Background: The definition of the optimal dose of a new compound and the design for identifying this dose have been set up for treatment with cytotoxic activity. The optimal dose is typically the Maximum Tolerated Dose (MTD) defined by the occurrence of severe toxic side effects during the first course of treatment. In the era of molecularly targeted therapies (MTA), where the mechanistic relations between the dose, the activity and the toxicity are probably different, how relevant these definitions and designs are?
Methods: We present here a review of the outcomes of potential interest for defining the optimal dose and we present several statistical innovative dose finding methods. Longitudinal data (tumour growth, repeated evaluation of toxic side effects) and continuous outcomes appear particularly promising for early phase trials.
Conclusions: Phase I dose finding method for molecularly targeted agents should continue to identify the MTD, but the dose recommended for phase II trials should not be systematically defined as the MTD and should incorporate other endpoints. In particular, more sensitive and more discriminatory endpoints are necessary. They needn't be good surrogate of the clinical benefit, but they should allow for ranking several doses. Experience from other medical specialties, such as phase II dose-ranging trials could be considered.