiac output, the infant is usually ready to come off bypass. The average ECMO run for nondiaphragmatic hernia babies is 5 days.
Data on every patient who has been placed on ECMO are maintained in a central registry of the Extracorporeal Life Support Organization (ELSO) [19, 20], with all participating ECMO centers supplying detailed demographic and clinical information. As of July 2002 a total of 17,333 newborn infants had been treated with ECMO for respiratory failure. The most common condition has been meconium aspiration followed by congenital diaphragmatic hernia, primary pulmonary hypertension, sepsis, and hyaline membrane disease. Although the predicted mortality was more than 80%, 78% survived to be discharged from the hospital. There are major differences in survival rates by diagnosis with 94% of meconium aspiration babies surviving compared with a rate of only 54% for infants with congenital diaphragmatic hernias.
The population of neonates treated with ECMO has changed considerably over time [21, 22] in large measure because of the recent dramatic development of less invasive modes of therapy for neonatal respiratory failure. High frequency oscillatory ventilation, exogenous surfactant, inhalational nitric oxide, and permissive hypercapnia (allowing patients to temporarily sustain high pCO2 and low pO2 levels in order to remain at lower, safer ventilator settings) have all contributed to a declining need for ECMO in many instances. The accessibility of ECMO actually facilitated the development of these other forms of treatment as novel approaches could be more safely attempted with ECMO available for potential rescue. The number of infants who have undergone ECMO per year has declined from a high of 1,517 in 1992 to 773 in 2001 [20]. Infants with hyaline membrane disease are especially less likely to require ECMO today—over a 10-year period the percentage of ECMO infants with hyaline membrane disease decreased from 15% to 4% [21].
Outcome
How do ECMO survivors fare? These infants are at increased risk for complications both as a consequence of ECMO itself and from the antecedent hypoxia and acidosis. The most significant complication on ECMO is abnormal bleeding due mostly to the heparin anticoagulation and also from thrombocytopenia that results from platelets aggregating in the circuit. Approximately 15% of neonates on ECMO sustain an intracranial hemorrhage or infarction [21, 23].
A number of investigators have analyzed the long-term outcome of these patients [24–29]. Neonatal ECMO survivors have a relatively high incidence of respiratory abnormalities initially with 15% requiring supplemental oxygen at 28 days of age and 25% having pneumonia by the age of 5 years (which is twice the rate of pneumonia in normal children) [28]. At 5 to 10 years of age, 15% to 20% of ECMO survivors have a significant neurodevelopmental disability (mental, motor, sensorineural, or seizure disorder) [24, 26, 27]. Compared with other high-risk populations of neonates (eg, infants with severe respiratory distress who did not require ECMO or extremely low birth weight infants) the incidence of major disabilities is similar, suggesting that these babies' underlying conditions were possibly more contributive to their later difficulties than was ECMO itself [24–26]. It is perhaps remarkable that the vast majority of children after ECMO do not have major neurologic problems, and for those with deficits the disabilities do not seem to progress over time [27].
Encouraged by the impressive success of ECMO for neonates with respiratory disease, prolonged extracorporeal bypass therapy has been extended to older children and again to adults for both respiratory and cardiac support. Results, although promising, are not as favorable as for respiratory failure in newborns. Survival for older children requiring ECMO for respiratory support is 63%, for children needing cardiac support it is 55%, and for adults it is 50% [20].
What is the cost effectiveness of neonatal ECMO? A 1993 analysis reported the total hospital charges per ECMO patient to be approximately $50,000 [30]. Using the methodology of Vats and associates [31] and assuming a predicted mortality of 80%, a survival rate of 80%, and a 70-year life-span for survivors it is estimated that the cost of neonatal ECMO is approximately $1,000 per life-year saved in 1993 dollars. This amount compares very favorably with the costs of other lifesaving therapies that have been considered cost effective such as adult renal transplantation ($16,300/life-year), heart transplantation ($26,900/life-year), liver transplantation ($43,500/life-year), and bone marrow transplantation ($62,500/life-year) [31]. Neonatal ECMO is a real bargain.
What happened to Esperanza, the first neonatal ECMO survivor? After her successful treatment she was quickly adopted. She is now 27 years old and has two children of her own [2].
References
This article has been cited by other articles:
|
|
 |
|
  W. C. Oliver Anticoagulation and Coagulation Management for ECMO Seminars in Cardiothoracic and Vascular Anesthesia, September 1, 2009; 13(3): 154 - 175. [Abstract] [PDF]
|
|
|
|
|
|
S. Crow, A. C. Fischer, and R. M. Schears Extracorporeal Life Support: Utilization, Cost, Controversy, and Ethics of Trying to Save Lives Seminars in Cardiothoracic and Vascular Anesthesia, September 1, 2009; 13(3): 183 - 191. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
