Extracorporeal Membrane Oxygenation as Salvage in Pediatric Surgical Emergencies By Diana L. Farmer, Marc L. Cullen, Arvin I. Philippart, Fredrick E. Rector, and Michael D. Klein Detroit, Michigan • Extracorporeal membrane oxygenation (ECMO) has become an established therapy for acute neonatal and pediatric respiratory failure. On an institutional level, once an ECMO program is well established, ECMO can be viewed as a logical extension of critical care for multisystem organ failure. The question left unanswered is "Should anyone die without being offered ECMO?" The authors reviewed a 10-year clinical experience with ECMO and its application as salvage therapy in pediatric surgical emergencies. Eight patients with life-threatening multisystem organ failure, from diverse causes, were treated; the survival rate was 50%. All survivors were neurologically intact at the time of discharge. The success of ECMO, coupled with improvements in technique, apparatus, and expertise, has allowed application of ECMO as an invasive extension of intensive care to diverse patient groups. These results have encouraged the authors to expand their indications and to push the "envelope" in offering ECMO to critically ill infants and children with life-threatening organ failure.
Copyright © 1995by W,B, Saunders Company INDEX WORDS: Extracorporeal membrane oxygenation, emergency, pediatric.
EN EXTRACORPOREAL membrane oxyenation (ECMO) was introduced, in the early 1980s, it was used specifically for the treatment of full-term newborns dying of respiratory failure usually caused by diseases characterized by pulmonary artery hypertension, such as meconium aspiration syndrome and congenital diaphragmatic hernia. It was later extended to the support of children with low cardiac output after open heart surgery and then to nonneonates dying of respiratory failure who had primarily pulmonary parenchymal disease. ECMO has been available at the Children's Hospital of Michigan for 11 years and is routinely performed in the pediatric and neonatal intensive care units. Given its availability, we have come to view it as an extension of critical care and have used it in many nonstandard situations in an attempt to salvage critically ill pediatric surgical patients. This report describes our experience with ECMO in this clinical setting. MATERIALS AND METHODS
and six children with pediatric surgical disease who were dying of respiratory failure or multisystem organ failure, whom we attempted to salvage with ECMO. The age range of these 8 patients was 3 days to 27 months. The duration of the ECMO run ranged from 43 hours to 448 hours. Four patients survived and four died. Of the four patients who died, two died on ECMO (of complications of ECMO) and two died after decannulation (Table 1).
CASE REPORTS Patient l This boy was born at 36 weeks' gestation (birth weight, 3,600 g) with Beckwith-Wiedemann syndrome and a giant omphalocele. His Apgar scores were 4 and 9. He was intubated shortly after birth because of mild respiratory distress. He was transferred to Children's Hospital, where primary repair of the omphalocele was performed. Postoperatively he had progressive respiratory distress and poor perfusion, despite pressors. On postoperative day 5, he was placed on ECMO. His ventilator settings at that time were PIP (peak inspiratory pressure) of 46, rate of 65, and PEEP (positive end-expiratory pressure) of 4, with arterial blood gas (ABG) values of pH 7.37, Pco2 45, and Po2 38. His condition improved slowly on ECMO, and he was decannulated on ECMO day 7. He remained intubated for an additional 2 months and was discharged 1 month later, doing well on nasal cannula oxygen.
Patient 2 This girl was born at 32 week's gestation (birth weight, 2 kg) with gastroschisis, and presented shortly after birth to our hospital. Her Apgar scores were 7 and 9, and she required no oxygen. She had staged closure of the gastroschisis with a silo. Postoperatively, the patient required pressors, including dopamine, dobutamine, epinephrine, and nitroglycerin, to maintain adequate perfusion and urine output. Reduction of the bowel contents resulted in hypoxemia, and ultimate abdominal closure was completed 10 days after the initial procedure. She also did poorly after the second operative procedure, having hypoxemia and hypotension. On postoperative day 20, she was placed on ECMO. Ventilator settings at the time of ECMO were PIP of 36, rate of 50, and PEEP of 8, with ABG values of pH 7.58, Pco2 40, and Po 2 28. Her ECMO course was characterized by marginal perfusion and relative hypoxia. After 6 days on ECMO, the patient had exsanguinating hemorrhage from the neck. Exploration of a septic wound showed retraction of the central end of the carotid, which required mediastinal exploration for control. After prolonged eardiopulmonary resuscitation, the patient died.
