Extracorporeal Membrane Oxygenation (ECMO) Dependent Acute Respiratory Distress Syndrome (ARDS): A Systematic Review and Meta-Analysis

Background: Extracorporeal membrane oxygenation (ECMO) has emerged as a newer method for managing severe acute respiratory distress syndrome (ARDS) and ARDS refractory to conventional management. However, its current role in the management of ARDS is not clear. Therefore, we conducted this meta-analysis to compare the mortality rates of ECMO over conventional management in ARDS. Methods: PubMed, PubMed Central, Embase, and Scopus were searched using appropriate keywords. We selected studies in adults with ARDS that compared the outcomes of patients treated with ECMO vs. conventional management. Cochrane Risk of Bias (RoB) 2.0 and the JBI (Joanna Briggs Institute) quality assessment tools were used for assessing the risk of bias in RCTs and observational studies, respectively. The I2 statistic was used to evaluate heterogeneity, and quantitative synthesis was performed using fixed or random effects to pool studies based on heterogeneities. Meta-analysis was conducted using Revman 5.4. Result: Twelve studies were included in this meta-analysis. As compared to the conventional management (mechanical ventilation: MV), patients treated with ECMO had lower odds of 30-days mortality (OR, 0.56; 95% CI, 0.37 to 0.84) and 90 days mortality (OR, 0.59; 95% CI, 0.41 to 0.85). However, there was no significant difference between in-hospital mortality (OR, 0.75; 95% CI, 0.40 to 1.41) and intensive care unit (ICU) mortality (OR, 1.00; 95% CI, 0.36 to 2.79). Similarly, length of hospital stays (LOS) (MD, 3.92; 95% CI, -6.26 to 14.11) did not show statistically significant differences across the two groups. However, the average ICU stay (ICU LOS) was 7.28 days longer in the ECMO group compared with the MV group (MD, 7.28; 95% CI, 2.55 to 12.02). Conclusion: Twenty-eight days and 90-days mortality were decreased in patients managed with ECMO compared with the MV group. Also, ICU LOS was found to be longer in the ECMO group. Furthermore, no statistical difference was found between the two groups for in-hospital mortality and hospital LOS.


Introduction And Background
Acute respiratory distress syndrome (ARDS) is one of the most common presentations in the intensive care unit (ICU). It has been managed conventionally by mechanical ventilation, and lung-protective ventilation has remained a cornerstone of ARDS management. With the discovery of extracorporeal membranous oxygenation (ECMO), it is considered a tool for managing severe ARDS. ECMO is a modified cardiopulmonary bypass circuit that provides gas exchange and ensures systemic perfusion to sustain the patient's life in pulmonary and cardiac failure refractory to conventional therapy. It alleviates the need for high airway pressures, thereby allowing the lungs to rest and prevent the effects of high pressures in the airway [1]. The associated risk of complications related to ECMO in patients with refractory ARDS is found to be coagulopathies, infections, hypoxia, ischemia, multi-organ failures, and others [2]. The two randomized controlled trials (RCT) could not confirm the superiority of the technique over more conventional management [3,4]. However, studies in the recent past show that ECMO and ventilator techniques have better survival rates and have improved six-month disability-free survival [5,6,7]. Given the lack of adequate data that compares the use of ECMO with other modalities of management in refractory ARDS, our study aims to evaluate the overall outcomes and outcome predictors, the etiologies, and the risk factors associated with ECMO dependent ARDS.

Review Objectives
To compare mortality and length of hospital and ICU stay in patients with ARDS managed with ECMO to conventional treatment with mechanical ventilation.
To compare serious adverse events among patients with ARDS managed with ECMO to conventional treatment with mechanical ventilation.

Methods
We used the PRISMA guidelines for this metanalysis [8]. The protocol has been registered in the International prospective register of systematic reviews (PROSPERO) (CRD42020215494) [9].

