Association of awake prone positioning with clinical outcomes in severe immune checkpoint inhibitor-associated pneumonitis with acute hypoxemic respiratory failure: a propensity-matched retrospective cohort study

Objective Severe immune checkpoint inhibitor (ICI)-associated pneumonitis can lead to acute hypoxemic respiratory failure and substantial mortality. This study evaluated the association between awake prone positioning (APP) and clinical outcomes in patients with clinically diagnosed severe ICI-associated pneumonitis. Methods In this single-center retrospective cohort study, 120 propensity-matched patients with clinically diagnosed severe ICI-associated pneumonitis and acute hypoxemic respiratory failure between January 2019 and September 2024 were analyzed. Infectious pneumonia, COVID-19 pneumonia, opportunistic infection, tumor progression, radiation pneumonitis, cardiogenic pulmonary edema, pulmonary embolism, diffuse alveolar hemorrhage, and other dominant alternative causes of respiratory failure were excluded through structured diagnostic adjudication. Patients were classified into the APP group ( n = 60, cumulative APP duration ≥10 h during the first ICU week) or control group ( n = 60) according to documented positioning exposure. Primary outcomes were dynamic PaO₂/FiO₂ ratio trajectories and endotracheal intubation rates; secondary outcomes included ICU length of stay and 28-day all-cause mortality. Sensitivity analyses addressed pandemic period, APP exposure timing, FiO₂ documentation, and competing risk for ICU discharge. Results APP was associated with a higher PaO₂/FiO₂ ratio at 168 h (268 ± 52 vs. 201 ± 49 mmHg, p < 0.001) and a lower intubation rate [11/60 (18.3%) vs. 27/60 (45.0%), p = 0.001; adjusted OR = 0.24, p = 0.002]. ICU stay was shorter in the APP group (8.5 vs. 12.3 days, p < 0.001). The APP group also had lower observed 28-day mortality than the control group [11/60 (18.3%) vs. 21/60 (35.0%), p = 0.039]. Findings were directionally consistent in sensitivity analyses, although effect estimates were attenuated in APP initiation-based and landmark analyses. Protocol-defined adverse events were comparable between groups. Conclusion In this retrospective propensity-matched cohort study, APP was associated with improved oxygenation and lower intubation rates in patients with clinically diagnosed severe ICI-associated pneumonitis. Associations with shorter ICU stay and lower observed 28-day mortality should be interpreted cautiously because of the observational design, potential residual confounding, and possible survivor or tolerance bias.