Clinical characteristics, risk factors, and risk stratification of interstitial lung disease during systemic treatment for lung cancer: a multicenter retrospective development and external validation study

Abstract Background Interstitial lung disease (ILD) is a major pulmonary complication of lung cancer treatment. We characterized treatment-associated ILD, identified risk factors, and developed and externally validated a simple risk score using routine clinical variables. Methods This multicenter retrospective cohort included 871 patients (75 ILD events) at the development center, 308 (32) at external validation center, 99 (8) at center 3, and 104 (3) in a 2026 temporal cohort. Logistic regression was performed in the development cohort. A four-factor score assigned one point each for smoking history, any comorbidity, immunotherapy exposure, and inflammatory abnormality. Results Among 1382 patients, 118 developed ILD (8.54%). In development, smoking was 64.0% vs 30.8% (ILD vs non-ILD; OR 3.99, 95% CI 2.41–6.59; P < 0.001); in external validation, 71.9% and 47.1% (OR 2.87, 95% CI 1.28–6.43; P = 0.009). In the four-factor model, smoking (aOR 3.37, 95% CI 1.99–5.69), immunotherapy (aOR 3.77, 95% CI 2.23–6.39), and inflammatory abnormality (aOR 2.25, 95% CI 1.31–3.86) remained independently associated with ILD; comorbidity showed a borderline association (aOR 1.66, 95% CI 0.97–2.84). The score achieved AUCs of 0.771 in development and 0.756 in external validation. ILD incidence reached 29.38% and 25.84% in the high-risk groups. Conclusions Treatment-associated ILD reflects host susceptibility, treatment exposure, and inflammatory activation. Smoking, immunotherapy exposure, and inflammatory abnormality were independently associated with ILD, comorbidity added host-related information. A four-factor routine-variable score showed moderate, stable discrimination in external validation and identified a distinct high-risk group. It supports pretreatment and early-treatment monitoring stratification, not ILD diagnosis.