Eligibility Criteria
Types of studies: We included prospective as well as retrospective observational studies and randomized clinical trials, which compared the mortality rate, clinical improvement and recovery, length of hospital stay, adverse effects of ECMO, mean difference of clinical improvement, and healing among patients receiving ECMO for ARDS as compared to conventional/conservative treatment. We have only included the studies after 2000 as there have been significant changes in ECMO management.
We have not included editorials, comments, viewpoint articles, systematic reviews, and meta-analyses. In addition, we have not included studies in which ECMO is used for the management of cases other than ARDS and the studies which have not mentioned our outcome of interest.
Types of participants: We included all patients suffering from ARDS > 18 years of age receiving ECMO or treated with conventional or conservative treatment. We have not included non-ARDS patients, less than 18 years of age, or pregnant patients.
Outcomes: We compared mortality at different time durations, cause of death, hospital and ICU length of stay, days on mechanical ventilation, number of days alive, and post-discharge mortality rates between patients receiving ECMO to those receiving conventional treatment with mechanical ventilation.

Search Methods
Two authors (PB and DBS) independently searched and evaluated the quality of the studies done in the past decade, identified via electronic search in PubMed, PubMed Central, Embase, Scopus, and Google Scholar databases.

Data Collection and Analysis
We extracted the data for quantitative synthesis through Covidence and did the analysis using RevMan5.4 (London, UK) [10,11]. Assessment of heterogeneity was done using the I-squared (I2) test. We used a random/fixed effect for the pooling of selected studies.
Selection of studies: Articles from the literature search were imported to Covidence, and duplicates were removed. Two researchers independently screened the titles and abstracts of all articles included. The conflicts were resolved by discussing with a third reviewer, and articles were finalized for full-text review. The same procedure was used to carry out a full-text review of the screened articles to include in the study.
Data extraction and management: Two researchers independently extracted data from the included studies, and any discrepancies were resolved through discussion. The extracted data were entered into Revman5.4. We evaluated the quality of studies thoroughly and considered only the outcomes of our interest.
Assessment of risk of bias in included studies: We used the Cochrane ROB 2.0 tool to analyze our RCTs, and we used the Joanna Briggs Institute (JBI) quality assessment tools to assess the risk of bias in our prospective and retrospective observational studies ( Figure 1 and Tables 1, 2) [12,13]. We used RevMan 5.4 for creating a summary of preferences for RCTs using the Cochrane ROB 2.0 tool.    Assessment of heterogeneity: The I-squared (I 2 ) test was used to assess heterogeneity [21]. We interpreted the I-squared (I 2 ) test based on the Cochrane Handbook for Systematic Reviews of Interventions.
Assessment of reporting biases: We assessed the reporting biases through predetermined outcome reporting documentation.
Data synthesis: Statistical analysis was performed using RevMan 5.4 software. Odds ratio (OR) was used to estimate discrete outcomes with a 95% confidence interval (CI). We analyzed the mean differences among the two groups for continuous outcomes using mean and standard deviations when available or after calculating mean and standard deviation when the median, sample size, and interquartile range were reported. Mean and SD was calculated to form median and interquartile range (IQR) using the following estimation for continuous variables (LOHS, ICU LOS) [22].
The fixed/random-effects model was used according to heterogeneities.
Subgroup analysis and investigation of heterogeneity: We presented forest plots to visualize the degree of variation between studies.
Sensitivity analysis: For sensitivity analysis, we examined the effect of the study based on their type (RCT and non-RCT) by excluding non-RCT studies when appropriate and re-running the analysis to find any differences. In addition, non-randomized studies were excluded for sensitivity analysis to find any alterations in the outcomes after removal.

Result
Twelve thousand three hundred fifty-seven studies were imported from a database search for screening. After removing duplicates, the title and abstracts of 9478 studies were screened. Eight thousand three hundred ninety-five studies were excluded, and full-text eligibility of 1083 studies was assessed. One thousand seventy-one studies were excluded for definite reasons. Twelve studies were included in the quantitative analysis, and 11 were included in the qualitative analysis ( Figure 2).

Qualitative Summary
Qualitative details of included studies are presented in Table 3.

Quantitative Analysis
Eleven studies were included in the quantitative synthesis.