The records of all children placed on ECMO at the Children's Hospital of Michigan were reviewed. Data were obtained from ECMO files, the ECMO data base, and the patients' medical records. From 1984 to 1994, 356 patients were treated with ECMO. Two hundred fifty-four were newborns (survival rate, 81%), 73 were treated for cardiac support (survival rate, 56%), and 28 were older children treated for primarily pulmonary parenchymal disease (survival rate, 43%). In the latter group were two neonates
From the Departments of Surgery, Wayne State University School of Medicine and the Children's Hospital of Michigan, Detroit, MI. Presented at the 25th Annual Meeting of the American Pediatric SurgicalAssociation, Tucson, Arizona, May 14-17, 1994. Address reprint requests to Marc L. Cullen, MD, Department of Surgery, Children's Hospital of Michigan, 3901 Beaubien Blvd, Detroit, M148201. Copyright © 1995 by W.B. Saunders Company 0022-3468 / 95/3002- 0035503.O0/0
Journal of Pediatric Surgery, Vo130, No 2 (February), 1995: pp 345-348
FARMER ET AL Table 1. Timing of ECMO in Relation to the Pediatric Surgical Procedure
Ecmo v Operation
No. of Hours on ECMO
1 2 3 4 5 6 7 8
Ornphalocele Gastroschisis Tracheal stenosis Pulmonary agenesis & tracheal stenosis Hirschsprung's disease Renal transplantation Incarcerated hernia Malrotation midgut volvulus
Newborn Newborn 6 mo 4 mn 1 mo 27 mo 2 wk 2 mo
5 days postoperatively 20 days postoperatively 2 days preoperatively 2 days preoperatively 3 days postoperatively 10 days postoperatively 4 days postoperatively 1 day postoperatively
165 176 488 43 118 260 212 275
Outcome Survived Died Died Died Died Survived Survived Survived
A 6-month-old boy was transferred to Children's Hospital after a workup for croup had shown distal tracheal stenosis. He remained hypercarbic (CO2 > 100) despite maximal ventilatory support. He was placed on ECMO for stabilization before undergoing a bronchogram and cardiac catheterization. Tracheal resection and a right bronchoplasty were performed on ECMO. He could not be weaned from ECMO at the completion of the operation. His postoperative ECMO course was complicated by mediastinal bleeding, which required bedside exploration. His condition eventually improved, and a postoperative bronchogram showed a widely patent anastomosis and an adequate left bronchus. ECMO flows were weaned, and the patient had a reasonable trial off, on 100% oxygen. Subsequent sepsis and ongoing bleeding led to adult respiratory distress syndrome (ARDS). He was decannulated on day 20, ventilatory failure progressed, and he died.
Three months after primary cadaveric renal transplantation for bilateral hypoplastic kidneys, this 2-year-old boy underwent operative correction of a transplant artery stenosis. Postoperatively he had significant intraabdominal bleeding, and progressive respiratory failure developed. Hypoxia hypercarbia, and barotrauma (bilateral pneumothoraces) were the indications for ECMO. Cytomegalovirus pneumonia was treated with Gancyclovir (Syntex Laboratories, Inc, Palo Alto, CA). His pulmonary status improved, but the ECMO run was complicated by severe intraabdominal bleeding, which resulted in an abdominal compartment syndrome that caused oliguria and poor ventilatory exchange. While on ECMO, he had exploration for control of bleeding and evacuation of the clot. Abdominal decompression with clot evacuation was required in the intensive care unit on several occasions. He was decannulated after 12 days on ECMO and was extubated 5 days later. He continued to improve and was discharged 1 month later, doing well.
Patient 4 This 4Wmonth-old boy, hospitalized from birth with right pulmonary agenesis and left bronchial obstruction, had undergone aortopexy elsewhere. Persistent problems with co2 retention and dying episodes led to transfer to Children's Hospital for assessment. After air transfer to our hospital, on mechanical ventilation, the child had persistent co2 retention unresponsive to highfrequency oscillation. He was placed on ECMO, and a bronchogram was performed. A left thoracotomy and division of the ligamentum arteriosum were performed on ECMO, with relief of expiratory obstruction. He was decannulated in the operating room after observing good ventilation, improvement in compliance, and normal ABG values. He did well for the initial 6 hours, but then arrested; he was returned to the operating room for exploration to rule out tamponade. No tamponade was found. The bronchus was patent, but deflation of the lung could not be maintained consistently. Ultimately, the patient died in the operating room. The postmortem examination showed acquired stenosis of several lobar and segmental bronchi.
