Pharmacometric evaluation and improvement of models and study designs - applied in diabetes

Abstract: Pharmacometric models are increasingly used to improve the efficiency of the drug development process and increase our understanding of the studied underlying pathophysiological system. These models require assumptions for handling different types of data and the different model components, and the appropriateness of such assumptions must be carefully inspected for unbiased conclusions. The aim of this thesis was to develop new models, that by acknowledging the complexity of the data captures more information, and novel methodologies for model evaluation, as well as applying models to improve study designs, with practical illustrations in the therapeutic area of diabetes. Two new models were developed. An integrated minimal model was developed to enable clinical trial simulations in presence of endogenous insulin secretion while deriving the important physiological indices for clinical diagnosis. A multi-state model was developed for improved handling of survival data in presence of competing risks and interval-censored data. New methodologies for model evaluations were developed that include residual modeling and linearization for assessing possible improvements of the structural and statistical model components as well as using simulations to assess the captured information from the data between structurally different models. A mapping approach for parameters carrying similar information between different models was developed, allowing the derivation of physiological indices from the integrated glucose insulin model. Models were also successfully applied with the purpose of improving study designs, either based on anticipated drug effect or for assessment of physiological indices. In conclusion, new more informative models were developed by acknowledging the complexity of the data, novel methods were proposed and applied for model development/evaluation process, and models were used to improve study designs for clinical trials and clinical diagnosis.