Predictors of Immune-Related Pneumonitis in Gastric Cancer Patients Receiving Immune Checkpoint Inhibitors: Model Development and Internal Validation in a Single-Center Cohort

Objective To identify peripheral blood predictors of CIP in gastric cancer patients treated with ICIs and develop a risk prediction model based on routine inflammatory and immune parameters. Methods In this single-center retrospective cohort study, 224 gastric cancer patients undergoing perioperative therapy were enrolled, including 112 in the chemo-immunotherapy group and 112 in the chemotherapy-only group. Peripheral blood markers-such as CRP, LDH, NLR, PLR, eosinophils, T-cell subsets, and cytokines—were analyzed. Multivariate logistic regression identified independent predictors, and a risk prediction model was constructed and internally validated using ROC and decision curve analysis. Results The incidence of CIP in the chemo-immunotherapy group was 12.0%, with a median onset of 47 days. Patients who developed CIP exhibited significantly higher CRP, LDH, NLR, and IL-6 levels, and a lower CD4/CD8 ratio. Multivariate analysis identified NLR, CD4/CD8 ratio, and IL-6 as independent predictors. The combined model achieved an AUC of 0.887 with good calibration and clinical utility. A nomogram and simplified scoring table were developed for clinical application. Transcriptomic analysis indicated that differential genes associated with ICIs were mainly enriched in the HIF-1 signaling and oxidative stress pathways. Survival analysis showed that patients receiving chemo-immunotherapy had superior overall and progression-free survival compared to chemotherapy alone. Conclusion Elevated NLR, decreased CD4/CD8 ratio, and increased IL-6 are independent predictors of CIP in gastric cancer patients treated with ICIs. The developed risk prediction model demonstrates robust performance and practical value for early identification and individualized management of CIP.