Longitudinal Respiratory Subphenotypes and Differences in Response to Positive End-Expiratory Pressure and FIO2 Ventilation Strategy in COVID-19 ARDS

Abstract

Background: In patients with ARDS, positive end-expiratory pressure (PEEP) titration remains a challenge and recommendations are not in agreement. In mechanically ventilated patients with COVID-19, subphenotypes based on different respiratory trajectories have been identified, but their heterogeneity in response to PEEP/FIO<inf>2</inf> strategy remains understudied. Research Question: Can these previously determined subphenotypes be detected early in the course of mechanical ventilation, and do these subphenotypes moderate the association between PEEP and FIO<inf>2</inf> ventilation strategy and mortality? Study Design and Methods: Retrospective analysis of invasively ventilated patients with COVID-19. Patients were categorized into 2 treatment groups: high PEEP/low FIO<inf>2</inf> strategy and low PEEP/high FIO<inf>2</inf> strategy. To replicate previously described longitudinal respiratory subphenotypes, hereafter named the low-power or high-power subphenotype, a prediction model was created. The primary outcome was the interaction between PEEP/FIO<inf>2</inf> strategy and subphenotype, with mortality as the dependent variable. Results: Of the 1,464 patients included in this analysis, 361 patients (25%) were allocated into the high PEEP/low FIO<inf>2</inf> strategy and 1,103 patients (75%) were allocated into the low PEEP/high FIO<inf>2</inf> strategy. A prediction model consisting of respiratory data of the first 2 days of invasive ventilation (area under the receiver operating characteristics curve, 0.88) assigned 908 patients (62%) to the low-power subphenotype and 556 patients (38%) to the high-power subphenotype. The high-power subphenotype was characterized by higher minute volume, mechanical power, ventilatory ratio, and driving pressure. The association between PEEP/FIO<inf>2</inf> ventilation strategy and ICU mortality was moderated by the subphenotype (P = .03), with high PEEP/low FIO<inf>2</inf> ventilation being associated with lower mortality in the low-power subphenotype (OR, 0.46; 95% CI, 0.31-0.67; P < .001) and not in the high-power subphenotype (OR, 0.85; 95% CI, 0.57-1.28; P = .44). Interpretation: In this study, high PEEP/low FIO<inf>2</inf> ventilation was associated with improved mortality only in one of the subphenotypes, suggesting that such subphenotypes influence heterogeneity of PEEP and FIO<inf>2</inf> effect and should be considered in personalized ventilation strategies. Clinical Trial Registry: ClinicalTrials.gov; No.: NCT05954351; URL: www.clinicaltrials.gov