Personalized mechanical ventilation guided by lung ultrasound in patients with ARDS: a pilot phase of a randomized clinical trial

Background: The “Personalized Mechanical Ventilation Guided by Lung UltraSound in Patients with Acute Respiratory Distress Syndrome” (PEGASUS) study aims to evaluate personalized mechanical ventilation (MV) in patients with acute respiratory distress syndrome (ARDS) compared to the standard of care. However, misclassification and misaligned MV strategies were shown to be harmful. We therefore aimed to assess the interobserver agreement of lung ultrasound (LUS) between the local investigator and an expert panel in classifying ARDS subphenotypes alongside protocol adherence and safety endpoints, as a pilot phase of the ongoing PEGASUS study. Methods: The first 80 mechanically ventilated patients with moderate-to-severe ARDS were enrolled in the ongoing PEGASUS study, a randomized clinical trial (RCT), and were included in the pilot phase. Focal or non-focal subphenotypes were classified using a LUS. Positive end-expiratory pressures (PEEP), tidal volumes (VT), the application of recruitment manoeuvres, and proning were performed according to randomization arm and subphenotype. Safety limits for MV followed current guidelines. Agreement in subphenotype classification between local investigators and a panel of three experts was evaluated using Cohen’s κ coefficient. Results: In 68 out of 80 exams, the images were of sufficient quality for assessment. The interobserver agreement for the lung morphology had a Cohen’s kappa of 0.72 (95% CI 0.53–0.9) and accuracy of 88% between local investigator and the expert panel. Misclassification occurred in 8/68 exams (11.8%). Among these 8 misclassified cases, 6 (75%) also showed disagreement between experts due to different LUS scores of the anterior regions. Tidal volume and PEEP were generally set according to the protocol. An exception was the TV in the non-focal ARDS patients randomized to personalized MV, where the median (6.2 ml/kg/PBW) was above the target range (4–6 ml/kg/PBW). Patients exceeding safety limits of MV were low. Conclusion: In the pilot phase of an ongoing international subphenotype-targeted RCT, we found that local investigators’ assessments agreed with expert panel consensus assessments in the large majority of cases, and nearly always when the expert panel assessment was unanimous. Protocol adherence was sufficient, but tidal volume in the non-focal subphenotype deserves attention during continuation of the study. Trial registration: The study was registered on clinicaltrial.gov (ID: NCT05492344, date 2022-08-05).