ABSTRACT
This chapter seeks to desirable treatment effects as Efficacy, and undesirable effects as Toxicity. The designs deal with early phase trials in settings where the effects of the treatment regime on Efficacy and Toxicity in humans are not yet understood. One may consider a phase I–II design to be like a Labradoodle, which is obtained by cross-breeding a Labrador Retriever with a Poodle. One may identify three general strategies for defining optimization criteria in phase I–II trials. Each strategy takes a particular approach to quantifying the trade-off between Toxicity and Efficacy, either in terms of their probabilities or the events themselves. The prior always has important consequences in an early phase trial, regardless of how "uninformative" it may appear to be. Toxicity and Efficacy most often are defined as binary events, although more elaborate phase I–II designs may define them as ordinal categorical or event time variables.
