Abstract
Background Despite being a groundbreaking cancer therapy, immune checkpoint inhibitors (ICI) can lead to potentially life-threatening toxicity with checkpoint inhibitor pneumonitis (CIP). While treatable, it is easy for clinicians to miss the symptoms of CIP, which can lead to a delay in diagnosis and worsening respiratory function. There is no consensus approach to systematically identifying patients at risk of developing CIP. Thus, we sought to create a workflow that could inform patient selection for ICI therapy based on previously reported risk factors for CIP development. Materials and methods We retrospectively identified 250 patients with lung cancer treated with at least one dose of an ICI over 20 months. Data were collected on comorbidities, cancer type and stage, performance status, ICI cycles, biomarkers, prior curative treatment, diagnostic evaluation, antibiotics, steroids, progression, and survival. A single-blinded radiologist characterized radiographic patterns of suspected CIP cases. Results Among 97 patients who received steroids while admitted to the hospital, 12 (6%) had at least one sign or symptom suggestive of CIP. Chronic obstructive pulmonary disease and non-small cell lung cancer subtypes correlated with suspicion of having CIP. CIP was confirmed in five patients (42%) and ruled out (mimics) in seven (58%). Median times until symptoms were 17 months and one month for confirmed and mimic cases, respectively. The median time to confirm or exclude CIP was 5 ± 4 days. Most suspected cases underwent thoracic imaging, blood cultures, and empiric antibiotics. Radiographic patterns in suspected cases included ground glass opacities, organizing pneumonia, acute interstitial pneumonia/acute respiratory distress syndrome, bronchiolitis, radiation recall pneumonitis, hypersensitivity pneumonitis, and post-radiation fibrotic changes. Conclusions CIP mimics are common in clinical practice; therefore, it is reasonable to empirically treat suspected cases with shorter courses of steroids until diagnostic clarity is achieved. This proof-of-concept study demonstrates that this novel workflow can identify the true incidence of CIP, inform treatment decisions, and lead to the development of implementation studies to improve patient care directly.