Extracorporeal membrane oxygenation (ECMO) is being increasingly used in acute respiratory failure. It is employed as a rescue intervention when conventional measures including titration of PEEP and prone positioning fail to achieve the desired effect. Historically, two randomized controlled trials (RCTs) had failed to demonstrate efficacy; however, these studies were performed several decades ago, when ECMO techniques were less refined. The CESAR study, performed many years later, demonstrated a significant improvement in the primary outcome of death or disability at six months in ECMO treated-patients. This study used devices with roller pumps, in contrast to the centrifugal pumps that are currently the preferred technique.
Combes et al., in their RCT, compared ECMO with conventional care in severely hypoxic patients with acute respiratory distress syndrome (ARDS). Patients in the control group could crossover to ECMO in case of severe, refractory hypoxemia. They found no significant difference in the primary endpoint of 60-day mortality. The secondary endpoint was treatment failure at 60 days: mortality in the ECMO group, and mortality or crossover to ECMO in the control group. The secondary endpoint was significantly more favorable with ECMO.
It may not be prudent to reject ECMO therapy in acute respiratory failure based on the findings of this study. To begin with, the investigators assumed improved survival by 20% with ECMO. This was based on two previously published studies- the PEEP study by Mercat et al. and the CESAR trial. However, such a large effect size entailed a small sample size, increasing the likelihood of a type II error. Clearly, it would have been unethical to decline ECMO to patients who were dying of hypoxia; however, crossover from the control to the intervention arm also makes the results of the study difficult to interpret.
Thirty-five patients who developed refractory hypoxemia in the control group were crossed over to receive ECMO. At the time of crossover, the median P/F ratio was 51, and the median Sao2 was 77%. Many of these patients were on the verge of severe cardiovascular failure; nine patients suffered cardiac arrest prior to initiation of ECMO and seven underwent veno-arterial ECMO. None of these patients may be expected to survive with continued conventional care; however, the use of ECMO resulted in 60-day survival in 15 of 35 (43%) patients. Another weakness of this study is the large number (72%) of potentially eligible patients who were excluded; 166 (16%) were excluded because they were already on ECMO. Blinding is not feasible in a study of ECMO; however, investigator bias cannot be excluded.
Perhaps the inclusion criteria for ECMO initiation also need to be considered. In our practice, we would probably not consider ECMO in a patient with a P/F ratio of 80 for six hours, particularly if there is an improving trend. Nor would I be too keen with a PCO2 of 60 mm Hg and pH of 7.25 for 6 hours. Only 62% of patients who underwent ECMO were prone ventilated; in our practice, we would attempt prone ventilation almost always before we consider ECMO.
At the end of the day, it may well be difficult to definitely prove the efficacy of ECMO against conventional care in a randomized controlled trial. First, it may be unethical to withhold ECMO in patients who are dying of hypoxia; approval for such a trial may be denied by most ethics committees. Second, to generate a sufficient sample size to demonstrate a clear effect in patients with refractory hypoxemia may take an inordinately long period. This multicentric study took nearly six years to recruit 249 patients. Therefore, I feel, given the current level of evidence, it may be appropriate to initiate ECMO based more on clinical judgment and local feasibility.