Purpose: Most current dose-seeking phase 1 trials include an expanded cohort at phase-2-recommended dose (P2RD) to better characterize the drug safety or to obtain a better estimate of secondary endpoints. Nevertheless, the sample size of this expanded cohort has generally not been justified.
Materials and methods: We reviewed 330 phase 1 trials (1998-2008). We estimated the rate of patients experiencing dose-limiting toxicity (DLT) at P2RD. Next, we estimated the probability of observing 1, 2 or 3 DLT in different fictive cohorts (from 8 to 22 patients).
Results: In the literature, the rate of patients experiencing DLT at P2RD was 367/2433, or 15.0%. We drew a table estimating the probability of observing 1, 2 or 3 DLTs in the different fictive cohorts. For example, in a cohort of 16 patients, the probabilities of observing 1, 2 or 3 DLTs are about 92.6%, 91.3 and 91.1% respectively.
Conclusion: This simple tool could provide a justification for the sample size of an expanded cohort when DLT remains the metric for dose-seeking.