Preoperative Computed Tomography-Based Prediction and Patterns of Lymph Node Metastasis in Renal Pelvis and Ureteral Urothelial Carcinomas.

The accurate preoperative prediction of lymph node (LN) metastasis is essential to optimizing surgical management in renal pelvis urothelial carcinoma (RPUC) and ureteral urothelial carcinoma (UUC). This study evaluates the predictive value of preoperative computed tomography (CT) findings in detecting LN metastasis and determining primary metastatic LN location based on the tumor site. This retrospective study included 48 RPUC and 97 UUC patients who underwent surgery with lymph node dissection (LND) between 2005 and 2023. Preoperative CT images were assessed for tumor size, location, multifocality, peritumoral fat infiltration, hydronephrosis grade, LN status, and metastatic LN location. Logistic regression and receiver operating characteristic (ROC) curve analyses identified predictive factors for LN metastasis, while Pearson's chi-square and Fisher's exact tests determined the association between locations of LN metastasis and primary tumor sites after categorizing UUC into upper and lower UUC. In RPUC, 13 of 48 patients had LN metastasis, with tumor size and peritumoral fat infiltration emerging as significant predictors ( < 0.05). In UUC, 39 of 97 patients had LN metastasis, with tumor size and hydronephrosis grade being significant predictors ( < 0.001). An optimal tumor size threshold of 4 cm was identified for predicting LN metastasis in UUC, and 4.4 cm for RPUC. Additionally, a hydronephrosis grade of 3 or higher was found to be a strong predictor in UUC. ROC analysis showed high accuracy, yielding an AUC of 0.907 in RPUC and 0.904 in UUC. Cross-validation supported the robustness of these findings. Primary LN metastatic sites were predominantly ipsilateral hilar nodes in RPUC and ipsilateral pelvic nodes in lower UUC ( < 0.001). Preoperative CT imaging provides a reliable, noninvasive tool for predicting LN metastasis in RPUC and UUC. Identifying key imaging-based predictors can facilitate risk stratification and surgical decision-making, particularly regarding the necessity and extent of LND.