[Accelerated titration design].

To reduce the number of patients treated at low and biologically inactive doses in phase I trials of anticancer agents, attempts to decrease the number of patients per dose level and to conduct a larger dose escalation have been made. Among them, accelerated titration designs were proposed and evaluated by simulation; designs 2 and 4 were reported to be acceptable (J Natl Cancer Inst 89: 1138-1147, 1997). Both designs 2 and 4 included only one patient per cohort during the initial accelerated phase. Dosage steps for the accelerated phase were defined using the modified Fibonacci method for design 2 and 100% escalation for design 4, respectively. The accelerated phase continued until one patient experienced dose-limiting toxicity or two patients experienced grade 2 toxicities. Dose escalation was conducted based on the information from the first course in design 2 and from the first three courses in design 4. In the simulation, both designs successfully reduced the total number of patients and the number of undertreated patients without increasing the number of overtreated patients. However, the safety of design 4 was assured as long as all patients received three courses of chemotherapy, which is unusual in phase I studies in Japan. Decision-making on dose escalation based on the information on toxicity in three courses might be cumbersome. Therefore, in Japan, design 2 would be recommended among the proposed accelerated designs. The performance of the design should be investigated by applying it to actual phase I studies and by evaluating the number of undertreated and overtreated patients.