Optimizing first-time-in-human trial design for studying dose proportionality

Mixed effect models are becoming common in analyzing data from clinical trials involving several measurements for each subject. However designs using mixed effect models have not received as much attention. Data collected from first-time-in-human studies are frequently used for studying dose proportionality. Two design types were evaluated theoretically for a hypothetical study conducted at six doses evenly spaced on a log scale for a drug whose AUC-dose relationship can be described by a mixed effect power model. In sequential panel design, subjects receive consecutive doses and in alternate panel design, subjects receive nonconsecutive doses. Sample sizes required to characterize the AUC-dose relationship to the same level of precision by both designs were calculated at given inter- and intra-subject variabilities. The conclusion is that an alternate panel design always requires fewer subjects than a sequential panel design. In many common cases, less than half as many subjects are required.