Prognostic factors in second-line treatment of urothelial cancers with gemcitabine and paclitaxel: German Association of Urological Oncology (AUO) trial AB 20/99.

Abstract

4586 Background: Clinical response in second-line treatment of urothelial cancers after cisplatin-failure may differ substantially. Progressive disease during treatment with survival of only a few weeks as well as long-term survival of several years is possible. Aim of this analysis was to validate the prognostic model for survival developed by Bellmunt et al. in a different patient-cohort. Methods: Baseline parameters (BPs) of 102 patients (pts.) derived from the largest phase III trial of second-line chemotherapy with gemcitabine and paclitaxel (GP) were compared to the phase III cohort of vinflunine versus best supportive care (n = 370). Uni- and multivariate analyses of BPs with respect to overall survival (OS) were performed. Pts. were stratified according to the Bellmunt model (ECOG > 1, hemoglobin < 10 g/dl, hepatic disease) and altered models. OS of subgroups was compared by log-rank test. Accuracy of the different models to identify subgroups with favorable and poor survival was tested by ROC-analysis. Results: BPs were comparable. The identification of prognostic subgroups by Bellmunt with significant difference in OS could be validated (11.8 [95%CI 6.3-17.3], 8.1 [95%CI 4.8-11.4], 3.2 [95%CI 0.0-7.9] months, P=.007). The prediction of poor (AUC=0.660, P=.023) as well as favorable (AUC=0.634, P=.036) prognosis was possible. A different hemoglobin cut-off value (<12 g/dl) was identified as a poor prognostic factor in a multivariate analysis (HR 1.485, P<.001) and resulted in a better discrimination of pts. OS (14.1 [95% CI 7.5-20.7], 10.9 [95% CI 4.9-16.9], 5.7 [95% CI 4.6-6.8], 3.2 [95% CI 1.5-3.1] months, P<.001). However, apart from the Bellmunt parameters lymph-node-only disease was the strongest predictor of prolonged OS and overruled every other single Bellmunt parameter in a multivariate analysis (HR 0.506, P=.001). Conclusions: The Bellmunt risk factor model was validated in the GP cohort. Lymph-node-only disease was identified as a favorable factor for OS which overrules the prognostic model. Therefore, in these patients every treatment option should be taken into account, even if anemia or impaired performance status is present.