Imaging features and prognostic significance of immune checkpoint inhibitor–related pneumonitis in NSCLC

Abstract

Background

Immune checkpoint inhibitors (ICIs) have demonstrated substantial therapeutic efficacy in the treatment of non-small cell lung cancer (NSCLC); however, their clinical application is associated with unique immune-related adverse effects (irAEs). Among these adverse events, immune checkpoint inhibitor-related pneumonitis (CIP) is rare yet serious, which may potentially result in severe respiratory failure, thereby requiring close clinical monitoring. Research specifically focusing on CIP in NSCLC patients treated with PD-1 inhibitors remain limited. This study targets this distinct cohort to comprehensively investigate the clinical and radiological determinants associated with overall survival, applying time-dependent covariate Cox regression to capture the dynamic impact of prognostic factors over time.

Methods

A total of 102 NSCLC participants who received immunotherapy with programmed cell death protein-1 (PD-1) inhibitors and then developed CIP were retrospectively enrolled in this study. Univariate and multivariate time-dependent covariate Cox regression models were constructed to determine associations between CIP features and survival benefits of CIP patients.

Results

The incidence of CIP was 15% (102/680) with a median onset time of 4.6 months. Fifty-one patients (50.0%) were identified as having organizing pneumonia (OP) pattern, followed by nonspecific interstitial pneumonia (NSIP) pattern in 28 patients (27.4%), hypersensitivity pneumonitis (HP) pattern in 6 patients (5.9%), and diffuse alveolar damage (DAD) pattern in 2 patients (2.0%). Additionally, 15 patients (14.7%) were classified as unclassifiable pattern. Kaplan-Meier analysis and Log-rank test indicated that CIP located around the tumor and with reticular opacity were associated with poorer prognosis (P = 0.023, P = 0.013). Compared to those with CIP grades 2–4, patients with CIP grade 1 demonstrated survival benefit with border-line significance (P = 0.049). Multivariate time-dependent covariate Cox regression analysis showed that CIP improvement or not (χ² = 6.81, P = 0.009), percentage of neutrophils (χ² = 24.13, P < 0.001) and albumin (χ² = 31.48, P < 0.001) at the time of CIP diagnosis were independent influencing factors for overall survival (OS) in NSCLC patients with CIP.

Conclusions

CIP without improvement or resolution, a high percentage of neutrophils and elevated albumin level of peripheral blood examination were independent predictors for the prognosis of NSCLC patients, which may have an implication for treatment.