Abstract
<title>Abstract</title> <p>Background Accurate preoperative staging of upper tract urothelial carcinoma (UTUC) remains challenging. This study aimed to develop site-specific scoring systems for predicting muscle-invasive UTUC without ureteroscopic findings. Methods This retrospective multicenter study initially included 386 patients who underwent radical nephroureterectomy for UTUC at Tottori University Hospital and affiliated hospitals between January 2015 and December 2021. Patients with clinically node-positive disease or those who received neoadjuvant chemotherapy were excluded. Preoperative variables included age, sex, body mass index, hydronephrosis, urinary cytology, clinical T stage, tumor size, and multifocality. Univariable and multivariable logistic regression analyses were performed separately for ureteral and renal pelvic tumors to identify the predictors of muscle-invasive disease. Results Among 356 eligible patients, 157 (44.1%) had muscle-invasive UTUC. Multivariable analysis identified positive urinary cytology and clinical T stage as independent predictors of muscle-invasive disease in both ureteral and renal pelvic tumors, whereas hydronephrosis was an additional independent predictor in ureteral tumors. Scoring systems incorporating these variables achieved areas under the curve of 0.865 for ureteral cancer and 0.745 for renal pelvic cancer. Optimal cutoff scores based on the Youden index were ≥ 3 for ureteral tumors and ≥ 2 for renal pelvic tumors, with sensitivity rates of 85.4% and 68.0% and specificity rates of 77.4% and 73.6%, respectively. Conclusion Site-specific scoring systems based on routinely available preoperative variables accurately predicted muscle-invasive UTUC without requiring ureteroscopic findings. These models may facilitate preoperative risk stratification, although external validation is required.</p>