Introduction
Despite improvements in the management of severe acute respiratory distress syndrome (ARDS), in-hospital mortality remains high, exceeding 40% [
1]. Some patients with severe ARDS and refractory hypoxemia, hypercapnia or uncontrolled high airways pressures may benefit from venovenous-extracorporeal membrane oxygenation (VV-ECMO) [
2]. One of the main goals of VV-ECMO is to rest the lungs by applying a so-called “ultra-protective” ventilation strategy, combining significant reductions of the tidal volume (VT) and intrathoracic pressures [
3,
4], to enhance prevention of ventilator-induced lung injury (VILI).
Prone positioning (PP) is an effective first-line intervention to treat ARDS [
5], as it improves gas exchanges [
6] and lowers mortality [
7]. However, the response to PP remains unpredictable [
8,
9] and may differ from one patient to another [
7,
10]. Despite these differences, PP is associated with greater survival [
10]. This effect on mortality regardless of impact on respiratory mechanics could be explained by a more uniform distribution of VT leading to a reduction of VILI [
11], but this effect was not well studied.
Although the use of PP combined with ECMO is still controversial [
12], few studies have suggested that it could improve oxygenation and static compliance [
13,
14], thereby preventing subsequent VILI, when associated with “ultra-protective” ventilation. To date, PP physiological effects on regional ventilation and the optimal positive end-expiratory pressure (PEEP) level in this specific population with very low VT and altered static compliance are unknown. Moreover, due to the extreme severity of these patients, and the inherent risks of PP in this specific population, such as decannulation, teams could be reluctant to perform PP.
In that context, electrical impedance tomography (EIT) could be a promising tool to describe the respective impacts of PP on regional ventilation and possible change of the “optimal PEEP” level on PP. Indeed, EIT is bedside, real-time, non-invasive, functional and radiation-free imaging of the lungs, which provides a regional dynamic lung analysis [
15,
16]. Its performance in the context of ECMO-supported severe ARDS was validated, showing that it could be a useful tool to titrate the optimal PEEP in this severely ill population [
17].
We hypothesized that EIT could help to monitor the impact of PP on regional ventilation and optimal PEEP highlighting potential beneficial effects to prone ECMO-treated severe ARDS patients. The aims of our study on ECMO-supported severe ARDS patients were therefore to describe (1) the PP impact on regional ventilation; (2) the PP influence on the optimal PEEP level; and (3) to identify different EIT patterns depending on static compliance gain.
Discussion
To our knowledge, PP physiological effects in a population of patients with severe ARDS on ECMO have not yet been thoroughly studied. Based on EIT-monitoring, the main findings were: (1) although PCG+ and PCG− had different tidal volume distribution before PP, all patients with severe ARDS on ECMO exhibited a shift of the VT distribution and EELI from the ventral-to-dorsal ROIs resulting in an increased of the local compliance and the VTdorsal/VTglobal ratio and; (2) EIT-estimated optimal PEEP decreased with PP, highlighting the potential reduction of atelectasis with PP.
By defining PP “responders” on PaO
2 improvement, other ventilatory monitoring tools, such as thoracic tomodensitometry) [
8] or lung ultrasonography [
9], have failed to predict PP response at baseline. Because we anticipated that gas exchanges may be more likely influenced by ECMO settings than PP effects, we chose to use modification of static compliance to identify two profiles of PP response. Interestingly, 62% patients were classified as being PCG+ . Their PP sessions were associated with a significant decrease of PaCO
2, which is, a well-known marker of PP response [
10,
30]. In addition, this subgroup had higher body mass indexes and more frequent viral pneumonia, highlighting the potential benefits to pursue PP in these patient subgroups.
These data suggest that EIT could offer a promising bedside, dynamic, non-invasive, functional analysis of lung mechanics that could predict and monitor potential PP “response” on ECMO. Our results mainly underscored immediate PP effects, which continued to evolve during the 16-h procedure, illustrating the need for long PP sessions to obtain the best benefits [
6,
31,
32]. Consistent with previous publications [
13,
14], we found that on-ECMO PP was simple, feasible and safe with no PP-related complications of the ECMO circuit.
