Editorial 

December 1, 2023

Prone Positioning During Venovenous ECMO for Severe ARDS

Ricardo Teijeiro-Paradis, Niall D. Ferguson

JAMA. Published online December 1, 2023. doi:10.1001/jama.2023.22456

Care of patients with severe hypoxemic respiratory failure receiving mechanical ventilation has advanced significantly in the last 25 years due to a focus on minimization of iatrogenic injury, largely ventilator-induced lung injury. In addition to limitation of tidal volumes and airway pressures, the use of prone positioning is a technique for lung protection.¹ Regional variability predisposes to overdistension of nondependent and cyclical opening and closing of dependent lung regions ,which in turn drive ventilator-induced lung injury.² Prone-position ventilation induces changes in chest wall and respiratory system compliance, improving gas exchange and lung mechanics. Over a series of clinical trials, clinicians and scientists learned that longer daily application of prone positioning (16 hours per day) in patients with moderate to severe hypoxemia (ratio of Pao₂ to fraction of inspired oxygen <150 mm Hg) culminated in the large mortality reduction demonstrated in the PROSEVA trial.³ Prone positioning improved oxygenation on average in that trial, but there was no association between changes in oxygenation and survival, further supporting that the main benefit of prone positioning is less ventilator-induced lung injury.⁴ The use of routine prone positioning is strongly recommended in current clinical practice guidelines.⁵

At the same time, another therapy, venovenous extracorporeal membrane oxygenation (VV-ECMO), emerged as an effective therapy in a subgroup of patients with very severe acute respiratory distress syndrome (ARDS). With this technique, large-bore cannulas allow venous blood to be pumped through a membrane lung, adding oxygen and removing carbon dioxide, before returning it to the right heart. This artificial lung allows mechanical ventilator settings to be significantly reduced, decreasing driving pressure and ventilator-induced lung injury and improving survival.⁶

Combining these 2 effective therapies—ie, placing patients already undergoing VV-ECMO in the prone position—has been proposed as a strategy to further improve outcomes. Part of this rationale is based on the fact that reduction in ventilator pressures during VV-ECMO can lead to derecruitment of the lung, which could favor ventilator-induced lung injury (atelectrauma), hypoxemia (despite ECMO), or liberation failure.⁷ Prone positioning during VV-ECMO might reduce decruitment or mitigate its consequences.⁸ Simultaneous use of both strategies has been reported in numerous observational studies, which have been summarized in recent meta-analyses.^8-10 A conventional meta-analysis reported a consistent improvement in survival with prone positioning added to VV-ECMO (74% vs 58%).⁸ An individual patient meta-analysis had similar crude results; however, these were attenuated in multivariable analysis, not reaching statistical significance.¹⁰

In this issue of JAMA,¹¹ the PRONECMO Investigators report the results of their multicenter randomized clinical trial comparing early prone positioning with usual care in 170 patients receiving VV-ECMO in 14 French intensive care units. The primary outcome was successful ECMO weaning within 60 days, whereby patients had to survive without further ECMO or lung transplant 30 days after ECMO removal. Both groups received similar cointerventions and followed the same ultraprotective ventilation strategy and standardized liberation protocol. Baseline characteristics were similar between groups (94% had COVID-19 ARDS) and nearly all patients received prone-position ventilation prior to ECMO (99% and 94%). Successful ECMO weaning occurred in 44% (38/86) and 44% (37/84) of patients in the prone and supine groups, respectively. Prone positioning during ECMO did not improve survival at days 60 and 90, did not improve ECMO-free or mechanical ventilation–free days, and did not shorten intensive care unit or hospital length of stay. There were no reported differences in the rate of complications.

We commend the PRONECMO Investigators for conducting this needed, rigorous, and well-designed randomized trial. Based on these results, routine prone positioning during VV-ECMO does not facilitate earlier liberation from ECMO or improve outcomes. This suggests that on average, prone positioning does not facilitate further lung protection than that already provided with an ultraprotective ventilation strategy facilitated by VV-ECMO (tidal volume <3 mL/kg, plateau pressure ≤24 cm H₂O, and driving pressure ≤14 cm H₂O). This contrasts with much of the previously published observational data, suggesting a significant amount of treatment indication bias in those data. In addition, the fact that almost all patients had ARDS secondary to COVID-19 may also play a role in these different results.

Several questions remain unanswered. The PRONECMO Investigators enrolled all patients regardless of the response to ECMO cannulation. In contrast, at least 5 of the 11 observational studies claiming survival benefit from prone positioning during ECMO had specific indications for the intervention, such as refractory hypoxemia despite optimization of ventilation and ECMO or failure to wean from ECMO with or without evidence of consolidation on chest imaging.^7,12-19 Indeed, if patients are adequately supported with VV-ECMO and have their mechanical ventilatory support reduced to very low levels that are unlikely to propagate ventilator-induced lung injury, it is unclear how early prone positioning would further reduce ventilator-induced lung injury and mortality. It remains unknown whether different results would be achieved if tested in patients in whom prone positioning has stronger potential biological indications.

Like ECMO centers included in the PRONECMO trial, we target lung ultraprotective ventilation during VV-ECMO at our center, lowering tidal volume, driving pressure, and respiratory rate after cannulation. Not infrequently, our patients develop complete lung collapse/consolidation with resulting tidal volumes far below anatomical dead space. Such derecruitment is exacerbated by the lack of respiratory effort due to deep sedation and paralysis often needed to maintain adequate ECMO flow and safe inspiratory pressures. A similar phenomenon may explain the progressive decrease in respiratory system compliance observed in both groups in the PRONECMO study (eFigure 1 in the article’s Supplement 1). This observation is relevant, as prone positioning requires sufficient gas flow to generate alveolar pressure gradients and redistribute alveolar stress and strain more homogeneously.¹ Unless prone positioning results in considerable mobilization of pulmonary secretions, redistribution of aerated units is unlikely and could explain the lack of improvement in oxygenation in this trial. Whether prone positioning facilitates liberation from VV-ECMO at later stages when airway pressures are liberalized or when spontaneous breathing is allowed, also remains unknown.

We consider this first randomized trial of prone positioning during VV-ECMO to be an important addition to the existing literature, which suggests against routine adoption of prone-position ventilation during ECMO. Results of the PRONECMO trial may lead us to turn our backs on routine prone-position ventilation during VV-ECMO and face forward toward novel group-specific interventions targeting lung- and diaphragm-protective ventilation, early awakening, and patient mobility.