Patient 5 This 1 month old had undergone a colostomy for Hirschsprung's disease and had enterocolitis complicated by septic shock, metabolic acidosis, and acute tubular necrosis. His pre-ECMO creatinine level was 1.9 mg/dL. Cardiovascular support included dopamine and dobutamine. Pre-ECMO ventilator settings were PIP of 50, PEEP of 8, and rate of 80, with ABG values of pH 7.48, Pco2 33, Po2 42. Because of progressive septic shock and hypoperfusion after the colostomy, the patient was placed emergently on ECMO. His condition stabilized, and he was decannulated on ECMO day 5. At the time of decannulation, he was doing well, requiring no pressor support; however, recurrence of sepsis resulted in death.
Patient 7 This 11-day-old boy presented with a 48-hour history of an incarcerated hernia, which was reduced in the emergency room. He was admitted to the hospital, and his condition deteriorated, leading to laparotomy. He was found to have necrotic bowel with a proximal ileal perforation. He underwent small bowel resection, with ileostomy and mucous fistula formation. Postoperatively he had gram-negative septic shock with severe multiorgan failure while on maximal support with pressors and volume replacement. His maximal ventilator settings pre-ECMO were PIP of 60, rate of 50, and PEEP of 10, with ABG values of pH 7.17, Pco2 59, and Po2 57. He was anuric and in disseminated intravascular coagulation. He was placed on ECMO in an attempt at salvage. He had a 10-day ECMO run, including hemodialysis and continuous arteriovenous hemofiltration. With the cardiovascular support of ECMO and appropriate antibiotic therapy, he was able to recover from his sepsis and was decannulated. He underwent ileostomy takedown with reanastomosis I month post-ECMO and was discharged doing well, with no obvious long-term sequelae.
Patient 8 This 2-month-old boy who had a small omphalocele, which was repaired at birth, presented with bilious vomiting. The workup showed malrotation with midgut volvulus. He had exploration, and 15 cm of bowel was resected. He did poorly postoperatively, having oliguria and sepsis. He had reexploration, and additional bowel was resected, a stoma was created, and a silo was placed to facilitate abdominal closure. He continued to be septic postoperatively, requiring dopamine, dobutamine, and epinephrine, which failed to improve his cardiovascular system and urine output. He
ECMO FOR SALVAGE
was placed on ECMO as salvage. Ventilator settings before ECMO were rate of 45, PIP of 42, and PEEP of 6, with ABG values of pH 7.34, Pco2 39, and Po2 90. Despite a 12-day run on ECMO, he remained anuric, with increasing blood urea nitrogen and creatinine levels, and was unable to be dialyzed. There was only marginal improvement in his pulmonary status. Therefore he was removed from ECMO and given to his mother, with the expectation that he would die. However, he did not die, and was decannulated approximately 12 hours after stopping the ECMO. His renal function returned, and he was extubated 48 days after decannulation and ultimately discharged. At the time of discharge, he required no pulmonary support and had normal renal function.
Using ECMO to resuscitate a critically ill surgical patient can provide end organs with improved perfusion of oxygenated blood in hopes that the underlying disease process will respond to treatment or time. It does not have a specific therapeutic effect, as it does in the newborn with pulmonary hypertension, in which it provides a pulmonary vasodilator (oxygen) and unloads the right side of the heart. In the 11 years since the inception of the ECMO program at Children's Hospital of Michigan, there have been 8 patients for whom ECMO was used unconventionally as cardiopulmonary Support in an
attempt to save their lives. None of the patients met extant "criteria" for placement on ECMO. These patients were believed to have potentially reversible disease (ie, no intracranial disease) but were about to die on conventional support. Two patients had airway anomalies and were unable to be ventilated, despite maximal support including high-frequency oscillation in the more recently treated patient. The other six patients were in the unrelenting downward spiral of unresponsive septic shock, which we see about once a year. These patients were on maximal ventilator settings for prolonged periods with still marginal ABG values. Perfusion could not be maintained (ie, blood pressure and urine output) despite massivevolume support, dobutamine, dopamine, and epinephrine. Four of the six patients who otherwise would probably have died are alive and well today. However, of course we are imperfect in our ability to predict death, even for the most hopeless-appearing patient. It is difficult to envision a randomized trial, given mixed diagnoses and use as resuscitation. Our experience with these patients may assist others in deciding to use ECMO as resuscitation in this manner.
