[The role of the expansion cohort in phase I trials in oncology: guidelines of the phase I HUB].

At the end of the dose escalation step of phase I trials in oncology, it is increasingly frequent to include patients in expansion cohorts. However, the objective of the expansion cohorts, the number of patients included and their justification are insufficiently explained in the protocols. These cohorts are sometimes of considerable size. The aim of this article is to outline the methodology of expansion cohorts in order to provide recommendations for their planning in practice. This work has been undertaken in collaboration with the statisticians of the early phase investigation centers (CLIP(2)), supported by INCA. First, we have outlined the recent articles published on the expansion cohorts in phase I. We then proposed recommendations, in terms of objectives and number of patients to be included, to guide investigators and facilitate the use of these expansion cohorts in practice. Manji et al. have identified 149 phase I clinical trials using expansion cohorts in oncology with a review of the literature between 2006 and 2011 (Manji et al., 2013). Objectives of the expansion cohort were reported in 111 trials (74%). In these trials, safety was the most reported objective (80% of trials), followed by efficacy (45%). According to this review, the number of patients included in these cohorts was insufficiently justified. This result was confirmed by the study of literature that we conducted over the period 2011-2014. We propose to define the number of patients to be included in expansion cohorts in terms of (1) their objectives, (2) the statistical criteria and (3) the clinical context of the trial. The toxicity study remains the primary objective to evaluate in the expansion phase. In some contexts, the activity study is considered as co-primary objective, either for identifying preliminary signs of activity in studies like screening, or for studying the activity when the target population is known. This study is then considered as phase I/II, and experience plans of phase II can be adapted for planning expansion cohorts. Recommendations for the size of expansion cohorts are proposed. Despite the exploratory character of the expansion cohort, a justification of their size based on assumptions statistically defined is recommended in order to provide an interpretable conclusion and to quantify the risk of errors.