RESUMO
BACKGROUND: Evaluation of new therapies for cancer has suffered a paradigm shift in the last years. The use of innovative and more efficient designs is a priority for the scientific community; nevertheless, the use of this kind of design is not yet wide spread. PURPOSE: In this paper will examine the effectiveness of adaptive designs compared with traditional designs in phase II clinical trials. METHODS: We reviewed a group of abstracts records between 1980 and 2008 and extracted data regarding statistical design, year of publication, kind of evaluated product, localization, sample size and results of the trials. RESULTS: Nine hundred and eighty-nine clinical trials were identified and from them 333 traditional designs and 19 adaptive designs were included in the review. Two hundred statistical papers were located and 16 were included in the review. The most frequent designs were Standard up and down designs, continual reassessment methods and its variation and designs with Bayesian approaches. More than 80% of the studies evaluated different schemes of chemotherapy. Adaptive designs evaluated only drugs and not any kind of treatment combination and the most often localizations evaluated in both designs were lung, haematology malignancies, and colon cancers. CONCLUSIONS: Adaptive designs are more efficient from the statistical point of view but they are not yet widely used because of complex and computationally intensive methods needed, substantial effort for planning the trials and lack of regulatory guidance.