[Validation of pre-cystectomy nomograms for the prediction of locally advanced urothelial bladder cancer in a multicentre study: are we able to adequately predict locally advanced tumour stages before surgery?].

objectivePre-cystectomy nomograms with a high predictive ability for locally advanced urothelial carcinomas of the bladder would enhance individual treatment tailoring and patient counselling. To date, there are two currently not externally validated nomograms for prediction of the tumour stages pT3-4 or lymph node involvement. MATERIALS AND

methodsData from a German multicentre cystectomy series comprising 2,477 patients with urothelial carcinoma of the bladder were applied for the validation of two US nomograms, which were originally based on the data of 726 patients (nomogram 1: prediction of pT3-4 tumours, nomogram 2: prediction of lymph node involvement). Multivariate regression models assessed the value of clinical parameters integrated in both nomograms, i.e. age, gender, cT stage, TURB grade and associated Tis. Discriminative abilities of both nomograms were assessed by ROC analyses; calibration facilitated a comparison of the predicted probability and the actual incidence of locally advanced tumour stages.

resultsOf the patients, 44.5 and 25.8% demonstrated tumour stages pT3-4 and pN+, respectively. If only one case of a previously not known locally advanced carcinoma (pT3-4 and/or pN+) is considered as a staging error, the rate of understaging was 48.9% (n=1211). The predictive accuracies of the validated nomograms were 67.5 and 54.5%, respectively. The mean probabilities of pT3-4 tumours and lymph node involvement predicted by application of these nomograms were 36.7% (actual frequency 44.5%) and 20.2% (actual frequency 25.8%), respectively. Both nomograms underestimated the real incidence of locally advanced tumours.

conclusionsThe present study demonstrates that prediction of locally advanced urothelial carcinomas of the bladder by both validated nomograms is not conferrable to patients of the present German cystectomy series. Hence, there is still a need for statistical models with enhanced predictive accuracy.