Implementation of a novel statistical design to evaluate successive treatment courses for metastatic transitional cell carcinoma. A Phase II trial at the M. D. Anderson Cancer Center

Abstract

4543 Background: Patients undergoing chemotherapy often receive several successive courses of treatment whereby physicians use an adaptive strategy to decide whether to repeat the same regimen or switch to an alternate regimen. Although this medical practice is quite common, current clinical trial design does not account for such within-patient treatment decision strategies. Methods: We adopted a novel trial design best described as a ‘play-the-winner-and-drop-the-loser’ algorithm (Statist Med 19:1011, 2000) using 4 different chemotherapy regimens for a total of 12 two-stage strategies. This methodology can be used to select one best treatment, or a best pair of treatments as a two-stage strategy. Patients with metastatic urothelial carcinoma were treated with a 6-week course of therapy and then re-randomized to alternate therapy if they did not have at least a 40% reduction in their tumor. After the second 6-week course, a successful regimen or combination was defined as one which provided a >= 90% susta...