Per Larsson, Eva Weibull, Astra Draco AB, Lund, Sweden, Søren Pedersen, Lone Agertoft,
Kolding Hospital, Kolding, Denmark
The semi-simultaneous design is an alternative to the traditional crossover design in bioavailability studies. The test and reference treatments are given on the same study day with a short time interval. Since the plasma concentrations then reflect both administrations, a pharmacokinetic model is required to evaluate the study. For a study with semi-simultaneous design an evaluation based on non-linear mixed effect models is well suited.
The semi-simultaneous design was applied to a bioavailability study of inhaled drug in children 4-6 years. The intravenous reference dose was given three hours before the inhaled dose. The advantages of this design was a reduced number of vein punctures, fewer blood samples and only one study day compared to a traditional crossover study. In young children, these advantages were considered to be of major importance to increase compliance.
When the study was designed, knowledge of the pharmacokinetics in adults was used to predict the outcome and to assure that the doses were sufficiently high to detect the drug in plasma. Individual plasma concentration curves for children were simulated using the estimates of variances obtained from a mixed effect model analysis of adult data. The evaluability of the study was investigated by fitting the original model to simulated data. The simulation work gave increased confidence in the design and was important for internal and external acceptance of the study.
The results from the study were evaluated using a mixed effect model based approach. Two different models described the data equally well: a three-exponential i.v. model with first order uptake of inhaled drug and a two-compartment i.v. model with first order uptake of inhaled drug plus an additional first order uptake of swallowed drug. The semi-simultaneous design used in this study could not differentiate between these two models. The possibility of this ambiguity was not predicted from the adult data underlying the simulations. However, given adult data from other studies, the three-exponential model was the most likely alternative. Therefore, the results of this study could not be unambiguously interpreted without using prior knowledge of the pharmacokinetics of the drug.
Reference: PAGE 6 () Abstr 660 [www.page-meeting.org/?abstract=660]
Poster: poster