**2. Low tidal volume ventilation (LTVV)**

The first of these, low tidal volume ventilation (LTVV), has remained the mainstay of recommended management strategies for ARDS for the past two decades as a result of clear mortality benefits elicited in randomized controlled studies [4]. It is based on the idea that the functional amount of aerated lung available to

participate in tidal ventilation during ARDS is much smaller than normal [5], and that the potential detrimental effects of permissive hypercapnic acidosis (to a certain extent) are outweighed by the prevention of alveolar stretch-mediated injury during mechanical ventilation. While the ideal tidal volume for any given individual patient remains a matter of debate, the principle of lung protection from excessive tidal volume-induced lung injury, with its consequent reduction in mortality, has been confirmed by meta-analyses of multiple randomized controlled trials as well as two prospective cohort studies [3, 6–8]. Conversely, poor adherence to LTVV in ARDS has been prospectively associated with worse mortality [6, 7].

As pulmonary edema fluid increases, progressive de-recruitment of alveolar gas-exchanging units occurs. In moderate to severe forms of ARDS, the degree of de-recruitment may be so severe that even the delivery of 100% oxygen is insufficient to maintain an acceptable level of hemoglobin oxygen saturation/oxygen delivery for a critically ill patient. Additional reasons to avoid high amounts of oxygen include the potential for increased generation of reactive oxygen species resulting in increased tissue injury [9]. In these circumstances, it becomes necessary to recruit collapsed alveolar units to participate in gas exchange via the application of positive end-expiratory pressure (PEEP). Depending on the severity of gas exchange impairment, the application of PEEP typically ranges between 8 and 20 cm H2O. However, when oxygenation is acutely and dangerously low, typically shortly after induction and intubation of a severe ARDS patient, high amounts of PEEP (up to 45 cm H2O known as a recruitment maneuver) are sometimes temporarily applied for short periods of time to emergently recruit collapsed alveolar units. It should be noted that the routine, non-judicious application of recruitment maneuvers has been shown to increase mortality in ARDS patients [10], and therefore it is reserved only for emergent, salvage situations.

The reason for this lies in the observed heterogeneity of PEEP-responsiveness amongst ARDS patients. CT chest imaging studies have identified groups of patients who demonstrate near-immediate anatomic recruitment of collapsed alveolar units after the application of PEEP versus those who require longer periods of time or who do not exhibit any appreciable anatomic recruitment following a recruitment maneuver and/or increases in applied PEEP [11, 12]. Patients who fall closer to this latter group along the spectrum of PEEP-responsiveness are susceptible during PEEP application to alveolar overdistension in well-aerated regions of the lung (with resultant stretch-mediated injury) [12] as well as hemodynamic compromise resulting from an imposition on ventricular preload [13]. Furthermore, the excessive application of PEEP out of proportion to the degree of responsiveness in any given patient may result in functional decruitment due to decreasing perfusion of aerated alveoli.

Prone positioning is utilized to manage patients with relatively low levels of PEEP responsiveness. It reduces the recruitment threshold by reducing the compressive effects of the heart and abdomen as well as by causing more even pleural pressure distribution [14, 15]. The contribution of PEEP non-responsiveness to alveolar overinflation and augmentation of lung injury is demonstrated by significant decreases in mortality with the use of prone positioning in severe ARDS across multiple studies [16–23].

Non-homogenous distribution of pleural pressures during severe ARDS also contributes to the development of the injurious phenomenon of "pendelluft" or "swinging air" during spontaneous breathing efforts occurring while ARDS patients are deeply sedated but not paralyzed on mechanical ventilation [24]. This leads to overstretch of dependent aerated lung during early inflation with air moving to these regions without a change in tidal volume. However, the contribution of pendelluft to the perpetuation of lung injury during severe ARDS is currently unclear, as data from two large randomized controlled studies examining the use of neuromuscular blockade are conflicting with regards to a mortality benefit attributable to this management strategy [25, 26].

Much of the work done to improve clinical management and outcomes in ARDS patients has validated the concept of abrogating further injury by lung protective measures. Despite many advances along these lines, mortality from ARDS remains high [27], particularly in very severe cases where extreme levels of acidosis and hypoxemia limit the safe applicability of lung protective strategies. It is within this niche that a role for extracorporeal support has developed, which allows substantial, additional lung protection in extreme circumstances via a marked reduction in intrinsic gas exchange requirements. The injured lung can therefore be rested and allowed to recover without further exacerbation.

