Evaluation of human trial design.

Randomized trials are considered the gold standard for evaluating chemopreventive agents. Compared with clinical trials, some specific design issues arise, requiring appropriate solutions: choice of chemopreventive agent(s); dose and chemical formulation of agent; choice of population; trial design; use of 'run-in phases' and pilot trials; sample size; monitoring of compliance; and duration and completeness of follow-up. Choice of the population, dose and length of intervention depend on the mechanism of action of the chemical(s) under study and its relationship with the natural history of the disease. The relationship between given and effective dose at a cellular level may also represent an efficacy determinant. A factorial trial design may be chosen on the basis of costs and of possible interaction between agents. Run-in phases and pilot studies are an efficient method for enhancing compliance and feasibility, although selection of good compliers may influence the expected frequency of outcome(s). Possible changes of exposure to the chemopreventive agent in the control group, due to spontaneous trends in the population, can reduce the power of the study and should be carefully monitored, together with compliance and side-effects. Length of follow-up and choice of validated end points represent crucial decisions in terms of validity of results and of their relevance for cancer prevention.