Predicting individual outcomes after radical cystectomy: an external validation of current nomograms.

OBJECTIVE

To determine whether published nomograms, i.e. The International Bladder Cancer Nomogram Consortium (IBCNC) and the Bladder Cancer Research Consortium (BCRC) statistical models for predicting disease recurrence and survival of patients after radical cystectomy, are feasible for routine use in intermediate-volume institutions in Europe, as although these nomograms had high accuracy by internal validation tests, they stem from high-volume centres and have not been validated elsewhere and thus their general applicability remains unproven.

PATIENTS AND METHODS

We externally validated the published nomograms. Information about treatments, pathological details, and recurrence and survival status was retrospectively collected from 246 patients. The expected survival according to the nomograms was calculated. The predictive accuracy of the proposed models was calculated by Harrell's concordance indices. To assess the independent prognostic value of the variables proposed by IBCNC and BCRC, stepwise multivariable Cox regressions based on Akaike's Information Criterion (AIC) for the different endpoints were used. A best model for prediction was created on the basis of our data.

RESULTS

The IBCNC and the BCRC nomograms showed an improvement in the predictive accuracy of recurrence, all-cause and bladder-cancer-specific survival after radical cystectomy of up to 4% compared to Tumour-Node-Metastasis stage-based predictions. According to the calculated AIC values for the different models, all nomograms performed better than the TNM classification.

CONCLUSIONS

The BCRC and IBCNC nomograms provided accurate predictions when they were applied to an external cohort of patients from low- to intermediate-volume centres. The prediction of recurrence and survival based on the addressed nomograms is better than TNM stage-based prediction. The application of such nomograms can be supported on a sound basis, but further amendments are warranted.