Extracorporeal membrane oxygenation (ECMO) is increasingly being used in patients with noncardiac respiratory failure, with a 70% relative increase in the use of ECMO in patients with acute respiratory distress syndrome (ARDS) between 2008 and 2012 (1). In 2015, it was estimated there were around 300 centers worldwide capable of providing ECMO support (2). This number has continued to increase in the last few years despite a lack of high-quality evidence supporting the role of ECMO in respiratory failure.
Until recently, much of the data examining this question has been observational. One retrospective study reported on the outcomes of 68 patients with severe H1N1 influenza-associated ARDS who were treated with ECMO in ICUs across Australia and New Zealand (3). The mortality rate in these patients was 21% (95% CI, 11–30%); much lower than their predicted mortality, suggesting a possible beneficial effect of ECMO in this population (3). Another retrospective observational study from Saudi Arabia reported on patients with refractory hypoxemia caused by Middle East respiratory syndrome coronavirus infection and found lower in-hospital mortality in those receiving ECMO (65% vs 100%; p = 0.02) (4).
The Conventional Ventilatory Support Versus ECMO for Severe Adult Respiratory Failure (CESAR) trial examined the role of ECMO in patients with severe ARDS of any etiology using an open-label, concealed allocation, randomized controlled trial (RCT) design (5). CESAR randomized 180 patients with severe ARDS to either conventional management or referral to an ECMO-capable center with consideration for treatment. Only 75% of patients randomized to the intervention actually ended up receiving extracorporeal support. Although there was uncertainty about the effect on mortality (relative risk [RR], 0.75; 95% CI, 0.53–1.06), those in the intervention arm had higher rates of survival to 6 months without disability (RR, 0.69; 95% CI, 0.05–0.97) and higher quality-adjusted life years (0.03) (5). In addition to the fact ECMO was not delivered to a quarter of those randomized to the intervention arm, there was significant variation in ventilation practices in the conventional ventilation arm, some inconsistent with best practices in ARDS (e.g., tidal volumes > 6 mL/kg). CESAR has not had a significant impact on clinical practice perhaps due to the perceived limitations of the trial design.
The ECMO to Rescue Lung Injury in Severe ARDS (EOLIA) RCT (6) was designed to overcome some of the limitations that were inherent to CESAR. EOLIA enrolled 249 patients from 64 centers across France, Canada, and the United States with very severe ARDS as defined by the study authors despite optimal conventional mechanical ventilation to receive ECMO or conventional mechanical ventilation. As opposed to CESAR, 97.6% of patients who were randomized to the intervention arm in EOLIA actually received ECMO. Also, patients randomized to the conventional ventilation arm in EOLIA were strongly encouraged to receive all evidence-based ventilation strategies and adjuncts for ARDS; 90% were proned, whereas 100% received neuromuscular blocking agents (6). The sample size calculation for EOLIA (n = 331) assumed a 60% 60-day mortality in the conventional ventilation arm, and the trial was powered for a 20% absolute reduction in mortality with 80% power and an alpha level of 5%. Interim analysis, performed after 240 patients were enrolled, found the results had crossed the prespecified lower boundary of the stopping rule, and this led the Data Safety Monitoring Board to recommend stopping the trial early for futility in the primary outcome.
Results from EOLIA demonstrate an 11% absolute reduction in 60-day mortality with ECMO (RR, 0.76; 95% CI, 0.55–1.04). Although the point estimate suggests benefit, the 95% CI includes a possibility of harm. In absolute terms, based on these results, if we treated 1,000 severe ARDS patients with ECMO, our best guess is that we would prevent 109 deaths as compared with conventional ventilation, although this could be anywhere from 205 fewer deaths to 18 more deaths with ECMO. The authors highlight their “key secondary endpoint,” treatment failure, a composite of 60-day mortality and need for rescue ECMO (crossover) in the conventional ventilation arm. As part of EOLIA study protocols, patients randomized to the control arm were allowed to crossover to ECMO for refractory hypoxia defined as oxygen saturation less than 80% for greater than 6 hours despite use of available adjunctive therapies. A total of 35 patients (28%) ended up crossing over and received rescue ECMO and as such, those randomized to ECMO experienced less treatment failure at 60 days (RR, 0.62; 95% CI, 0.47–0.82). Of note, among those patients who crossed-over from conventional treatment to ECMO, the mortality rate was higher than those patients that did not crossover (57% vs 41%). There was no significant difference in adverse events between study arms.
