Phase I trial design: are new methodologies being put into practice?

Abstract

BACKGROUND: The primary goal of phase I studies is to efficiently and accurately determine the recommended dose of a new agent for further investigation. Issues of concern ranging from the ethics of these trials to selection of starting dose and rapidity of dose escalation, have led to suggested modifications of the traditional phase I design. We wanted to assess the frequency with which these new approaches are being applied to recent phase I trials and, if possible, their impact. METHODS: Reports of phase I trials of single agent cytotoxics published between 1993 and 1995 were identified by computer search and review of cancer journals. Data on starting dose, toxicology, dose escalation method, definition of dose limiting toxicity (DLT), actual maximum tolerated dose (MTD) and recommended phase II dose were abstracted. RESULTS: Reports of 46 phase I trials were identified: 27 were the first clinical studies of 17 new cytotoxic agents (group A) and 19 were repeat studies of 14 agents (group B). Starting doses in group A were based on preclinical animal toxicology (usually mouse or dog) and for group B on previous clinical experience. Dog toxicology appropriately influenced starting dose in 3 of 6 trials. The majority of group A (19/27) studies employed modified Fibonacci dose escalation; group B studies commonly escalated doses by fixed increments. The definition of DLT was highly variable across studies. MTD was usually defined as the dose level at which > 2/6 patients experienced DLT but several studies required 3-4/6 patients. In 30 trials, the recommended phase II dose was one dose level below the MTD; but in 10 trials the terms MTD and recommended phase II dose were considered synonymous. CONCLUSION: Despite proposed new methodologies (particularly dose escalation) for phase I trials, very few are being employed in practice. A concerted effort should be made to prospectively evaluate these to determine which provides the best combination of safety and efficacy. In addition, the lack of standardization in the definition of limiting toxicity is surprising. Those involved in drug development should strive for agreement on the acceptable degree of toxicity for phase II dose selection.