The destructive sampling necessary to obtain tissue-specific pharmacokinetics of radiolabeled ligands does not permit true time-dependent models to be estimated. To adjust for this, small groups of animals are sacrificed at varying time points and the percent injected dose per gram (PID) of radiolabeled ligands is summarized for each tissue of interest. Usually the average PID at each time point is modeled to obtain estimates of pharmacokinetic parameters for each tissue when utilizing summarize#x2013;regress procedures. In this paper, it is recommended that the median should be the summary measure at each time point to be modeled in these circumstances. The discrepancy that exists between modeling the average and median PID is demonstrated by an actual experiment. Finally, simulation studies are performed and the results are compared when using these two summary measures.
展开▼