A pretreatment nomogram to predict muscle-invasiveness in high-risk upper tract urothelial carcinoma (ROBUUST 2.0 collaborative group)

Abstract

BACKGROUND: The ability to predict muscle invasion in the final pathology of upper tract urothelial carcinoma (UTUC) patients after radical nephroureterectomy (RNU) potentially influences the selection of the most appropriate treatment modality. The present study aims to develop a model predicting muscle-invasive status in high-risk UTUC. METHODS: The ROBUUST (RObotic surgery for Upper tract Urothelial cancer - UTUC - STudy) 2.0 dataset is an international, multicenter registry of patients undergoing curative surgery for UTUC between 2015 and 2022. Data about high-risk patients, classified according to EAU and NCCN prognostic stratification criteria, who underwent RNU were retrieved. The primary outcome was the identification of muscle-invasiveness. Two multivariable models, differing in the inclusion of biopsy-related data, were fitted with pT stage results at final pathology. Their predictive ability was calculated using the area under the receiver operating characteristic curve and decision curve analysis (DCA). A nomogram was developed using the model demonstrating the highest area under the curve (AUC) and clinical net benefit. RESULTS: In the overall cohort, 1558 patients met the inclusion criteria, with 934 patients having ≥pT2 disease. Patients in the ≥pT2 cohort had significantly worse oncological outcomes in terms of metastases, all-cause, and cancer-specific deaths (all P<0.001). The biopsy-related model had the highest AUC (74%) and the highest net benefit in DCA. The DCA showed an improvement in the clinical risk prediction of muscle-invasiveness, and a reduction in the number of upfront or unnecessary RNU, at every ≥pT2 probability threshold. CONCLUSIONS: The proposed prognostic model is a valuable tool for estimating the risk of muscle-invasiveness in high-risk UTUC patients, owing to its optimal predictive ability and user-friendly design.