Design and results of phase I cancer clinical trials: three-year experience at M.D. Anderson Cancer Center.

Abstract

PURPOSE Alternatives to the standard design for conducting phase I trials are proposed with increasing frequency. This study was undertaken to determine how phase I trials are currently conducted and to provide a basis for evaluation of evolving methodology. SUBJECTS AND METHODS All published phase I trials from a single institution over a 3-year period were reviewed to determine the method of selection of the recommended dose for a phase II trial of a new agent, type and extent of toxicity, number of patients treated at the recommended dose, and clinical response. RESULTS All 23 published trials used the standard method of entering cohorts of patients at increasing dose levels and observing toxic effects to determine the dose recommended for phase II study. Among 610 patients, 26% were treated at or within 10% of the recommended dose and 35% were treated with less than 50% of the recommended dose or on a trial that yielded no recommended dose. Among 18 trials using agents previously tested in humans, fewer patients were treated at much less than the recommended dose. For trials in which myelosuppression was dose-limiting, the estimated probability of serious myelosuppression associated with the recommended dose ranged from 23% to 66%. Nineteen patients (3%) responded to therapy. CONCLUSION This summary of phase I trials recently conducted at M.D. Anderson Cancer Center confirms the need for alternative methods, provides baseline information against which alternatively conducted trials can be compared, and demonstrates some practical clinical trial issues not generally considered when alternative methods are proposed.