C104-11 Severe Immune Checkpoint Inhibitor Pneumonitis: A Multicenter Cohort Study of Clinical Predictors, Management Patterns, and Outcomes in Hospitalized Patients

Background Pneumonitis has the highest mortality among all immune related adverse events (irAEs) from immune checkpoint inhibitors (ICIs), which are used to treat a wide range of malignancies. Evidence remains limited as to which clinical markers or pre-treatment risk factors predict severe disease. Methods Patients were identified from a registry of patients who received immune checkpoint inhibitors at any University of North Carolina Health affiliate and were hospitalized for severe immune checkpoint inhibitor pneumonitis (ICI-P) between January 2013 and December 2023. We performed a systematic evaluation of clinical variables, lung function testing, imaging findings, laboratory testing, hospitalization management and vital status. Results Among the 90 patients with severe ICI-P requiring hospitalization, the average age was 67 and 51% were male. 38% (n = 34) of patients had documented pre-treatment pulmonary function testing. The average hospital length of stay for our cohort was 10 days (SD ± 8). Bronchoalveolar lavage (BAL) was performed in 27% (N = 24) of patients, 46% of which had >20% lymphocytes on their BAL. Almost all patients were treated empirically with antibiotics for infection. The mean interval from initial CT scan to initiation of steroids was 0.9 days (SD ± 2.6), and the mean steroid dosing was 1.92 mg/kg/day (SD ± 1.6). Inpatient mortality was 32% (n = 29). Preexisting Interstitial Lung Disease (ILD) (p = 0.010), Interstitial Lung Abnormality (ILA) (p < 0.001), oxygen support level (p = 0.003) and grade of pneumonitis (p = 0.004), were associated with longer lengths of stay (+13, +10, +2 per level increase and +7 days per higher grade level, respectively). For every day of increased time between index CT and steroid initiation, logistic regression revealed a 39% increase in hospital mortality (OR 1.39 CI=1.08 -1.79, p = 0.011). Findings of consolidations on CT appeared to be protective, although not statistically significant (inpatient mortality OR 0.48, CI 0.18- 1.27, p = 0.14). Conclusion Our study illustrates the high mortality associated with severe ICI-P requiring hospitalization and the need for prompt evaluation and treatment. Those with pre-existing ILD and ILA are at risk for longer hospitalizations, and our data suggests that a delay in steroid initiation (first line treatment for ICI-P) may increase odds of death during hospitalization. This study highlights the need for a consensus on diagnostic testing and threshold to treat when there is a high suspicion for ICI-P.