How might this new trial influence clinical guidance? The most recent ARDS clinical practice guidelines, which include CESAR but published before EOLIA, made no recommendation for or against the use of ECMO in patients with severe ARDS and instead recommended for more research in the field (7). Has the question regarding the role of ECMO in severe ARDS been definitively answered? Although EOLIA has provided important data examining this clinical question, we propose that further research is needed before ECMO can be considered part of the standard treatment protocol in the management of severe ARDS. There are a number of reasons why we believe this to be the case.
- 1) Certainty in the treatment effect—EOLIA did not demonstrate a significant difference in the primary outcome of 60-day mortality with ECMO as compared with conventional ventilation. The point estimate suggested benefit; however, given the possibility of harm included in the 95% CI, we are less certain about this effect. There are a few other important factors to consider. This trial was stopped early for futility. It is possible that if all 331 patients had been enrolled, as initially planned, that a statistically significant benefit would be demonstrated, although it is also equally possible that the opposite, a trend further toward no effect, would become more evident. The risks of stopping an RCT early for any reason, including benefit or futility, are well documented, and the potential impact on conclusions can be significant (8). A recent systematic review reported on 207 RCTs that were stopped early for benefit prior to achieving their predefined sample size (9). Of these truncated trials, 102 had at least one subsequent RCT performed examining the same research question. Only 41.6% of the subsequent RCTs demonstrated the same statistically significant results as the trial that was stopped early for benefit and 12% of the subsequent trials demonstrated results that were in a different direction from the initial RCT (9).
- The treatment failure results are less imprecise but not without other limitations. The statistical significance of this outcome is likely driven by the crossover of patients randomized to conventional ventilation who received late or rescue ECMO. Although the decision to crossover was protocolized, it was still up to the bedside clinicians to operationalize this decision. The protocol states this was allowed in the context of refractory hypoxemia; however, at the time of crossover, patients had a median oxygen saturation of 77% (interquartile range, 74–87). This suggests a number of patients were crossed-over to ECMO despite saturations that were above the 80% threshold defined in the protocol. In an unblinded trial, this could be an important source of bias, over-estimating the effect of ECMO on treatment failure, and limiting our ability to draw credible conclusions from this outcome.
- 2) Not all patient-important outcomes have been considered—the primary results presented in EOLIA are 60-day mortality and treatment failure. Although some other secondary outcomes such as hospital/ICU length of stay, use of life-support modalities, and adverse events are captured, longer-term clinical outcomes are not reported. The critical care community is increasingly realizing that survival, although important, may not be the most important outcome measure given the potentially devastating consequences of prolonged critical illness and post-ICU syndrome (10). Those randomized to ECMO in EOLIA spent a mean of 15 days on therapy and close to a month in ICU. It is important to understand the long-term morbidity and mortality associated with this intervention. Follow-up documentation and reporting of 1-year mortality, quality of life, and return to independent living would be crucial before considering the wide-spread use of ECMO in this population.
- 3) Population—severe ARDS or rescue?—if ECMO is beneficial in ARDS, which population is most likely to benefit? Current practice, at most centers with ECMO capabilities, is to use ECMO as a rescue intervention for refractory hypoxia in patients perceived to have recoverable disease (reversible etiology, limited comorbidities) (11). Although EOLIA intended to study the effect of earlier ECMO in the setting of severe ARDS, the patients enrolled more closely resemble a very severe, perhaps even refractory, cohort. Over 50% of patients were proned, and almost all received neuromuscular blocking agents. Ongoing hypoxia despite optimizing ventilation, administering adjuncts, and proning qualifies for most as refractory hypoxia. As such, EOLIA may provide guidance on how to manage similar patients, but clinicians should be extremely careful about generalizing these results to those with less severe disease (even those that qualify as severe ARDS) or those in which all other treatments and optimization strategies have yet to be employed.
- 4) Are the results generalizable?—the EOLIA trial enrolled patients from selected centers which either had the capability to provide ECMO or had to have ‘extensive experience in treating patients with ARDS ‘and the ability to operationalize getting patients on ECMO and transferred to an ECMO center within 2 hours after randomization if allocated to the intervention arm. It remains unclear whether the results of this trial would be generalizable beyond this small subset of centers. It is possible that patients meeting criteria at a less experienced center may still benefit from transfer to a more experienced center, even for optimization of conventional ventilation, prior to being considered for ECMO therapy. A small proportion of patients who are initiated on ECMO will not recover, even over an extended period of time, and subsequently may need to be considered for long-term support or transplantation. Given only a limited of centers are able to provide this high level of care, the implications of more broadly offering ECMO, beyond these highly selected centers, requires further investigation before routinely including ECMO in treatment algorithms.
