Zamacona MK, Gomeni R
GlaxoSmithKline, Clinical Pharmacokinetics Modelling & Simulation, Verona (Italy)
Background. Although crossover studies are preferred to establish In Vitro – In Vivo Correlations (IVIVC), parallel studies are also acceptable if a common reference treatment is used 1. The outcome of an IVIVC may be influenced by the experimental design and by the extent of inter- individual variability on the drug absorption/disposition kinetic properties of the compound.
Objective. To investigate the influence of the study design and inter-individual variability in the outcomes of a Level A IVIVC.
Methods. Plasma concentration time profiles following oral administration of three formulations (slow, medium and fast release) and after intravenous administration were simulated assuming a crossover design. A one- compartment model with first order absorption and elimination was used. Plasma concentration time profiles were simulated using NONMEM with different degrees of interindividual variability and measurement error. Two sets of simulations were produced: with bioavailability fixed to one and no inter-individual variability; and with bioavailability of 0.8 and an inter-individual variability low and high. The percentage of in vivo drug absorbed vs. time was calculated using numerical deconvolution from the simulated oral plasma concentration time profiles by using the intravenous data as unitary impulse response function. Individual profiles were used to compute the corresponding individual absorption profiles. A linear regression analysis was done to evaluate the degree of correlation between in vitro dissolution and in vivo release profiles. Internal predictability analysis was done to evaluate the accuracy of the In-Vivo predictions given the In-Vitro dissolution data. Mean absorption profiles for each formulation were used in the convolution step to predict the plasma concentration time profiles. Cmax and AUC values were estimated for the simulated and predicted profiles and the prediction error was calculated. In a subsequent analysis, the mean profiles in the deconvolution step with a common unitary impulse response function were used in the analysis to mimic the scenario of a parallel rather than a crossover design.
Results and conclusions. Crossover designs resulted always in good predictability but prediction error slightly increased when the increase of inter-individual variability. Parallel designs gave also in general good predictability values, but with prediction error higher than crossover designs.
References.
[1] Guidance for Industry: Extended release oral dosage forms: Development, evaluation, and application of in vitro/in vivo correlations. US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research. September 1997.
Reference: PAGE 12 () Abstr 448 [www.page-meeting.org/?abstract=448]
Poster: poster