A prognostic nomogram based on log odds of positive lymph nodes for non-metastatic upper tract urothelial carcinoma after radical nephroureterectomy with lymph node dissection

To evaluate the prognostic value of LODDS and to develop nomograms for overall survival (OS) and cause-specific survival (CSS) in patients with non-metastatic upper tract urothelial carcinoma (UTUC) after radical surgery. A retrospective cohort of 1,970 patients from the SEER database (2004–2022) was randomly split into training (70%) and validation (30%) cohorts. Separate multivariable Cox regression models compared LODDS, N classification, positive lymph node count (PLN), and lymph node ratio (LNR), each entered in an independent model alongside the same set of clinical covariates. Optimized OS and CSS models were built using Cox regression. Time-dependent AUC, calibration, and decision curve analysis (DCA) assessed nomogram performance. Age, T stage, and the lymph node indicators (N, PLN, LNR, LODDS) were independent predictors of both OS and CSS. Grade was an independent predictor of CSS in every model; in the LODDS-based OS model grade was not statistically significant (HR 1.28 (0.98–1.68), p = 0.073) and was therefore excluded per the pre-specified rule, so the OS nomogram comprises age, T stage, and LODDS, whereas the CSS nomogram retains grade. Among the four lymph node metrics compared, the LODDS-based model yielded the lowest Akaike information criterion (AIC); however, the absolute differences in discrimination were small, and the C-index was numerically slightly higher than that of N classification for OS but identical to it for CSS. In repeated 10-fold cross-validation, the absolute differences in discrimination were small and, after correction for overlapping folds, not statistically significant (OS corrected p = 0.225, CSS corrected p = 0.656), indicating that LODDS was numerically slightly higher for OS (ΔC-index + 0.002) and essentially identical for CSS (ΔC-index − 0.001), but was not statistically distinguishable from N classification. A Fine-Gray competing-risk analysis for CSS corroborated these associations. LODDS provided discrimination that was comparable to—though numerically slightly higher than—traditional lymph node metrics and may serve as a complementary prognostic marker in non-metastatic UTUC. The nomograms offer individualized survival estimates to aid prognostic counseling. External validation and biologically motivated refinement are warranted before any clinical implementation.