### **3. Studies**

Three studies, two major randomized controlled trials and one matched paired analysis [28–30], have evaluated the use of ECMO in the clinical care of severe ARDS patients. The Conventional ventilatory support versus Extracorporeal membrane oxygenation for Severe Acute Respiratory failure (CESAR) trial studied 180 patients with severe acute respiratory failure who were randomly assigned either to be referred to a single ECMO center in the UK or to undergo continued conventional management as outlined above [28]. Severe respiratory failure was defined by the presence of one out of two criteria. The first was hypercapnic respiratory acidosis with a pH < 7.20 which would limit allowance of further permissive hypercapnia via lung protective ventilatory strategies. The second used the Murray lung injury score which is based on the ratio of arterial oxygen tension to the fraction of inspired oxygen (PaO2/FiO2), PEEP, lung compliance, and chest radiograph appearance [31]. A score of greater than 3 defined severe respiratory failure. Survival without disability was significantly higher in the patients referred to an ECMO center (63% vs. 47%) [28]. However, limitations of the study in evaluating the actual efficacy of ECMO for severe ARDS included the lack of a homogenous ventilation strategy in the control group, and a high percentage of patients that were referred to the ECMO center but never placed on ECMO (25%).

The ECMO to rescue Lung Injury in severe ARDS (EOLIA) trial studied 249 patients with severe ARDS who received either early venovenous (VV) ECMO or conventional LTTV with late ECMO as a rescue modality [29]. Severe ARDS was defined as a PaO2:FiO2 < 50 mm Hg for >3 h or PaO2:FiO2 < 80 mm Hg for >6 h. The group receiving early ECMO was placed as soon as criteria for severe ARDS were met. The data safety and monitoring board overseeing the study stopped the trial early when interim results were largely in favor of ECMO [32]. However, in the final analysis the primary outcome of 60-day mortality, which remained in favor of ECMO, was not statistically significant (46% vs. 35%) [29]. Survival was also much higher in those who received ECMO within 2 days after onset of ARDS vs. those who received it later within about 6 days after onset (65% vs. 43%). Early cessation of the trial along with a high percentage of patients that crossed over from conventional LTTV to ECMO as a rescue modality may have biased the results away from benefits associated with early ECMO use.

During the H1N1 influenza pandemic of 2009, a study of 75 patients with severe ARDS was conducted in a matched pair design [30]. It found that transfer of patients to an ECMO center improved survival considerably (76.3% vs. 47.5%). 85% of transferred patients were placed on ECMO during the study.

One of the major limitations to conducting well controlled studies in this niche is the relative paucity in numbers of severe ARDS patients necessitating the coordination of large multicenter networks amongst relatively few centers with ample ECMO experience. This comes at great expense of time and resources and, as in the case of the EOLIA trial, risks underpowering of the study when looking at important differences in outcomes. Several meta-analyses have attempted to overcome these limitations. One of these reviewed two randomized trials and three observational studies, finding that 60-day survival was higher in severe ARDS patients who received VV ECMO (66% vs. 53%; RR 0.73, 95% CI 0.58–0.92) [33]. Using a Bayesian randomeffects network metanalysis, another study reviewed 25 randomized clinical trials ranking the relative effectiveness of 9 different interventions (including ECMO) in moderate to severe ARDS patients undergoing lung protective ventilation [23]. The two interventions with the highest-ranking probabilities of significantly lowering 28-day mortality compared with LTVV alone were prone positioning (PP) and VV ECMO (PP: RR 0.69, 95% CI 0.48–0.99; VV ECMO: RR 0.60, 95% CI 0.38–0.93). Finally, a meta-analysis that looked at pooled data from both the CESAR and EOLIA trials found a 90-day reduced mortality in patients who received ECMO (RR 0.75, 95% CI 0.6–0.94) [34].

