An interpretable LASSO-Boruta nomogram for non-muscle-invasive bladder cancer recurrence after intravesical...

Recurrence remains a major clinical challenge in non-muscle-invasive bladder cancer (NMIBC), highlighting the need for interpretable tools for individualized risk assessment. This study aimed to develop a transparent model that estimates recurrence risk in patients with NMIBC after transurethral resection of bladder tumor (TURBT) and intravesical instillation.

We reviewed 200 patients with NMIBC who received TURBT and postoperative intravesical instillation at the Department of Urology, The Affiliated Hospital of Xuzhou Medical University, from January to December 2020. Recurrence status was determined from follow-up records. We collected baseline variables, preoperative laboratory results, and clinicopathological data. Patients were divided into training and validation sets (7:3). We screened predictors with multivariate Cox regression, least absolute shrinkage and selection operator (LASSO)-Cox regression, and Boruta analysis before building the nomogram. Model performance was checked with receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis (DCA). Shapley additive explanations (SHAP) analysis was used for interpretation, and an R Shiny application was built.

Of the 200 patients, recurrence was observed in 90 cases. The final identified predictors were smoking, hypertension, hemoglobin (Hb), and serum Na. The nomogram showed stable performance, with time-dependent areas under the curves (AUCs) of 0.862, 0.918, and 0.948 in the training set and 0.725, 0.870, and 0.919 in the validation set at 12, 36, and 60 months, respectively. SHAP analysis ranked Hb as the main contributor; higher Hb and serum Na indicated lower recurrence risk, while smoking and hypertension indicated higher risk.

This model provided internally validated recurrence-risk estimates; external validation and usability testing are required before clinical use.