The American Society of Clinical Oncology held its annual meeting at the start of June and reported several findings from a number of clinical trials investigating new targeted agents. Among these was a Phase III trial of an antibody, ipilimumab, which inhibits cytotoxic T lymphocyte antigen 4 (CTLA4). CTLA4 regulates T cell-mediated immune responses and guards against the development of autoimmunity; ipilimumab improved overall survival in patients with progressive melanoma. These results required the collaboration of several international cancer centres, and the trial was statistically powered to address overall survival. However, as we assess more and more targeted agents in clinical trials there is growing concern that the standard progression from Phase I to Phase III is not the best way to rapidly identify the most promising drugs.

On page 514, Johann de Bono and colleagues discuss the need for better-designed clinical trials to minimise the costly failure of new drugs in Phase III trials. They argue that Phase I trials should be smaller and based on solid preclinical data that identify the mechanism of action of a new drug. Moreover, if a drug is likely to work on only a subset of patients with a particular type of cancer, then these should be the patients who are recruited to the early trials, rather than recruiting all cancer patients with advanced disease and then determining in which patients the drug works best, as has been done in the past. However, as de Bono and colleagues acknowledge, not all drug trials can be updated in this way. Ipilimumab, for example, should boost a patient's immune response to several different types of cancer, but why some patients respond and others do not is not yet known. Therefore, standard progression through Phase I–III clinical trials for such agents remains essential to fully understand the mechanism of action.