by Arne Ring
Date: 19.12.2001
Language: written in GER
Abstract: The estimation of pharmacokinetic parameters of absorption and disposition processes is one of the main focus topics of pharmacokinetic analysis. The validity of the estimation will be determined by the interaction of adequate modeling and appropriate statistical methodology.
This thesis investigates the effects of potential systematic errors (which are caused by model-misspecification) on estimation methods. The investigations are performed for certain absorption and disposition processes which cannot be analysed correctly on the basis of previous methodology.
To describe the absorption of modified drug release, the Inverse Gaussian (IG) distribution was validated as appropriate model for certain cases. This function consists of two (pharmacokinetically interpretable) parameters which characterise the rate and form of the absorption processes.
Effects of the usage of the IG in comparison to the traditionally applied exponential distribution were investigated e.g. in the field of bioequivalence analysis. One outcome was that only the so-called "intercept metrics" are able to reflect potential changes of the absorption process.
The characterisation of the disposition system was based on a whole body model, which describes the underlying distribution and elimination processes on organ level. Theoretical consequences of the drug disposition were examined by means of three anaesthetic drugs.
Subsequently, investigations of regression methods of the resulting concentration-time-curves were performed to find an appropriate statistical methodology for pharmacokinetic parameter estimation in cases of potential model miss-specification. Two methods were found statistically efficient: one being the weighted least squares, the other based on the entropy divergence principle.