Abstract / Summary
Background: Osimertinib is the standard of care for epidermal growth factor receptor mutation-positive non-small cell lung cancer, but heart failure hospitalization (HFH) remains a concern, and no practical tool exists to stratify HFH risk during therapy. Methods: In this retrospective cohort study using a Japanese claims database, 5802 patients newly initiating osimertinib were followed for HFH during treatment. Baseline predictors assessed before the index date were selected by 10-fold cross-validated least absolute shrinkage and selection operator (LASSO) logistic regression and internally validated by bootstrapping (B = 500), repeating predictor selection in each resample. Cumulative incidence functions and Fine–Gray models treated death as a competing event; cause-specific Cox models served as sensitivity analyses. Results: HFH occurred in 169 patients (2.9%) during a median on-treatment follow-up of 9.3 months, with 268 competing deaths. LASSO retained seven predictors. The apparent concordance index (C-index) was 0.731 (95% confidence interval, 0.691–0.771) and the optimism-corrected C-index 0.718. Observed incidence increased 10.2-fold across predicted-risk deciles (1.03–10.50%). Fine–Gray and Cox estimates closely matched the logistic model. Conclusions: Baseline cardiovascular vulnerability stratifies HFH risk during osimertinib therapy. The nomogram, not externally validated, is best used as a screening aid to identify patients who may benefit from intensified cardiovascular monitoring.