Box 53 Use of a twostage optimal Simon design in a randomized phase IIIII trial

Standard chemotherapy for metastatic germ cell cancer is the BEP (bleomycin, etopos-ide, cisplatin) combination. A phase I study suggested high activity levels if the drug paclitaxel was added to this regimen [21]. As this was a relatively rare group of patients, an efficient design was chosen to enable a combined phase II/III design to be carried out. Patients were to be randomized between BEP and paclitaxel-BEP (T-BEP) in several stages. The number of patients in the first two stages was determined using an optimal Simon two-stage design with complete response (CR) rate as the primary outcome measure. The expected CR rate with BEP in this group of patients was 65% (P1). The minimum activity level of interest for T-BEP was 80% (P2). With a = 0.1 and p = 0.05, the first step required forty-two patients in the T-BEP arm (an equal number were to be randomized to the BEP arm). If fewer than twenty-nine patients on T-BEP achieve a CR, the trial would stop. Otherwise it would continue until there were eighty-two patients on the T-BEP arm (again randomizing an equal number to BEP). If fewer than fifty-nine patients on T-BEP achieve a CR, the trial would stop. Otherwise, it would continue as a phase III trial to which all the patients randomized so far would also contribute. The primary outcome measure of the phase III trial would be progression-free survival, and it would be powered to detect a 10% increase in progression-free survival.

0 0

Post a comment