Confirmatory Phase III Population Pharmacokinetic Analysis
C. Hu(1), H. Zhou(1)
(1)Pharmacokinetics, Modeling & Simulation, Clinical Pharmacology Sciences, Centocor Research and Development, Inc., Chesterbrook, PA, USA
Objectives: An important objective in population pharmacokinetics analysis is to assess covariate influence. Results from covariate assessments may be used to guide potential dosing adjustment or drug labeling. The commonly used extensive exploratory approach may contain certain biases [1] that can lead to unwarranted dosing adjustments. Recently, we proposed a confirmatory approach aimed for regulatory submission purpose [2]. We herein provide refinements to the proposed approach, and apply it to several more case scenarios to gain further experience with the performance of this approach in contrast to the exploratory approach with phase III clinical trial data.
Methods: A pre-specified primary analysis is proposed based on phases I/II data and the phase III study design, together with two sensitivity analyses similar to [2]. Confidence intervals of covariate effect estimates were proposed as the means for assessing dosing adjustment needs, and refinements of time-adjustment were considered for the linear-model sensitivity analysis. The sensitivity analyses aimed to address potential concerns associated with the primary analysis, which may include misspecification of base and covariate models as well as the lack of robustness due to dosing/sampling time inaccuracies. The approach was applied to several phase III analyses covering different situations of small molecule and therapeutic protein, as well as different sample sizes.
Results: Differences between the proposed confirmatory approach and the exploratory analyses varied, in part depending on sample sizes and the amount of explorations. However, differences in main covariate effects between both approaches were usually small, which was consistent with both theoretical predictions and practical expectations. The proposed confirmatory approach provided a clearer understanding of the uncertainties embedded in the conclusions. The analysis time was also substantially shortened because the amount of exploration was vastly reduced.
Conclusions: The confirmatory approach provides a method that is more accurate in statistical theory, and also gives better understanding of uncertainties and robustness from practical considerations. It is relatively easy to implement with appropriate pre-planning prior to the analysis.
References:
[1] J. Ribbing and E.N. Jonsson. Power, selection bias and predictive performance of the population pharmacokinetic covariate model. Journal of Pharmacokinetics and Pharmacodynamics 2004; 31:109-134.
[2] C. Hu and H. Zhou. An improved approach for confirmatory phase III population pharmacokinetic analysis, Journal of Clinical Pharmacology, 48: 812-822, 2008.
