Posttreatment prognostic nomogram for patients with metastatic urothelial cancer completing first-line cisplatin-based chemotherapy

Background: Models to predict the outcome of patients with metastatic urothelial cancer (UC), based on pretreatment variables, have previously been developed. However, patients often request "updated" prognostic estimates based upon their response to treatment. Patients and Methods: Data were pooled from 317 patients enrolled in 8 trials evaluating first-line cisplatin-based chemotherapy in metastatic UC. Variables were combined in the Cox proportional hazards model to produce a nomogram to predict survival from the end of treatment. The nomogram was validated externally using data from a phase III trial. Results: The median survival from end of treatment was 10.65 months (95% confidence interval; 9.20-13.24); 69% of patients had died. Baseline and posttreatment variables were evaluated. Baseline performance status, baseline number of visceral metastatic sites, baseline white blood counts, and response to treatment were included in the final model. The nomogram achieved a bootstrap-corrected concordance index of 0.68. Upon external validation, the nomogram achieved a concordance index of 0.67. Conclusions: A model derived from pretreatment and posttreatment variables was constructed to predict survival from the completion of first-line chemotherapy in patients with metastatic UC. This model may be useful for patient counseling and for stratification of trials exploring "maintenance" therapy.