Toxicokinetic analyses with destructive sampling

Even with satellite animals, repeated sampling within a rodent species is often not performed. Usually, each animal contributes to the data at a single time point. When the animal is sacrificed to obtain the pharmacokinetic data, these animals are said to be destructively sampled. The number of animals in such studies is large and costly since often a study has many different dose levels, are dosed in both males and females, and samples are collected at many time points. As a com- 

The primary TK endpoints from toxicity studies include measures of exposure, such as area under the curve and maximal concentration, time to maximal concentration, and half-life. Usually, these metrics are obtained by calculating the mean or median concentration at each time point and then using noncompartmen-tal methods to obtain the pharmacokinetic parameter estimates. The problem with this approach is that it ignores between-subject variability. The variability between animals is lumped into a single animal and the estimate of the parameter is obtained. In doing so, the error associated with the parameters is biased because of the ignored between-subject variability. 

An early example of this methodology was presented by Burtin et al. (1996) who presented the results of a PopPK analysis from a 13 week toxicology study in male and female rats orally dosed once daily at four dose levels. Each animal provided one sample on the first day of dose administration and one sample after the last dose on Day 92 at one of five possible times (0.5-, 1-, 2-, 7-, and 24-h postdose) at the same time on each occasion. Two animals were sampled at each of the five times. They then compared the results of the PopPK analysis to a traditional noncompartmental approach. Both analysis methods came to similar conclusions, but the PopPK approach, resulted in greater mechanistic interpretations to the data. 

---