A clinical risk model predicts over 300 ILD-free days in urological cancer patients on immune checkpoint inhibitors

Abstract Background Immune checkpoint inhibitor (ICI)-induced interstitial lung disease (ILD) is a potentially fatal complication, yet practical risk stratification tools integrating systemic inflammatory markers and pre-existing pulmonary status are lacking in urological cancers. Methods This retrospective, single-center study included 180 patients with urological cancers who received ICI therapy between January 2018 and January 2025. Univariate and multivariate logistic regression using Firth’s penalized likelihood approach identified predictors of ICI-induced ILD. A combined risk model was constructed, and patients were stratified into high- and low-risk groups using the Youden index. Between-group differences in ILD-free time were quantified using the restricted mean survival time (RMST) with a truncation time of 2100 days. Results ILD occurred in 18 of 180 patients (10.0%). In 171 patients with available baseline data, multivariate analysis identified baseline C-reactive protein (CRP; OR 1.11, 95% CI 1.01–1.20, P = .020) and a history of pre-existing ILD (OR 7.86, 95% CI 1.24–53.1, P = .033) as independent predictors, yielding a combined model area under the receiver operating characteristic curve of 0.728. The low-risk group gained over 300 days of additional ILD-free time compared with the high-risk group (RMST difference: 349.2 days, 95% CI 145.8–803.4, P = .005), with consistent findings across cancer types and treatment regimens. Conclusions A two-factor model combining baseline CRP and history of pre-existing ILD enables clinically meaningful risk stratification for ICI-induced ILD in urological cancers. The over 300-day difference in ILD-free time between risk groups has direct implications for monitoring frequency and treatment planning.