Reply by Pan et al. to Haouzi et al.

From the Authors:

We appreciate Dr. Haouzi and his colleagues for their interest in our research letter (1). They reanalyzed our reported data and found a possible but nonsignificant coupling between lower compliance and greater alveolar Po2 (PaO)–PaO gradient. They then suggested that positive end-expiratory pressure (PEEP) should be titrated by reaching the highest compliance and lowest PaO–PaO gradient.

We want to point out that a possible association between compliance and PaO–PaO gradient among different patients makes physiological sense but may not be applied for PEEP titration in a given individual; the PEEP providing the highest compliance can be completely different from the PEEP providing the lowest PaO–PaO gradient. Actually, we have observed that patients with coronavirus disease (COVID-19)–associated acute respiratory distress syndrome (ARDS) from Wuhan often present “better” compliance and “worse” PaO–PaO gradient at low PEEP. We thus will discuss the optimal compliance and the optimal PaO–PaO gradient as two respective PEEP strategies.

Titrating PEEP by the optimal compliance has been proposed for several decades, but years of research have shown many pitfalls and limitations. 1) Plateau pressure can be measured by performing varied durations of end-inspiratory occlusion, and the pressure value can change according to viscoelastic properties, pendelluft, or simply the presence of leaks. This technical issue is not trivial. A preset 0.2- to 0.3-second end-inspiratory pause minimizes this issue, providing more reliable plateau pressure as an indicator of the maximal lung distension (2). 2) Some physiological studies using electrical-impedance tomography suggested that a high PEEP guided by “best” compliance of the whole respiratory system can be substantially higher than the PEEP based on regional compliance or on the dorsal fraction of ventilation reaching 50% and that the chest wall could play a role in these discrepancies (3). 3) In contrast, when substantial tidal recruitment is present at low PEEP, compliance may be increased by this tidal recruitment (4). Using this “best” compliance would therefore favor ongoing repeated recruitment and collapse. 4) The optimal compliance approach has been tested in a large randomized controlled trial, showing no benefit on outcome (5).

The PaO–PaO gradient can be a useful physiological indicator during clinical practice, but we cannot rely on it for PEEP titration because of the following considerations. 1) Calculating the PaO–PaO gradient by using a simplified alveolar gas equation assumes that the respiratory quotient is 0.8 in all patients. This might not be the case in our critically ill patient with ARDS. Particularly, three patients in our study (1) received extracorporeal membrane oxygenation, an intervention that can remarkably alter the respiratory quotient. 2) For a given FiO, titrating PEEP by reaching the lowest PaO–PaO gradient is similar to seeking the “best” oxygenation but can be misled by an increase in PaCO. In other words, both an increase in the PaCO2 and an increase in PaO can reduce the calculated PaO–PaO gradient. 3) Although it may be practically preferable to assume the PaO–PaO gradient represents the shunt in patients with ARDS, optimizing PEEP by reaching the lowest shunt can be misdirected by a reduction in cardiac output. Indeed, classical studies have shown that higher PEEP often reduces cardiac output and then reduces the intrapulmonary shunt (6). Improved PaO and PaO–PaO gradient can then be observed with higher PEEP, but the price may be a lower oxygen delivery to the tissues because of the reduction in cardiac output.

For these multiple reasons, we prefer to rely on the direct assessment of lung recruitability to guide the PEEP setting. Our primary goal is to reduce the risk of ventilator-induced lung injury, such as atelectrauma and overdistension. Our study has demonstrated that the assessment of lung recruitability is feasible at the bedside by only using the ventilator of the patient, even in very constrained situations like COVID-19 (1).