- 5) Impact on resources—the resource implications associated with ECMO therapy are substantial. One economic analysis, based on a mean duration of 9.5 days of ECMO therapy (a shorter duration than was seen in EOLIA) found totally costs of ECMO to be around $70,000 USD for each procedure (12). There are also considerations around safe patient transportation, staffing, ICU flow, and training. Given these costs and the ubiquitous deficiencies in healthcare spending, many centers, even in high-income nations, will have challenges initiating sustainable ECMO programs. This intervention would be almost impossible to operationalize in middle-income or low-income nations. As such, if ECMO does become part of recommended therapy in severe ARDS, this will have a significant impact on increasing health inequity. Before considering ECMO as standard of therapy, we require comprehensive and context-specific economic evaluations demonstrating cost-effective care. In addition, more research on how to overcome barriers and to facilitate implementation of ECMO use in developing countries is highly needed.
Based on the evidence to date, treatment in expert centers that can perform ECMO may have promise for certain patients with refractory ARDS. However, given the limitations described above, further study is required before this costly life-support modality becomes standard of care. Given the persisting uncertainty, the lack of long-term follow-up data and the resource/cost implications, any strong recommendation for or against ECMO therapy is not justified (13). We look forward to future research, which is both warranted and required, with the aim of better delineating the role of ECMO in those with ARDS. Further research priorities should include better defining the patient population with ARDS who is most likely to benefit from ECMO, what is the optimal time to initiate ECMO in a patient with ARDS, and long-term outcomes for those treated with ECMO.
1. Natt BS, Desai H, Singh N, et al. Extracorporeal membrane oxygenation
for ARDS: National trends in the United States 2008-2012. Respir Care 2016; 61:1293–1298
2. Butt W, MacLaren G. Extracorporeal membrane oxygenation
2016: An update. F1000Research 2016; 5
3. Davies A, Jones D, Bailey M, et al. Extracorporeal membrane oxygenation
for 2009 influenza A(H1N1) acute respiratory distress syndrome
. JAMA 2009; 302:1888–1895
4. Alshahrani MS, Sindi A, Alshamsi F, et al. Extracorporeal membrane oxygenation
for severe Middle East respiratory syndrome coronavirus. Ann Intensive Care 2018; 8:3
5. Peek GJ, Mugford M, Tiruvoipati R, et al; CESAR trial collaboration: Efficacy and economic assessment of conventional ventilatory support versus extracorporeal membrane oxygenation
for severe adult respiratory failure (CESAR): A multicentre randomised controlled trial. Lancet 2009; 374:1351–1363
6. Combes A, Hajage D, Capellier G, et al; EOLIA Trial Group, REVA, and ECMONet: Extracorporeal membrane oxygenation
for severe acute respiratory distress syndrome
. N Engl J Med 2018; 378:1965–1975
7. Fan E, Del Sorbo L, Goligher EC, et al; American Thoracic Society, European Society of Intensive Care Medicine, and Society of Critical Care Medicine: An official American Thoracic Society/European Society of Intensive Care Medicine/Society of Critical Care Medicine Clinical Practice guideline: Mechanical ventilation
in adult patients with acute respiratory distress syndrome
. Am J Respir Crit Care Med 2017; 195:1253–1263
8. Guyatt GH, Briel M, Glasziou P, et al. Problems of stopping trials early. BMJ 2012; 344:e3863
9. Murad MH, Guyatt GH, Domecq JP, et al. Randomized trials addressing a similar question are commonly published after a trial stopped early for benefit. J Clin Epidemiol 2017; 82:12–19
10. Gajic O, Ahmad SR, Wilson ME, et al. Outcomes of critical illness: What is meaningful? Curr Opin Crit Care 2018; 24:394–400
11. Alessandri F, Pugliese F, Ranieri VM. The role of rescue therapies in the treatment of severe ARDS. Respir Care 2018; 63:92–101
12. Mishra V, Svennevig JL, Bugge JF, et al. Cost of extracorporeal membrane oxygenation
: Evidence from the Rikshospitalet University Hospital, Oslo, Norway. Eur J Cardiothorac Surg 2010; 37:339–342
13. Atkins D, Best D, Briss PA, et al. Grading quality of evidence and strength of recommendations. BMJ 2004; 328:1490