Baseline lung mechanics and predominant lesions differed between PCG+ and PCG− patients, and could explain these different responses between subgroups. Indeed, baseline tidal impedance were mostly distributed in ventral ROIs only in PCG+ . This finding suggests a collapse of the dorsal lung ROIs and functional aerated ventral ROIs in PCG+ , whereas ventral ROIs were over distended in PCG− at baseline. Consistently, previous observations reported that PaCO
2-based beneficial PP effects mainly depend on the lung recruitment/derecruitment ratio [
11,
33]. In addition, EIT-determined local compliance increased at a higher percentage than global compliance, suggesting a potential negative PP impact on another lung ROI, not captured by EIT at this thoracic level. Pertinently, Bikker et al. described different EIT-pattern responses to a PEEP trial when they were evaluated at two different thoracic levels. Hence, it cannot be excluded that different response patterns at different thoracic levels might also occur during PP.
Based on these results, should we decide to perform PP based on predicted PP-response for ECMO patients? The relevance of this question appears low. Indeed, the increase of EELI in the dorsal regions was observed in all patients, as previously reported with other tools [
34,
35] and in other populations. Moreover, it is worth noting that PP significantly impacted on regional VT distribution and optimal PEEP levels in all patients, regardless of their static compliance modifications. Our study suggests that global static compliance or gas exchanges are not good surrogate of the impact of PP on regional ventilation. Indeed, improvement of local compliance, VT and EELI redistribution, were observed even for patients with lower global static compliance at the end of PP. Our preliminary results suggest striving to prone all ECMO-supported ARDS patients regardless their predicted response in terms of static compliance improvement. These results are consistent with previous studies suggesting that PP benefits are independent of the oxygenation/decarboxylation responses [
10] and may be more related to less VILIs.
One of our main results is the impact of PP on optimal PEEP evaluated with EIT. EIT-based optimal PEEPs decreased significantly after 16 h of PP, highlighting the remarkable PP impact on respiratory mechanics. To date, this aspect of PP management has only been studied in patients with healthy lungs and yielded controversial results. Spaeth et al. [
36] reported that PP was required to increase PEEP to avoid lung collapse in patients with healthy lungs after lumbar spine surgery, whereas Petersson et al. [
37] suggested that application of PEEP during PP was associated with increased ventilation/perfusion mismatch in healthy subjects.
Our study has several limitations. First, EIT provides only a cross-sectional lung-region evaluation, which may differ from whole lungs [
22]. This distinction might explain the differences between local and global compliance variations reported herein. Second, we chose to apply a 16-h PP session, as described by Guerin et al. [
7]. However, our findings cannot rule out that the PP impact could evolve beyond 16 h, enhancing the PP impact on local and global ventilation. That hypothesis was also supported by Kimmoun et al.’s observations, after subjecting 17 severe ARDS patients on VV-ECMO to one or more 24-h PP sessions [
14]. In this study, patients exhibited major compliance improvement that persisted 24 h after returning to SP. Third, we defined optimal PEEP as the best compromise between lowest overdistension and collapse, as previously described [
20]. We cannot exclude that different optimal PEEP identification methods, e.g., inhomogeneity index [
38] or dependent silent spaces [
39], would have yielded different results. However, we can acknowledge that the selected optimal PEEP’s influence on our results was probably limited because ventilator settings were unchanged throughout the entire protocol. Lastly, our study enrolled relatively few patients and our promising results need to be confirmed in larger studies.
Acknowledgements
This work was done in the Service de Réanimation Médicale, Assistance Publique–Hôpitaux de Paris, Groupe Hospitalier Pitié–Salpetrière, Paris, France.
The Pulmovista® electrical impedance tomograph was provided by Dräger (Lübeck, Germany) during the study period. Dräger had no role in the study design, collection, analysis, and interpretation of the data, writing the article, or the decision to submit the article for publication.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.