We studied with lung CT twenty mechanically ventilated patients with COVID-19, mostly with moderate to severe hypoxemia with 10 cmH2O of PEEP. According to the guidelines, these were good candidates for a higher PEEP [1]. Our results can be summarized as follows. First, the individual CT response to increasing PEEP from 10 to 15 cmH2O was extremely variable, ranging from net recruitment to net hyperinflation. Second, net hyperinflation occurred in half of the patients, more commonly those with less severe hypoxemia and higher compliance. Third, different responses to increasing PEEP may be partly explained by different superimposed pressures.
PEEP has been used during mechanical ventilation since the first description of ARDS by Ashbaugh and colleagues in 1967 [12]. During ARDS, pulmonary edema makes lungs heavier. The superimposed pressure, which is the product of density (~ weight) and vertical distance, accordingly increases along the sterno-vertebral axis of the lung. The dorsal regions, where the superimposed pressure is maximal, collapse under the weight of those above [8, 13]. If this non-aerated tissue maintains perfusion, some mixed venous blood will return to the left heart desaturated (venous admixture), and hypoxemia will develop. Moreover, compliance will probably decrease, and PaCO2 will possibly increase because fewer alveoli remain open to ventilation [4]. PEEP counteracts this increased superimposed pressure to keep the dorsal lung regions aerated [13]. According to the above model, oxygenation and compliance should increase, and PaCO2 should decrease with recruitment [14, 15]. Unfortunately, this is not always the case, and recruitment can hardly be predicted or detected from changes in respiratory system mechanics and gas exchange alone [4, 16].
Any PEEP level high enough to recruit the dorsal lung can overdistend the ventral one where the superimposed pressure is minimal, and collapse is rare [17, 18]. Therefore, a reasonable approach to setting PEEP during ARDS can be to search for the best compromise between maintaining recruitment and avoiding overdistention (and a large drop in cardiac output) [19–21]. In this study, we measured the changes in the non-aerated and over-aerated lung volumes when PEEP was increased from 10 to 15 cmH2O. Some hyperinflation occurred in all patients, and exceeded recruitment in half of them. On average, the volume of the over-aerated compartment increased ventrally where, according to the EIT, compliance decreased as for overdistention. Ventilation was then redistributed towards the dependent lung. Of note, these changes occurred despite seemingly protective ventilatory settings [22, 23], with relatively low plateau and driving airway pressures.
Net recruitment was associated with lower compliance (≤ 45 ml/cmH2O) and more severe hypoxemia (PaO2:FiO2 ≤ 123 mmHg). By contrast, net hyperinflation was associated with less profound physiological derangement. These data suggest that a PEEP of 15 cmH2O is preferable in patients with more severe disease, probably more severe than currently recommended (PaO2:FiO2 ≤ 200 mmHg) [1]. As for the others, if lung collapse remains an issue with 10 cmH2O of PEEP, prone positioning may be more indicated as it favors recruitment without hyperinflation [24].
With the limitation of a small sample size, compliance with 10 cmH2O of PEEP was the variable most strongly associated with the individual CT response to increasing PEEP. As shown in Figure E2, and in line with previous knowledge on ARDS unrelated to COVID-19 [10], a higher compliance was associated with a larger normally aerated lung compartment. Net hyperinflation with 15 cmH2O of PEEP could be then readily explained by milder pulmonary disease (with lower superimposed pressure). However, in this and other studies [3, 25], a higher compliance was also associated with a larger over-aerated lung compartment. Herein, in patients with net recruitment, the volume of the over-aerated lung was 129 (36–277) ml with 10 cmH2O and 175 (67–375) ml with 15 cmH2O of PEEP. In those with net hyperinflation, it was 361 (113–589) ml and 507 (228–776) ml. Even so, the latter group had the highest, not lowest, compliance at both PEEP levels. Large hyperinflation without clear signs of overdistention (i.e., a lower compliance) might reflect tissue destruction with loss of elastic recoil, as in chronic emphysema [8]. In line with this hypothesis, new emphysema can be noted in some survivors and non-survivors of COVID-19 [26, 27], and pulmonary air leak is a quite common complication of this disease [28, 29].
Whether patients with COVID-19 should be generally ventilated with a lower or higher PEEP remains controversial. For example, thirty-nine international experts could not reach a consensus on this issue in 2021, as the number of those favoring a lower or higher PEEP was approximately the same [30]. Our present findings support the recommendation by other authors to use a higher PEEP in patients with COVID-19 with caution [31, 32], and possibly only in those with lower compliance [33]. Moreover, they extend the results of previous studies, showing that the response to increasing PEEP differs largely among mechanically ventilated patients with COVID-19 and is frequently less positive than expected for the severity of hypoxemia [3, 34–38]. Nonetheless, our study is novel in that we directly looked at changes in lung aeration rather than inferring them from other (sometimes imprecise [39]) variables, such as changes in regional or global compliance.
Some of the limitations of this work deserve a comment. First, the sample size was small and we did not correct for multiple tests, so our findings should be interpreted cautiously. Second, we enrolled patients early during their ICU stay. In a later phase of the disease, the response to increasing PEEP will be more probably negative even in patients with low compliance [37, 40, 41], due to the possible occurrence of fibrosis. Third, in defining the individual net CT response to a higher PEEP, we treated lung recruitment and hyperinflation as equally good or bad. This approach is questionable [42]. Harms of overdistention might be larger than benefits of recruitment [43], but it is hard to define by how much. Fourth, we did not assess the impact of PEEP on patient-centered outcomes such as mortality.