One of the major concerns for all in vivo experiments is intra- and inter-subject variability, which can be a great source of inaccuracy. The aim of this study is, therefore, to estimate the ability of parallel vs. cross-over design studies in order to describe the relative pharmacokinetic performance of the studied drug formulations. We analyzed the data from a drug development program that examined the performance of innovative abiraterone acetate formulations against the identical reference product in three stages. In stages 1-3, groups A-F were dosed with the reference product once in a parallel manner. Stage 4 was performed to evaluate the intra-individual variability (IIV) by repeated administration of the reference product to the same animals. Although the geometric mean (90% CI) values of abiraterone AUClast in groups A-F were similar to the IIV group (24.36 (23.79-41.00) vs. 26.29 (20.56-47.00) mg/mL·min·g), the results generated in the isolated parallel groups provided imprecise estimates of the true AUClast values ranging from 9.62 to 44.62 mg/mL·min·g due to chance. Notably, in 4 out of 15 possible pair comparisons between the parallel groups, the confidence intervals did not include 100%, which is the true ratio for all comparisons tested after identical formulation administration to all groups. A cross-over design can significantly improve the methodology in short-term comparative pre-clinical pharmacokinetic studies, and can provide more precise and accurate results in comparison to more traditional pre-clinical study designs.
Keywords: abiraterone; cross-over design; in vivo study; pharmacokinetics; rat; variability.