In theory, the application of extracorporeal membrane oxygenation (ECMO) in severe
respiratory failure allows lung treatments varying from a lung at rest (continuous
positive airway pressure) to all different levels of ventilatory support or even pure,
spontaneous breathing. Although ECMO is increasingly used worldwide, very little is
known about the respiratory settings applied during the course of ECMO, and even less
is known about the optimal “balance” of ventilatory and extracorporeal support to
minimize ventilator- or ventilation-induced lung injury, and the optimal conditions
for lung healing and repair. In this issue of the Journal, Schmidt and coauthors (pp.
1002–1012) present an international, multicenter, prospective cohort study (LIFEGARDS
[Ventilation Management of Patients with Extracorporeal Membrane Oxygenation for Acute
Respiratory Distress Syndrome]) in which data from 350 patients with ECMO in 23 international
ICUs were collected during a 1-year period (1). In addition to demographics, the authors
carefully compiled data regarding the ventilator settings applied before and during
ECMO, the use of adjunctive therapies, and ICU and 6-month outcomes. The authors and
their participating centers should be congratulated for providing the community with
such sound data from different countries and ICUs, as well as the preferential ventilator
settings used before and during the application of ECMO. The primary outcome measured
was 6-month mortality, but the study also provides data on the type and use of adjunctive
therapies, as well as the changes in driving pressure and mechanical power before
and during the ECMO run. Some of these observational data are in part confirmatory
and quite striking (2, 3). This study included only ICUs with an annual ECMO volume
of more than 15 cases, and all of the participating centers treated a median of 30
patients with ECMO in the year before the study. Therefore, they could be clearly
classified as “experienced.” In this context, it is more than striking that the prone
position was not used in more than 26% of the patients, especially when a plateau
pressure of 32 cm H2O was applied. Instead, the fact that a reported 15% of patients
were turned to prone even during the ECMO course gives reason to hope that proning
will be more regularly applied also in patients without ECMO. In contrast, with a
Vt of 6.4 ± 2.0 ml/kg, patients were ventilated close to the magic “protective” value.
However, the ventilatory setup as a whole led to a plateau pressure of 32 ± 7 cm H2O,
a ventilatory rate of 26 ± 8, a driving pressure (ΔP) of 20 ± 7 cm H2O, and a mechanical
power of 26 ± 12.7 J/min. It is interesting to note that after the ECMO initiation,
while the reduction in DELTAP was only 30%, the reduction in mechanical power was
as great as 75%, reflecting the importance of the frequency for energy transmission.
With an overall 6-month survival of 61%, the study presents impressive outcome findings.
The changes in respiratory settings after ECMO initiation resulted in both ΔP and
power values below the thresholds that have been considered “critical” in both experimental
and clinical studies (4–7). It is thus not surprising that the ventilator settings
applied during the first 2 days after ECMO onset had no impact on survival, whereas
age, immunocompromised state, extrapulmonary sepsis, and lactate and fluid balance—all
of which could be considered indicators for the general severity of disease—were positively
correlated. Given the ΔP and power values observed before ECMO was initiated, it is
not unexpected that each day of delaying intubation to ECMO was also positively correlated
with a higher 6-month mortality. Moreover, higher spontaneous respiratory rates during
the first 2 days of ECMO were associated with higher 6-month mortality.
The strength of this study, which used data from different ICUs in 10 different countries,
lies in the amount and quality of the data and the homogeneity of the treatment, including
the use or nonuse of adjunctive measures. At the same time, this is also a limitation,
as these data certainly do not reflect the real world of patients with ECMO treated
in non-university hospitals or in hospitals with lower ECMO volumes and less experience
in treating patients with severe respiratory failure and/or acute respiratory distress
syndrome. In addition, the study describes how the patients were ventilated after
the onset of ECMO, but it does not provide the reasons for the chosen partitioning
between gas exchange across the native lungs and the artificial lung, or the rationale
behind each specific ventilatory pattern. It is also unclear why a Vt of 3.7 ± 2.0
ml/kg ideal body weight and a respiratory rate of 14 ± 6, including 8 ± 11 spontaneous
breaths at a positive end-expiratory pressure of 11 ± 3 cm H2O, was chosen. This study
clearly identifies crucial questions for further research: how much unloading of the
lungs is most beneficial for healing and repair, and what is the best composition
(i.e., ventilatory pattern) of the chosen load? It seems reasonable to choose a ventilator
setting that enables the greatest alveolar ventilation (i.e., the highest amount of
CO2 removal) with the lowest price to pay (resulting power). A simplified mathematical
approach makes it possible to determine for any given power the combination of Vt
and frequency that will result in the highest alveolar ventilation (see
Figure 1A). The ECMO settings applied will determine how low the power could theoretically
become to reach equivalent CO2 removal. Figure 1B demonstrates the reduction in power
achieved in LIFEGARDS, the EOLIA (Extracorporeal Membrane Oxygenation for Severe Acute
Respiratory Distress Syndrome) trial (8), and the animal experiment by Araos and colleagues
(9), with the goal of near-apneic ventilation. Ultimately, the question remains as
to what creates the best conditions for an organ accustomed to rhythmically expanding
and relaxing: more rest or more movement?
Figure 1.
(A) Mechanical power (MP) normalized per kilogram of body weight delivered during
mechanical ventilation before and after onset of extracorporeal membrane oxygenation
(ECMO) in the LIFEGARDS (Ventilation Management of Patients with Extracorporeal Membrane
Oxygenation for Acute Respiratory Distress Syndrome) and EOLIA (Extracorporeal Membrane
Oxygenation for Severe Acute Respiratory Distress Syndrome) studies, as well as in
the experimental study by Araos and colleagues (9), indicating a reduction (in percent)
of MP attributed to the respiratory rate (RR) or the Vt. (B) We built a model for
an MP (here we use the one delivered during ECMO in the LIFEGARDS study, 6.6 J/min)
and a given dead space (200 ml) to establish the best combination of Vt and RR, with
the aim of maximizing alveolar ventilation. Each column represents the alveolar ventilation
at each different RR (left y-axis), and the light blue line represents the associated
Vt (right y-axis). Positive end-expiratory pressure was kept constant (11 cm H2O)
in this model, as were the airway resistances. bpm = breaths/min.
Schmidt and coauthors did a great job of letting us know where—at least in experienced
centers—we actually are on this issue. The LIFEGARDS study provides a more than solid
basis for us to move forward.