Discus sion T. Weber (St Louis, MO): ECMO is a wellestablished treatment for newborn respiratory failure secondary to a relatively specific list of diagnoses. Attempts to extend the use of ECMO beyond these diagnoses into pre- and postoperative cardiac support, pediatric and adult respiratory failure, and as a bridge to heart or lung transplantation have been generally anecdotal, controversial at best, and have prompted numerous calls for prospective randomized studies to test the real utility and efficacy of ECMO in these and other clinical settings. This report by Dr Farmer and her colleagues is another anecdotal account of the use of this very expensive, laborintensive technique as salvage therapy for newborns and infants who have had, or will require, pediatric surgical procedures. Although the authors are to be congratulated on their results in saving 50% of these critically ill children, careful scrutiny of these patients shows interesting--if not controversial--aspects of their care. For example, two of the infants deteriorated after repair of abdominal wall defects and were placed on ECMO. One would have to question whether reoperation and decompression of the abdominal cavity would not have resulted in the same
outcome. Two infants with congenital tracheal stenosis were treated, and one of them died of ARDS, sepsis, and hemorrhage, presumably aggravated by the ECMO run. A child postrenal transplant surgery was placed on ECMO because of hemorrhage and its complications, and the hemorrhage predictably worsened. Finally, one of the septic infants was removed from ECMO to be allowed to die, but the child subsequently lived, bringing up the question of whether ECMO was even needed in this particular case. This series obviously generated a number of questions. Exactly what disorder are you treating? It is respiratory failure. ARDS. poor perfusion, low cardiac output, acidosis, or a combination of these? Are there specific indications for the use of ECMO in these patients, such as blood gas values, oxygenation index, [A-a]Do2 hemodynamic measurements. or urine output? Finally, can you now answer your own question, "should any infant be allowed to die without being offered ECMO?" J.-m. Laberge (Montreal, Quebec): I do not know if you have ever seen papers published on the cost of
dying. The cost of dying, I think, will go up in Michigan with this treatment, and I hope there are no HMO representatives in the room because you won't get referrals for dying patients anymore. K. Anderson (Los Angeles, CA): In your first patient, you performed primary repair of a giant omphalocele. Now that seems like an oxymoron to me. I. and maybe a number of other people in the audience, want to know how you can accomplish primary repair of a giant omphalocele. What do you do that the rest of us are not able to do? G. Haase (Denver, CO): At our center, the use of ECMO. specifically in the neonate, has decreased because of the dramatic effect of oscillatory ventilation and nitric oxide. I will limit the question to the two patients in your group who died because of their complicated airway situation. Were alternative methods of treatment employed? You briefly mentioned oscillation, but I did not hear anything about the length of time of therapy. Could you provide any details about other modalities that were attempted before ECMO was instituted for those patients with airway problems? J. Vacanti (Boston, MA): I have a question related to reimbursement for clinical care. This study, by anyone's definition, would be considered experimental. In your state are there increasing constraints for doing innovative therapy and obtaining reimbursement for these sorts of experimental interventions? W. Weintraub (Philadelphia. PA): I was very concerned about the choice of treatments offered almost each of these patients before giving ECMO. but the part that mortifies me is that to do an uncontrolled, clearly random, study that shows horrendous results and yet come to the conclusion that you need to do more ECMO. sooner, strikes me as not scientific. B. Harris (Boston. MA): Since you have convinced us that your group does not know who is going to live and who is going to die, I wonder if you p i a n t 0 continue using E C M O in such cases, and what your selection criteria will be. D.L. Farmer (response): The common denominator in this clinical report is that these were patients who
FARMER ET AL
we believed were about to die and for whom all other possibilities had been exhausted. Half of them are alive today. It is very difficult to determine when a patient i s going to die, and we clearly are imperfect in this assessment. To address the specific questions: The patients with abdominal wall defects were two of the early patients in the study. The patient with giant omphalocele had skin closure only, performed with adequate flaps. The patient with gastroschisis had a silo placed, which was opened, to determir!e that bowel necrosis was not the cause of the patient's problem. Regarding the patients with tracheal bronchial sten0sis, one did receive high-frequency oscillatory ventilation before ECMO, and the other case occurred before oscillatory ventilatio n was available. Many aggressive things are done in the ICU now. Patients get central lines. Patients get balloon pumps. Patients get operated on in the ICU. Ma W centers ligate PDAs in the ICU and perform other procedures. I think the boundaries become blurred: ECMO can simply be aggressive cardiopulmonary support. Dr Vacanti, I do not know what to say about reimbursement. We are in an environment where we can still try to do everything we can to save our patients who have normal neurological function and reversible disease processes. Assessing the cost of life is a complex ethica ! challenge. I would like to conclude by telling you about our last patient, who was a normal 3 year old who got chicken pox. Within 48 hours he had disseminated varicella and was in multiorgan system failure and a b o u t to die. He had intelligent parents. We discussed the fact that we really did not know whether w e could help him, but that we felt he would surely die soon and that we would try to place him on ECMO. He has had complete neurological, hemodynamic, and renal recovery. Such gratifying experiences make us think that despite the well-thought-out questions that have been raised, it is worth pursuing this aggressive course in certain cases.