Notes
Physiologically based pharmacokinetic models are mathematical models that simulate the processes of release, absorption, distribution, metabolism and excretion after drug administration based on the physicochemical and biopharmaceutical properties of the drug substance, the properties of the formulation and the physiological characteristics of humans or animals. They are used to predict pharmacokinetic parameters such as the drug bioavailability, maximum concentration of the active substance in blood plasma, area under the curve of plasma concentration as a function of time, etc. Physiologically based pharmacokinetic models are useful at various stages of drug development, however their application in the generic pharmaceutical industry for regulatory purposes is not yet so widespread. The purpose of the doctoral thesis was to develop physiologically based pharmacokinetic models for various active pharmaceutical ingredients and formulations with as low prediction error as possible. Using the developed models, we aimed to evaluate the effects of changes in various parameters on the drug bioavailability and to predict the bioequivalence of formulations in the development of generic drugs. We wanted to test whether physiologically based pharmacokinetic models could be used to predict the in vivo behavior of formulations in which the active ingredient is present in amorphous form in one formulation and in crystalline form in another formulation. Additionally, we aimed to test the applicability of the models for predicting the bioequivalence of a test and reference drug in fed study. We also wanted to investigate the applicability of physiologically based pharmacokinetic models to evaluate the effects of gastrointestinal properties on drug bioavailability and to determine parameters that significantly affect the variability of drug pharmacokinetics in in vivo studies. We first developed a physiologically based pharmacokinetic model for a drug, in which the active ingredient was present in amorphous form in the test formulation and in crystalline form in the reference formulation. During the development of the model, we determined how to introduce the difference between the formulations into the model and which model had the lowest prediction error. The developed model predicted the pharmacokinetic parameters of the test and reference formulations with less than 10% prediction error. Using virtual clinical trials, we also predicted the bioequivalence of the test and reference drug in the fasted state, which is consistent with the results of the in vivo bioequivalence study. Key question during the development of models was how to develop a model that has the lowest prediction error. We have found that models are best developed in a stepwise fashion and validated accordingly in the interim. Thus, pharmacokinetic parameters (clearance, volume of distribution, distribution constants) are best determined from the plasma concentration profile after intravenous drug administration. In this case, drug release and absorption do not affect the shape of the plasma concentration curve. Plasma concentration profiles after oral administration of the solution are most useful for determining the drug permeability in the gastrointestinal tract, since in this case the dissolution of the active ingredient does not affect the rate and extent of absorption. If these profiles are not available, plasma concentration profiles following oral administration of an immediate-release capsule or tablet can be used. There are several options for entering dissolution profiles into the model - direct entry of the dissolution profile, fitting the z-factor to the dissolution profile or adjusting the particle size of the active ingredient. When developing an individual model, it is necessary to test different methods and use the one resulting in the model with the lowest prediction error. One of the models, which was developed in a stepwise fashion, was developed based on plasma concentration profiles after intravenous administration of the active ingredient, after oral administration of different doses of the same active ingredient in immediate-release formulations and after oral administration of the same active ingredient in modified-release formulations. The developed model was used to predict the effect of changes in the in vitro dissolution rate on the drug bioavailability. In vitro in vivo correlation could not have been developed according to regulatory guidelines, therefore we developed an in vitro in vivo relationship based on physiologically based pharmacokinetic model. A validated physiologically based pharmacokinetic model was proposed as an adequate alternative to in vitro in vivo correlation when evaluating the effect of changes in the in vitro dissolution profile on drug bioavailability. Next, we wanted to use physiologically based pharmacokinetic models to predict the bioequivalence of drugs in the fed state. Physiologically based pharmacokinetic models can be used to predict the effect of food on drug bioavailability by altering relevant parameters (e.g. pH and volume of gastrointestinal fluid, gastrointestinal transit time). The previously developed model for the amorphous and crystalline forms of the same drug was also used to predict the effect of food on the in vivo performance of the formulations and to predict the bioequivalence of the test and reference formulations in fed state. The modeling results agreed with the in vivo results of the bioequivalence study. We developed six additional models for drugs belonging to classes I, II and III according to the biopharmaceutical classification system. The models were developed based on literature and in vitro data on active ingredients and formulations as well as on the results of an in vivo bioequivalence studies with drug administration in fasted state. The models were properly validated and then used to predict bioequivalence in fed state. We predicted that the test and reference formulations were bioequivalent in five cases and non-bioequivalent in one case, which was consistent with the results of in vivo bioequivalence studies. In the final part of the research, we studied the variability of the gastrointestinal tract conditions and how the literature values match the default values of physiological parameters in the gastrointestinal tract model in GastroPlus%. Based on the results of the literature review, we adjusted the physiological parameters of the model in the fasted and fed states. The modified gastrointestinal tract models were used to determine the effect of variable gastrointestinal characteristics on the bioavailability of two drugs. Physiologically based pharmacokinetic models were developed for a delayed-release tablet and for an immediate-release tablet. We found that the pharmacokinetics of the delayed-release tablet were most influenced by the timing of gastric emptying. However, for the immediate-release tablet containing poorly soluble active pharmaceutical ingredient, the variability of dissolution and absorption was most significantly influenced by pH and fluid volume in the gastrointestinal tract. Using parameters sensitivity analysis and virtual clinical trials, we determined which parameters, in addition to gastrointestinal conditions, influence the in vivo variability in the pharmacokinetics of these two drugs. In the case of a delayed-release tablet, an important parameter affecting pharmacokinetic variability is clearance, which is particularly noticeable for the part of the population that poorly metabolises the drug substance. In the case of an immediate-release tablet, variability in the permeability of the active substance in the gastrointestinal tract contributes to the variability in pharmacokinetics in addition to the pH and fluid volume in the gastrointestinal tract. Using the developed models, we were able to describe the plasma concentration profiles of all subjects that deviated significantly from the mean plasma concentration curve. Through the presented research work, we have demonstrated the applicability of physiologically based pharmacokinetic models in the development of generic drugs. The use of physiologically based pharmacokinetic models and virtual clinical trials could lead to a reduction in the number of clinical trials, such as fed bioequivalence studies, as we have shown that the outcomes can be predicted on the basis of a properly developed and validated model. Thus, reducing the number of clinical trials could lead to lower exposure of healthy volunteers to medication, reduce time and cost of drug development, and help bring high-quality and effective medicines on the market faster.