The SARS-CoV-2 pandemic spanning the last few years has dramatically increased experience across critical care centers around the world in managing patients with severe ARDS, including in the use of ECMO for the most severe cases. Initial observations by some experts suggested COVID ARDS was much different in its need for lung protective strategies compared to non-COVID-related disease [35]. However, subsequent detailed analyses revealed a similar distribution of severity, lung compliances, recruitment thresholds, and response to lung protective measures, leading expert consensus back towards recommending the use of well-established lung protective strategies in COVID ARDS [35].

Due to a temporary increase in large numbers of patients with very severe forms of ARDS worldwide, experience with the use of ECMO in ARDS has grown during the pandemic. While the opportunity for controlled studies has been limited, much of this experience has been reported via observational series and retrospective studies.

A retrospective cohort study conducted in Wuhan, China, on critically ill COVID-19 patients between January 2020 and March 2020 concluded that those who received ECMO had significantly lower in-hospital mortality rates compared to those who received conventional therapy (58.8% vs. 93.5%, *P* = 0.001) [36]. Further analysis of the cause of death between these two groups revealed that zero of the patients in the ECMO group died from ARDS (0 vs. 51.6%, *P* = 0.000). When death occurred in the ECMO group it was more likely to be related to sepsis (17.6% vs. 0, *P* = 0.025). No differences were observed between the two groups for all other causes of death. The most common complication noted in the ECMO group was bleeding (84% of patients in ECMO group) [36].

Blazoski et al. noted in their literature review how poor the overall mortality of COVID-ARDS was compared to influenza-ARDS [37]. For instance, those admitted to the ICU with COVID have a 3.7 times higher risk of death than those who are admitted for influenza. Similar outcomes were found in Veteran's Affairs Hospitals, in which the risk of death was noted to be 5 times higher in those admitted for COVID than for influenza [37]. To see how ECMO affected these statistics, Blazoski et al. looked at the outcomes of ECMO in influenza vs. COVID patients during the

first wave of the COVID-19 pandemic. Patients with ARDS secondary to either influenza or COVID-19 that were placed on ECMO between August 1, 2010 through September 15, 2020 were compared in this retrospective study. Twenty-eight COVID patients and 17 influenza patients were included in this study with the survival rates being overall better in the influenza group compared to the COVID group (94% vs. 68%, respectively (*P* = 0.04) [37]. Further analysis of 30-day survival following VV ECMO decannulation also favored the influenza group compared to the COVID group, being 76% vs. 54% respectively. However, this finding was not statistically significant (*P* = 0.13). This study found that COVID patients tended to spend more time on ECMO support (21 days vs. 12 days, *P* = 0.25) and had higher rates of new infection (50% vs. 18%, *P* = 0.03) and bacterial pneumonia (36% vs. 8%, *P* = 0.24) when compared to their influenza counterparts [37]. COVID patients in this study were more likely to have received immunomodulatory therapy prior to ECMO initiation as part of their treatment, which may have played a role in their higher infection risk.

Jäckel et al. conducted an analysis looking at the use of VV ECMO in COVID-19 ARDS as compared to influenza related ARDS in a retrospective study of patients managed between October 2010 and June 2020 [38]. At 30 days following ECMO cannulation, 13.35% of COVID-19 ARDS patients vs. 44.7% with influenza were discharged alive from their ICU (*P* = 0.03). COVID-19 patients were also more likely to have fewer VV ECMO free days and longer ICU treatment duration than their influenza counterparts. 30-day mortality was noted to be higher in the COVID-19 group but wasn't found to be statistically significant. This may have been secondary to a smaller number of cases in the COVID group compared to the influenza group (15 vs. 47) [38].

Given the high cost in labor and other resources that are associated with ECMO support, prediction models of survival on ECMO have long been sought, and experience during the pandemic expanded knowledge specifically for severe ARDS patients. Zayat et al. conducted a single-center, retrospective observational study examining all severe COVID-ARDS patients who received ECMO support between March 1, 2020 to April 20, 2020 [39]. A total of 83 pre-ECMO variables including biomarkers, risk scores, and demographics were evaluated for predictiveness of survival. Procalcitonin, IL-6 and NT-proBNP were all remarkably higher in non-survivors versus survivors. Data also validated the Respiratory Extracorporeal Membrane Oxygenation (RESP) Score [40] as a viable survival prediction tool for patients with severe COVID ARDS who undergo ECMO support.

