Strategies to adjust positive end-expiratory pressure in patients with ARDS

In Reply:

Using esophageal manometry to estimate pleural pressure helps delineate contribution of lung versus chest wall mechanics to clinically observed airway pressures. Transpulmonary pressure provides a truer estimate of distension of aerated lung parenchyma (ie, stress of the ventilated lung regions) than airway pressures alone. In preclinical ARDS models, injury ascribed to atelectrauma can occur from high shear forces generated during repetitive tidal re-opening of fluid-filled or collapsed small airways.¹ Drs Sklar and Goligher argue that chest wall mechanics are of negligible relevance to lung recruitability and individualizing lung-protective PEEP. Yet negative P_L values (in which pleural pressure exceeds airway pressure) predispose to airway closure and maintaining airway pressure above pleural pressure reverses such closure,² lending mechanistic plausibility to protective effects of setting PEEP accordingly in patients at risk of biophysical lung injury.

Whether such a strategy affords clinically relevant lung protection remains unknown. EPVent-2 tested a clinical question: whether titrating PEEP to non-negative P_L was superior to empirical high PEEP for moderate-to-severe ARDS.³ Because resulting end-expiratory P_L values were similar between treatment groups, EPVent-2 unfortunately did not test the underlying scientific hypothesis: whether titrating PEEP to maintain non-negative P_L prevents clinically relevant lung injury. Secondary analyses are underway to address this interest.

Rather than the direct method of P_L estimation used in EPVent-2, Sklar and Goligher advocate for the alternative “elastance-based” method for estimating P_L. However, the elastance-based method was derived using assumptions incompatible with established respiratory physiology and has questionable value for estimating P_L.⁴

Regardless of strategy, increasing PEEP without concomitant reduction in tidal volume might exacerbate overdistension (volutrauma) of aerated lung regions, countering potentially protective effects against atelectrauma. Two possible strategies to overcome this issue are to simultaneously lower tidal volume so that end-inspiratory pressures do not increase, or to restrict applications of higher PEEP only to patients for whom recruited volume mitigates risk of overdistension in non-dependent regions. The latter might be signified by increased lung and respiratory system compliance (lower airway and transpulmonary driving pressures) with higher PEEP, as suggested by Dr Lundin and colleagues. Yet, even if such lung recruitment occurs, higher end-inspiratory pressures could portend volutrauma.

Titrating PEEP by oxygenation response, proposed by Sklar and Goligher, does not mitigate concern for volutrauma. Moreover, gas exchange is not a reliable surrogate for protection against ventilation-induced lung injury. In both the ARDSNet low tidal volume trial and the PROSEVA trial of prone positioning, increased post-intervention PaO₂: FiO₂ failed to predict survival with interventions that have proven survival benefit in ARDS.⁵ Increased oxygenation with higher PEEP may simply identify patients with less severe lung injury in whom impaired oxygenation is driven primarily by PEEP-responsive atelectasis. Whether higher PEEP benefits such patients, who might have less predisposition to biophysical injury, is unclear.

Lundin and colleagues note valid technical limitations, discussed elsewhere,⁴ regarding influence of mediastinal weight on direct estimation of P_L. Previous work by members of our group has demonstrated that, compared with upright positioning, added weight of mediastinal structures with supine positioning contributes mean (SD) 2.9 (2.1) cm H₂O to P_ES.⁶ The small contribution of mediastinal weight to measured P_ES is likely even less with semirecumbent positioning used during measurements in EPVent-2.

In response to Drs Yang and Yuan, racial/ethnic composition in EPVent-2 among the esophageal pressure-guided group was: 75% white, 11% black, 2% Asian, 1% Native American, 12% other/unknown. Among participants assigned to empirical PEEP-FiO₂, 69% were white, 15% black, 4% Asian, 1% Native American, 10% other/unknown. Hispanic ethnicity was reported in 19% of participants assigned to P_ES-guided PEEP and 12% to empirical PEEP-FiO₂. The modest sex imbalance in EPVent-2 is unlikely to be clinically or mechanistically important.