This study aimed to develop a triphasic contrast-enhanced CT-based habitat imaging method for preoperative prediction of World Health Organization/International Society of Urological Pathology (WHO/ISUP) grade in clear cell renal cell carcinoma (ccRCC). A retrospective analysis included 300 ccRCC patients from two centers. Center 1 data were used for training (n = 190) and internal validation (n = 82), and Center 2 for external validation (n = 28). All patients underwent triphasic CT scans. Tumor volumes of interest (VOIs) were manually delineated using 3D Slicer. CT values from the corticomedullary (CMP), nephrographic (NP), and excretory (EP) phases were extracted to assess enhancement. K-means clustering segmented tumors into four habitats, and volume fractions were calculated. Logistic regression identified significant predictors from habitat features and clinical variables. A nomogram was constructed and evaluated using receiver operating characteristic (ROC) curves, area under the curve (AUC), calibration curves, Hosmer-Lemeshow (HL) tests, and decision curve analysis (DCA). Gender, tumor size, and the volume fractions of Habitat 1 (F1) and Habitat 2 (F2) were independent predictors. These predictors were integrated into a nomogram that achieved AUCs of 0.794 (95% CI, 0.726–0.862) in the training cohort, 0.787 (95% CI, 0.678–0.897) in the internal validation cohort, and 0.781 (95% CI, 0.599–0.962) in the external validation cohort. The model showed acceptable calibration and yielded potential net clinical benefit in both validation sets. We developed and externally validated a CT-based nomogram for preoperative WHO/ISUP grade stratification in ccRCC; larger independent cohorts are needed to confirm generalizability.
Zhang et al. (Fri,) studied this question.