Bootstrap method-based estimation of the minimum sample number for obtaining pharmacokinetic parameters in...

Area of Science:

• Pharmacokinetics and Pharmacometrics
• Preclinical Drug Development
• Statistical Modeling in Biology

Background:

• Preclinical studies often use an arbitrary 3-6 samples per time point without justification.
• Lack of a theoretical basis for sample size determination can impact data reliability.
• Accurate sample size is essential for robust pharmacokinetic parameter estimation.

Purpose of the Study:

• To propose a practical, statistically-grounded method for determining the minimum sample number (Nmin).
• To establish a reliable approach for sample size calculation in preclinical pharmacokinetic studies.
• To address the need for theoretical justification in sample size selection.

Main Methods:

• Utilized Monte Carlo simulation and bootstrap resampling techniques.
• Developed a computer program MOMENT(BS) for pharmacokinetic parameter estimation (AUC, MRT).
• Created a new simulation program, MONTE1, to generate data with inter- and intra-individual variations.
• Proposed an index, S(2)CV (sum of squared coefficient of variation), to determine Nmin.

Main Results:

• The proposed approach was successfully applied to actual preclinical experimental data.
• Demonstrated the practical utility of the developed method for sample size determination.
• Highlighted the limitations of one-point sampling in assessing inter- and intra-individual variability separately.

Conclusions:

• The developed Monte Carlo simulation and bootstrap resampling approach provides a robust method for determining Nmin in preclinical studies.
• The S(2)CV index offers a quantifiable measure for sample size justification.
• Further consideration is needed for studies where inter- and intra-individual variability cannot be distinguished from one-point sampling data.