To the Editor:— We read with interest the article by Minto et al. , 1which discusses a method for combining pharmacokinetic and pharmacodynamic parameter sets from different studies, in which the effect site equilibration constant keois estimated based on the time of peak effect (tpeak) after bolus injection. We have successfully used this approach for approximately a decade in pharmacokinetic–pharmacodynamic models for educational simulations and described it in detail in 1998. 2We are grateful for the recommendation of Minto et al. for other investigators to acquire tpeak, because the availability of this parameter was cited as a limiting factor in establishing a complete parameter set in our article. Based on our study and experience, we would like to comment on the study presented by Minto et al.
Our first comment concerns caution regarding their statement:“As shown in the third remifentanil example, a directly observed tpeakis a model-independent gold standard for describing with a single parameter the time course of concentration in the effect site.” We can relay a comment made by one of the reviewers of our article, which we elaborated on in the discussion section:“The time of peak effect is used to derive the equilibration time constant for the effector compartment. This is a one data point estimate for application in educational simulations. It does not allow for an estimation of the variability of the keoobtained in this manner. Therefore this method is not presented asdf an alternative for the estimation of keobased on multiple plasma concentration and effect measurements in a simultaneous study.”2
We also expect that an estimate for keobased on the pharmacokinetic parameter set for the considered experimental group and on the tpeakfrom a different experiment gives, in general, better results than simply combining the pharmacokinetic parameters with the keoof the second experiment, which does not contain any information regarding the first group. Therefore, we do not contest the main conclusion that the tpeakmethod would yield better predictions of the time course of drug effect than the so-called naive method. However, we have some comments regarding the two simultaneous thiopental studies used to make this point by Minto et al. The unit disposition functions show that they have different pharmacokinetic parameter sets. The different keos combined with the different pharmacokinetic parameter sets result in similar tpeaks, as can be seen in figures 1 and 2 of the article by Minto et al. Given the considerable differences in subject population, this similarity in tpeakcould be serendipitous. Combining a pharmacokinetic parameter set with a tpeakfrom another experiment, which is almost identical to the tpeakof the first set, can be expected to lead to a better estimate for keo. Our article presents a detailed sensitivity analysis to evaluate how inconsistencies in parameter sets influence the derived parameters and model response. 2In our opinion, more data and similar sensitivity studies applied to the specific situation proposed by Minto et al. are required to sustain the above conclusion.
University of Florida College of Medicine, Gainesville, Florida. email@example.com