Discussion
We did not find any significant difference in in-hospital mortality and ICU mortality. A similar conclusion was found for in-hospital mortality in a meta-analysis pooled with one RCT and two observational studies [26]. However, in the same study, the in-hospital mortality reduction was seen with ECMO when results of observational studies were analyzed using a propensity score matching with replacement [26]. However, multiple meta-analyses which have reported 30 days mortality [1], 60-day mortality [1], or 90 days mortality [27], have found lower mortality rates with ECMO than that with conventional ventilation. This is similar to our outcome. A recent meta-analysis [28] had not seen any difference in mortality at 30 days with ECMO or conventional treatment, which had included studies before 2000 AD when the management of ARDS was different from that of recent times, and ARDS management now is evolved a lot in comparison to earlier days with modern technologies. This could have led to the difference in our findings.
We did not find any significant difference in length of hospital stay, but ICU stay was longer in the ECMO group, which is expected. The previous meta-analysis by Mendes et al. [29] has also noted an increase in ICU and hospital length of stay in patients treated with ECMO, which the authors have attributed to increased survival among the patients as compared to the patients treated conventionally. However, in our analysis of the length of hospital stay, Wang et al. [15] have reported the duration only for the survivors in both the cases of ECMO and non-ECMO groups, which could have possibly obscured any difference that could be attributed to increased survival and led to the result.
Because of the lack of reporting of adverse events, we could not analyze the difference in complications of ECMO. A few studies reported hemorrhagic complications in patients treated with ECMO but not for patients managed with conventional ventilation [2,17]. ECMO to Rescue Lung Injury in Severe ARDS (EOLIA) trial [23] reported an increased incidence of thrombocytopenia and bleeding events requiring transfusion in the ECMO arm compared to the control arm. In a systematic review, bleeding complications were seen in 29.3% of patients treated with ECMO with significant bleeding in 10.4%; the majority causes of bleeding were cannula bleeding (9.3%) [30]. Two studies included in our study reported the need for renal replacement therapy (RRT) [17,19]. Both of them reported increased requirements of RRT in patients treated with ECMO. However, in a recent meta-analysis, management with ECMO was not associated with an increase in RRT incidence [28].
We have included the two RCTs that have compared ECMO vs. conventional that have been done in the last two decades [7,23]. Furthermore, we have included other prospective and retrospective studies without any randomization. For example, in their combined meta-regression model, Vaquer S et al. [30] have shown an association between MV duration before ECMO support with mortality. However, in our meta-analysis, the time of the start of ECMO after mechanical ventilation is variable in the studies. Similarly, patients with a PAO2/FIO2 ratio of less than 150 have higher ventilator-free days; the effect was not seen in patients with a higher PAO2/FIO2 ratio [31]. However, there is considerable variation in the studies included in our metaanalysis in terms of the PaO2/Fio2 ratio. Also, there is venoarterial ECMO in selected patients in some studies. Similarly, there might have been wide variation in methods while managing patients with conventional management, including applying for prone positions, using concurrent steroids, and maintaining pressure and volume while managing patients. This could have introduced some bias in our study.
There have only been two RCTs in the last two decades that have studied ECMO vs. conventional management in ARDS. CESAR [7] showed a mortality benefit of ECMO while EOLIA [23] did not. However, both of them have their limitations. The meta-analyses that have pooled data only from these two RCTs have shown mortality and other benefits of ECMO in ARDS [27,29]. Thus, concerning all these studies and findings, it seems evident that patients benefit from ECMO in ARDS. However, the selection criteria for patients who will benefit is well defined, nor the appropriate time for initiation of ECMO in the patients is evident at present. Further studies should be conducted to shed light on these issues. Our meta-analysis pooled available data; however, a limited number of available papers and heterogenous population and variation in the individual studies are significant limitations of our meta-analysis.

Conclusions
We analyzed 12 studies in our study. We found lower odds of mortality at 28 days and 90 days in the ECMO group of patients compared with the MV group. Also, ICU LOS was found to be longer in the ECMO group. Furthermore, no statistical difference was found between in-hospital mortality and hospital LOS across the two groups. We could have limited exploration in the analysis due to the limited information available and relatively few studies. Further studies should be conducted to evaluate the outcomes of ECMO in ARDS.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.