Machine learning models can accurately predict survival outcomes after surgery for upper tract urothelial cancer, providing a potential tool to guide the intensity of postoperative management and follow-up. A large international analysis shows that data-driven models may help identify patients most likely to benefit from close monitoring rather than immediate adjuvant chemotherapy after radical nephroureterectomy.
Unmet needs in postoperative risk stratification
Upper tract urothelial cancer is generally stratified into high-risk and low-risk categories, and radical nephroureterectomy and bladder cuffing are considered standard treatments for nonmetastatic high-risk disease. However, there is currently no consensus regarding postoperative management, especially regarding close follow-up and selection of patients for adjuvant chemotherapy. Existing clinical tools provide limited guidance and create uncertainty in postoperative treatment planning.
Research design and machine learning approaches
Researchers retrospectively collected data from a large, multiethnic cohort of 3,129 patients who underwent radical nephroureterectomy for histologically confirmed upper tract urothelial carcinoma at institutions in Asia and Europe. A total of 637 Asian patients formed the training cohort and 2,492 European patients were used for external validation. Twenty supervised machine learning models were trained and tested to predict 3- and 5-year overall survival, cancer-specific survival, and disease-free survival.
A nomogram was constructed using eight independent prognostic factors: age, sex, tumor grade, pathological tumor stage, pathological nodal status, presence of carcinoma in situ, multifocality, and lymphovascular invasion. Model performance was evaluated using the area under the receiver operating characteristic curve.
Predictive performance and clinical significance
During model training, the logistic regression-based approach achieved the strongest performance, ranking highest in four out of six outcomes. The highest predictive accuracies were for 3-year and 5-year cancer-specific survival with area under the curve values of 0.85 and 0.84, respectively, and for 3-year disease-free survival with an area under the curve of 0.81. In external validation, the logistic regression model continued to perform well, ranking first in three of six outcomes, including 3-year cancer-specific survival with an area under the curve of 0.84.
The results of this study suggest that machine learning can provide robust prognostic estimates after surgery for upper tract urothelial cancer and may explain epidemiological differences between European and Asian populations. Although further clinical validation is required, these models could support individualized decision-making regarding adjuvant therapy and follow-up strategies in daily practice.
reference
Nicoletti R et al. Training and external validation of a machine learning supervised prognostic model for upper tract urothelial carcinoma (UTUC) after nephroureterectomy. scientific report. 2026;https://doi.org/10.1038/s41598-025-29043-w.
