Abstract
Methods: A prospective cohort study was conducted, including NMIBC patients who underwent re-TURBT at the Department of Urology, Zhongshan city People's Hospital, from June 2022 to May 2024. Morning urine samples were collected prior to re-TURBT to detect urinary Twist1 methylation, and a 3.0T MRI scan of the bladder was performed for VI-RADS scoring. Based on postoperative pathology results, patients were divided into residual tumor and non-residual tumor groups. Binary logistic regression was employed to identify independent predictors of residual tumor burden prior to re-TURBT. Two predictive models were subsequently developed. The diagnostic performance and clinical utility of these models were assessed using the receiver operating characteristic (ROC) curve and decision curve analysis (DCA). Results: The study ultimately included 52 patients who were initially diagnosed with NMIBC based on pathology. According to the pathological results after re-TURBT, the patients were divided into two groups: the tumor residue group (n=22) and the control group (n=30). Binary logistic regression analysis identified the VI-RADS score and urinary Twist1 methylation as independent predictors of residual tumor burden in NMIBC patients prior to re-TURBT. A predictive model incorporating these factors, along with the presence of visible hematuria within one week before re-TURBT, achieved a sensitivity of 95.45% and a specificity of 83.33% for diagnosing residual tumor burden. ROC curve analysis demonstrated an area under the curve (AUC) of 0.950 (95% CI: 0.884–1.000, P < 0.001). DCA revealed that the model provided a net benefit for threshold probabilities ranging from 0.10 to 0.92. Conclusion: The predictive model combining VI-RADS score, urinary Twist1 methylation, and visible haematuria exhibits excellent diagnostic performance for predicting residual tumour burden in NMIBC patients, offering significant guidance for clinical practice.