Prone positioning under VV-ECMO in SARS-CoV-2-induced acute respiratory distress syndrome

Background Infection due to severe acute respiratory coronavirus 2 (SARS-CoV-2) may lead to an atypical acute respiratory distress syndrome (ARDS) [1], requiring in the most severe cases veno-venous extracorporeal membrane oxygenation (VV-ECMO). The management of persistent severe hypoxemia under VVECMO requires a multi-step clinical approach including prone positioning (PP), which could improve oxygenation [2].


Background
Infection due to severe acute respiratory coronavirus 2 (SARS-CoV-2) may lead to an atypical acute respiratory distress syndrome (ARDS) [1], requiring in the most severe cases veno-venous extracorporeal membrane oxygenation (VV-ECMO). The management of persistent severe hypoxemia under VV-ECMO requires a multi-step clinical approach including prone positioning (PP), which could improve oxygenation [2].

Methods
We performed a retrospective study of patients with SARS-CoV-2-induced ARDS submitted to PP during VV-ECMO. We aimed to describe mechanical ventilation parameters and gas exchanges before and after PP. We assess the safety of PP and compare patients with PP under ECMO (prone ECMO group) to those maintained in the supine position (supine ECMO group). Patients were treated in accordance with the recommendation guidelines on ARDS [3]. During VV-ECMO, PP was considered in case of severe hypoxemia (PaO 2 /FiO 2 ratio below 80 mmHg) despite FDO 2 and FiO 2 both at 100% and in case of extensive lung consolidation (ECL) on chest imaging (> 50% of lung volume).

Results
We enrolled 208 COVID-19 patients. Among the 125 patients with ARDS, 25 (20%) required VV-ECMO, and 14 (56%) were placed at least once in PP for a total of 24 procedures with a median duration of 16 (15-17) h. The delay from ECMO implantation therapy to PP was 1.5 days [1][2][3]. The resultant changes in ventilator/ ECMO settings and blood gas analysis before and after PP are displayed in Table 1. The median PaO 2 /FiO 2 ratio improvement after PP was 28% . High responders (increase PaO 2 /FiO 2 ratio > 20%) were 62.5%, moderate-responders (increase PaO 2 /FiO 2 < 20%) were 16.7%, and non-responders (decrease PaO 2 /FiO 2 ) were 20.8%. We did not observe any major safety concerns but only pressure sores after 6 procedures, three minor hemorrhages at the injection cannula, and three moderate drops in VV-ECMO flow requiring fluid resuscitation. Pre-ECMO characteristics, ventilator/ECMO settings, and outcomes are exposed in Table 2. Patients in the prone ECMO group were less likely to be weaned from ECMO, and 28-day mortality rate was significantly higher.

Discussion
We report that during VV-ECMO, PP improved oxygenation without a change in respiratory system compliance and PaCO 2 at constant levels of minute ventilation and sweep gas flow. This does not suggest lung recruitment by PP but rather an optimization of ventilation and perfusion matching. Three explanations could be advanced for the mortality rate in the prone ECMO group (78.6%). First, prone ECMO patients may be more severe than supine ECMO patients. As described by Gattinoni et al., worsening patients progress from type 1 to type 2 (higher percentage of non-aerated tissue) [1], which is associated with a higher mortality rate [4]. Prone ECMO patients had much more consolidations, obviously because ECL was the main indication to be prone (n = 10/14). Furthermore, prone ECMO patients need a higher respiratory rate for a higher sweep gas flow suggesting that they may be exposed to a higher mechanical power, and they possibly had also a higher dead space. Second, postmortem biopsies, performed in 6 patients with ECL in the prone ECMO group, found a fibrin exudative presence both in the alveolar spaces and bronchioles followed by a fibroblastic phase [5] and raise the question of the use of corticosteroids (only one patient in the prone ECMO group). Third, as already described by Zeng et al. [6], more than half (8/11) of the patients died from septic shock and multiple organ failure, for which ECMO may be useless.

Conclusion
Prone positioning under VV-ECMO improves oxygenation in SARS-CoV-2-induced ARDS without compromising the safety of the patients. The high mortality rate in prone ECMO patients may be explained by the greater illness severity and the lack of an immunomodulatory therapy such as corticosteroids.  Data are expressed as number (%) or median [IQR]. Analyses were performed with the GraphPad Prism 6 software (San Diego, CA). All tests were two-tailed, with α level at 0.05. To compare the prone ECMO group to the supine ECMO group, we used the non-parametric Mann-Whitney test for continuous variables and the exact Fisher test for categorical ones SAPS II Simplified Acute Physiology Score, IQR interquartile range, CT computed tomography, PaO 2 arterial partial pressure of oxygen, FiO 2 fraction of inspired oxygen, PEEP positive end-expiratory pressure, ECMO extracorporeal membrane oxygenation, ICU intensive care unit