By Richard Kallet, MS, RRT, FCCM

Director of Quality Assurance, Respiratory Care Services, Department of Anesthesia, San Francisco General Hospital

Mr. Kallet reports he is a major stockholder in the the Asthma & Allergy Prevention Company, is a consultant for Getinge Group, and receives grant/research support from Nihon-Kohden.

SYNOPSIS: Treatment of very severe acute respiratory distress syndrome with venovenous extracorporeal membrane oxygenation vs. lung-protective ventilation with aggressive ancillary therapy use demonstrated only a trend toward improved 60-day mortality.

SOURCE: Combes A, Hajage D, Capellier G, et al. Extracorporeal membrane oxygenation for severe acute respiratory distress syndrome. N Engl J Med 2018;378:1965-1975.

The authors of this international, multicenter, prospective, randomized, controlled trial enrolled 249 subjects with very severe acute respiratory distress syndrome (ARDS) to receive either venovenous extracorporeal membrane oxygenation (ECMO) with lung-protective ventilation (LPV) or LPV using a higher positive end-expiratory pressure (PEEP) strategy. Both treatment arms apparently continued at least some ancillary therapies. Between intubation and randomization, approximately the same percentage of subjects received prone position, inhaled vasodilators, recruitment maneuvers, and neuromuscular blockade.

At randomization, there were no differences between treatments arms regarding time from intubation to randomization, pulmonary mechanics, ventilator settings, gas exchange, age, gender, ARDS etiology, presence of nonpulmonary organ dysfunction, or compromised immune status. Enrollment criteria included ARDS duration more than seven days and the presence of either profound hypoxemia or respiratory acidosis persisting for more than three to six hours (depending on severity). The authors of this trial used similar exclusion criteria as other large ARDS randomized, controlled trials.

Subjects in the ECMO arm were cannulated at 3.3 ± 2.8 hours of randomization, with an average therapeutic duration of 15 ± 13 days. Those managed in the LPV control arm continued to receive a low tidal volume (VT) and higher PEEP strategy with encouragement to continue ancillary therapies. LPV goals were well-maintained in both treatment arms. However, those in the ECMO arm achieved a greater reduction in mean VT (< 300 mL), plateau pressure (< 25 cm H2O), and driving pressure (< 14 cm H2O) over the first seven study days.

The intention-to-treat design resulted in an extraordinarily high crossover (28%) to ECMO therapy, occurring at 6.5 ± 9.7 days. Sixty-day mortality in crossover subjects was 57% compared to 41% in those not failing LPV. Overall, subjects in the ECMO arm exhibited a lower-trend 60-day mortality compared to controls (35% vs. 46%; P = 0.09), which was not statistically significant. The ECMO arm subjects also experienced significantly more days without requiring either prone positioning or renal replacement therapy. However, subjects in the ECMO arm demonstrated significantly higher incidences of bleeding requiring transfusion as well as thrombocytopenia.


What was vexing about this trial was its early termination (at the third interim analysis for crossing futility criteria) despite a clearly positive outcome trajectory. This represents one of several nettlesome issues with conducting a randomized, controlled trial in a particularly challenging study cohort. These issues were discussed elegantly in the accompanying editorials.1,2 Other problems included the high crossover and particularly slow enrollment rate (more than six years). The study was very well-designed and executed. Ethical considerations obviously and rightly dictated the need for crossover, which occurred at an appropriate juncture when acute cardiovascular failure became evident. The slow enrollment signified the earnest effort of investigators to test ECMO in precisely the appropriate patient population for whom this expensive and extraordinarily invasive therapy is indicated.

Notwithstanding the advances in medical science associated with evidenced-based medicine, it remains a human endeavor burdened with biases. Some of the most recent “negative” results in ARDS randomized, controlled trials have been dismissed outright by advocates, while simultaneously embraced by those inimical to a particular therapy (despite nuances in study design or execution that might preclude unequivocal acceptance). Thus, the initial hope that high-level evidence from large multicenter, randomized, controlled trials would engender a medical “Age of Reason” appears unfulfilled. Thus, a claim that this trial was underpowered downplays the fact that medical ethics dictates a priori boundaries, balancing competing needs to stop a trial early either for efficacy or to prevent unnecessary exposure to therapies carrying substantial risks. Moreover, the tacit assumption that strong statistical trends inevitably would result in efficacy “if only” more subjects were enrolled to increase study power represents perilous reasoning. It is plausible that with continued enrollment, the trend actually might reverse. One large randomized, controlled trial of which I am aware indicated clear therapeutic efficacy at the first interim analysis, which evaporated at the second interim analysis, only to reappear and cross efficacy boundaries at the third. Thus, trends and their magnitude can be deceiving. Further, the enrollment of approximately 42 subjects annually in more than 60 participating institutions raises ethical issues about expending limited economic and personnel resources on a therapy that, in a broader societal context, could be allocated for research that might benefit a more patients. It is improbable that ECMO will be abandoned in severe ARDS based on these results. The positive trend in mortality makes it reasonable to pursue ECMO as rescue therapy. Moreover, this trial reaffirmed the importance of minimizing parenchymal strain-stress, particularly in severe ARDS. ECMO’s ability to enhance carbon dioxide (CO2) excretion allowed a mean VT reduction that likely was ≤ 4 mL/kg and that also corresponded to plateau and driving pressures previously associated with improved outcomes. In this context, recent advances allowing extracorporeal CO2 removal during renal replacement therapy3 introduces the possibility to further reduce lung injury in severe ARDS in cases complicated by multiple organ dysfunction syndrome without further escalating the intensity of invasive therapy.


  1. Harrington D, Drazen JM. Learning from a trial stopped by a data safety monitoring board. N Engl J Med 2018;378:2031-2032.
  2. Hardin CC, Hibbert K. ECMO for severe ARDS. N Engl J Med 2018;378:2032-2034.
  3. Quintard JM, Barbot O, Thevenot F, et al. Partial extracorporeal carbon dioxide removal using a standard continuous renal replacement therapy device: A preliminary study. ASAIO J 2014;60:564-569.