Optimal timing of ECMO initiation for severe ARDS is also a matter of ongoing debate, with some experts favoring early institution when it appears that lung protective strategies will not be viable on the basis of severe acidosis and/or hypoxemia, while others favor its institution only after all other lung protective strategies have been attempted. Two studies conducted during the pandemic have contributed to this debate.

A cohort research study conducted by Giraud et al. included COVID-19 ARDS patients admitted to the Geneva University Hospital ICU between March 14 and May 31, 2020, who were supported on VV ECMO [41]. Amongst the 10 patients studied, mean durations of mechanical ventilation and ECMO were 7 ± 3 days and 19 ± 11 days, respectively. Six patients died in the cohort, leaving the study mortality at 60%. This study highlighted that survivors had a significantly shorter duration on mechanical ventilation prior to ECMO initiation compared to nonsurvivors (91 ± 58 h vs. 208 ± 34 h, *P* = 0.01) as well as a shorter amount of time on


**Table 1.**

*Summary of studies.*

ECMO (246 ± 102 days vs. 588 ± 294 days, *P* = 0.038) and in the ICU (17 ± 6 days vs. 32 ± 12 days, *P* = 0.016). Overall, this meant that those who received longer than 7 days on mechanical ventilation prior to initiation of ECMO in their study ultimately died. The study found no other pre-ECMO variable that was statistically significant in predicting survival. The investigators concluded that ECMO is a viable option for refractory hypoxemia in COVID-19 patients with ARDS, but that it should be considered early in their clinical course, as late initiation of ECMO therapy (beyond 7 days of mechanical ventilation) is likely futile [41].

The other retrospective cohort study contributing to the debate around timing was conducted by Kurihara et al. [42]. They similarly reported that COVID-19 ARDS patients who received more than 7 days of mechanical ventilation prior to VV-ECMO initiation had a very high mortality rate. In this study mortality was 100% when ECMO initiation was delayed beyond 7 days of mechanical ventilation. The COVID-19 ARDS patients who received 7 days or less of mechanical ventilation had a 63.1% mortality rate compared to 30.7% in the non-COVID-19 ARDS group. The investigators were unable to determine why the 7-day cut off was so significant. Since COVID-19 patients typically experience multiple episodes of proning, they decided to evaluate if the increased number of proning episodes in COVID-19 patients prior to VV ECMO cannulation affected post-ECMO mortality. They did not find a specific number of proning episodes that predicted mortality post-ECMO in the COVID-19 cohort.

## **4. Conclusion**

Taken altogether, the preponderance of research in the field has clearly demonstrated an exquisite sensitivity of the ARDS lung to stretch-mediated injury with strong signals for increased mortality when lung protective strategies are abandoned. It follows that in select cases of extremely severe ARDS, a role for extracorporeal support exists in which the injured lung is allowed to rest and recover. Although more work needs to be done, this hypothesis is supported by the current cache of clinical research observations including those derived from experiences during the COVID-19 pandemic (**Table 1**). In response to this data, current guidelines established by the international Extracorporeal Life Support Organization (ELSO) suggest consideration of ECMO in patients with severe ARDS and refractory hypoxemia (PaO2/ FiO2 < 80 mm Hg), or severe hypercapnic acidosis (pH < 7.25 with a PaCO2 ≥ 60 mm Hg) after optimal conventional management including a trial of prone positioning (in the absence of contraindications) [43]. Since increased duration of mechanical ventilation prior to the institution of ECMO is associated with worsened mortality, it is also recommended that optimal medical management be rapidly and maximally implemented, and transition to ECMO performed without delay when indicated. The only absolute contraindication to the use of ECMO in severe ARDS is anticipated nonrecovery without a feasible plan for decannulation or possibility of bridge to transplantation [43].

As advancements in ECMO device technology and implementation experience continue to reduce complication rates, it is anticipated that future research will be sufficiently powered to definitively refine optimal patient selection and timing of ECMO implementation in severe ARDS. Advancements in single site cannulation methods, device portability, and experience managing patients on VV ECMO without mechanical ventilation have allowed inroads to be made with regards to safe mobility-promoting therapy during extended ECMO support [44], which is independently associated with improved outcomes during critical illness [45]. It is hoped that these efforts will culminate in continued reductions in the high mortality rates associated with this otherwise devastating condition.
