Purpose: The variation in reported survival of patients with metastatic transitional-cell carcinoma (TCC) treated with systemic chemotherapy may be a consequence of pretreatment patient characteristics. We hypothesized that a prognostic factor-based model of survival among patients treated with methotrexate, vinblastine, doxorubicin, and cisplatin chemotherapy could account for such differences and help guide clinical trial design and interpretation.
Patients and methods: A database of 203 patients with unresectable or metastatic TCC was retrospectively subjected to a multivariate regression analysis to determine which patient characteristics had independent prognostic significance for survival. Patients were assigned to three risk categories depending on the number of unfavorable characteristics. Patient selection in phase II studies was addressed by developing a table of expected median survival for patient cohorts that had varying proportions of patients from the three risk categories.
Results: Two factors had independent prognosis: Karnofsky performance status (KPS) less than 80% and visceral (lung, liver, or bone) metastasis. Median survival times for patients who had zero, one, or two risk factors were 33, 13.4, and 9.3 months, respectively (P =.0001). The median survival time of patient cohorts could vary from 9 to 26 months simply by altering the proportion of patients from different risk categories.
Conclusion: The presence of baseline KPS less than 80% or visceral metastasis has an impact on survival. Reporting the proportion of patients with zero, one, and two risk factors will facilitate understanding of the relevance of the median survival in phase II trials. Phase III trials should stratify patients according to the number of risk factors to avoid imbalance in treatment arms.