Discussion
To our knowledge, this is the first systematic review and meta-analysis of the use of veno venous ECMO in treating severe COVID 19 related ARDS. We identified 12 independent studies (18-29) that provided data on the outcome of ECMO in COVID 19 patients. The initial studies testing the efficacy of ECMO in COVID 19 were small and pessimistic. A study by Ñamendys-Silva SA (30) looked at pooled mortality and efficacy from initial Chinese reports (only 17 patients). It included 2 of our included studies (22,25) and found a mortality of 82.3% (14/17) with no overall benefit of ECMO (z=0.57, p=0.56). Henry and Lippi (31) shortly followed by another pooled analysis which included also 2 of our included studies (22,23) comparing ECMO (total of 17 patients) and conventional mechanical ventilation and again found no benefit for ECMO in severe COVID 19 with pooled odds of mortality in the ECMO group versus the conventional therapy group not significantly different (OR: 2.00, 95%CI: 0.49–8.16) with no observable heterogeneity (I2 = 0%, Cochran’s Q, p-value = 0.99).
We performed a subgroup single arm meta-analysis for studies including another arm for invasive mechanical ventilation for severe ARDS due to COVID 19, but all our ECMO studies had 5 or more patients (22-25) as per inclusion criteria and again found no benefit or harm for ECMO in treating severe ARDS in COVID 19. The pooled odds of mortality in the ECMO group versus the conventional therapy group were not significantly different (p=0.273, 95%CI: 0.06–1.111). The mortality was high in both groups (87.5% vs 69.2%). We believe there are two limitations here, firstly, none of the larger more recent studies with better outcomes (20,21,26-29) had a control arm of treating COVID related ARDS with conventional mechanical ventilation. Secondly, we believe these are two different populations with patients having variable levels of ARDS severity, with those receiving ECMO treatment being potentially more critically ill in some cases, which could have impacted the outcome for mortality rates.
We have shown by our subgroup analysis that heterogeneity was mainly in the initial Chinese (I2=87%) studies and the following studies showed homogeneity from the USA (I2=0%, P=0.67) and France (I2=0%, p=0.86) with lower mortality associated with the use of ECMO for severe COVID 19 patients, although the number of studies was small. The three largest case series (20,21,27) included in this systematic review were all outside China and showed more promising results than the rest of studies with smaller numbers (18,19,22-26, 28,29). A mortality of 29.2% only in these series as compared to a mortality of 61.1% in the rest of the nine studies (seven Chinese series). This may reflect the fact that high flow ECMO centres achieve superior results. Our data thus confirm these results showing the beneficial effects of ECMO in large tertiary referral centres for the treatment of COVID 19 ARDS. This is consistent with previous findings of Barbaro and his colleagues (32) who found lower ECMO case-mix adjusted mortality in adult patients (adjusted OR, 0.61; 95% CI, 0.48–0.79) in higher volume ECMO centres. Our definition of high volume centres is like them as performing more than 30 cases/year.
The role of ECMO in improving outcomes in severely ill COVID 19 patients seems to be multi factorial. The ARDS observed in COVID 19 patients mostly fits the Berlin criteria (33) but Gattinoni and his colleagues (34) have proposed that the classic ARDS injury is only present in 20-30% of COVID 19 patients with decreased pulmonary compliance less than 40 ml/cmH2O while the non-ARDS type (present in 70-80%), the severe hypoxemia is associated with a respiratory system compliance of more than 50 ml/cmH2O. Hence, severe hypoxemia is primarily due to ventilation/perfusion (VA/Q) mismatch. Unlike classic ARDS, high PEEP pressures and prone positioning in this subgroup of COVID-19 patients do not improve oxygenation through the classic theory of the recruitment of collapsed areas. Although our meta regression showed that pre ECMO days of ventilation did not affect outcome, these patients could benefit from early ECMO to avoid ventilator-induced lung injury and this was the recommendation from the only two included studies in our analysis which had zero mortality in their ECMO patients (18,20)
Secondly, there is an increasing trend to show that COVID 19 infection is associated with a hypercoagulable and thrombotic state. Yin et al (35) studied the differences of coagulation features between severe pneumonia caused by the SARS-CoV-2 (COVID 19) and non-SARS-CoV-2 viruses and found that platelet count of the COVID19 group was significantly higher than that of non-COVID19 patients. Beyls and his colleagues (27) suggested that venous-Doppler ultrasonography of femoral and jugular veins should be performed routinely for severe COVID-19-related ARDS in preparation in case ECMO therapy is needed as they found a higher rate of ECMO related line thrombosis. ECMO circuits eliminate coagulation factors binding them irreversibly to their surface coating material. Systemic anticoagulation is usually utilized for ECMO and further aggravates the anti-coagulatory state on many levels (36). In their current COVID-19 guidelines, the “Extracorporeal life support organisation ELSO” recommended following existing anti-coagulation guidelines, with consideration given to an anti-coagulation targeted at the higher end of normal with vvECMO given the known hyper-coagulable status of COVID-19 patients (37)
In their meta-analysis Munshi and his colleagues (38) have shown a reduction in 60 day mortality in patients receiving ECMO for ARDS in comparison to conventional mechanical ventilation with an associated increased risk of bleeding. The improvement in ARDS outcome and the anti-thrombotic benefit shown in this meta-analysis is a hypothesis of the additional benefit in severely ill COVID 19 patients.
Two of our studies (21,23) suggested that higher mortality related to patients receiving ECMO or conventional mechanical ventilation for severe ARDS can be related to cytokine production. There is accumulating evidence suggesting that a sub-group of patients with severe COVID-19 disease have a cytokine storm syndrome in which a cascade of activated cytokines leads to harmful auto-amplification of inflammatory cytokine production leading to end-organ damage and increasing the risk of mortality. Among COVID-19 patients who have received ECMO, a strong positive correlation exists between mortality and high cytokine levels, most notably IL-6 (39).
Ruan et al (23) found that Interleukin-6 concentrations differed significantly between non survivors and survivors in their COVID-19 cohort, with non survivors having up to 1.7 times higher values. We could not study this in our meta-analysis as both studies did not mention absolute values for cytokines for the ECMO only group.
Our meta-analysis suggests that there is some potential role for ECMO in appropriately selected patients with severe COVID-19. Although the risk factors and variables that contribute to the optimal outcome are complex and reflect individual ECMO center experiences and available resources during the pandemic, it can be argued that it would be unethical to withhold ECMO (or consideration for referral to an experienced ECMO center) in patients who might potentially benefit from this therapy as suggested by Abrams and colleagues (40) when considering ECMO for ARDS due to all causes.
The planning and execution of a randomised controlled trial of this advanced intervention during the current pandemic is difficult. Current challenges include randomisation of markedly sick patients early with a higher risk of death, the need for engagement of many centres worldwide, the lack of the ECMO service in poorer and third world countries and the inconsistency of managing the control non ECMO group. As a result, a current study of ECMO in patients with COVID 19 related severe ARDS soon is unlikely. Thus, our meta-analysis can provide clinicians with the most comprehensive synthesis of all the available limited evidence for the outcome of vvECMO in adult patients with severe ARDS due to COVID-19, although further data collection and meta-analysis for larger studies are invited.