Modelling and simulation approaches for waiving in vivo pharmacokinetic formulation studies

The bioavailability and bioequivalency of oral drug depends on gastrointestinal tract physiology and drugrelated physicochemical and pharmacokinetic factors. In general, bioavailability of a new drug substance or new formulation is studied in vivo with healthy volunteers. In vivo bioequivalency studies are needed for generic drug products or if a formulation is significantly altered during clinical trials. In certain cases, in vitro dissolution studies can be used as a surrogate for in vivo bioavailability or bioequivalency studies, referred to as a “biowaiver”. These biowaivers are based either on in vitro-in vivo correlation (IVIVC) or the biopharmaceutical classification system (BCS). For drugs with dissolution rate-controlled absorption and level A IVIVC, a direct relationship between in vivo drug input and in vitro dissolution may be found. A BCS biowaiver can be utilised for BCS I drugs that have complete absorption due to high solubility and high permeability. In this thesis, in vitro dissolution methods and computer simulation models were developed to predict relative bioavailability and bioequivalency and to probe properties of drugs suitable for biowaivers. A level A IVIVC model with a stochastic approach was developed for a modified-release formulation series of levosimendan. Firstly, the criteria for selection of a dissolution method for the level A IVIVC model were evaluated. Secondly, a stochastic Bayesian approach was integrated with the level A IVIVC model in order to get a predictions of concentration-time profiles of different formulations. BCS biowaiver studies included literature data evaluation of immediate release formulations of ranitidine, a BCS III drug with high solubility and low permeability. Ranitidine was evaluated as a potential BCS biowaiver candidate. Generalised rules to estimate the risk of bioinequivalency and to suggest new potential biowaivers were evaluated by theoretical pharmacokinetic simulations. Gastrointestinal tract physiology, formulation type and drug solubility, dissolution, absorption and elimination rates were taken into account in the pharmacokinetic simulation model. A dissolution method using pH 5.8 and a rotation speed of 100rpm/min provided acceptable discrimination between formulations based on the level B IVIVC and comparisons of pharmacokinetic parameters (MRT and Tmax) to the dissolution profiles for levosimendan. The level A IVIVC model with Bayesian approach has good external predictability for the formulation series, although an averaged IVIVC model with the same data failed. Subject-specific in vivo data was utilised and predictions were obtained as probability distributions. The BCS III drug ranitidine was suggested as a biowaiver candidate based on data from the literature. Generally, the simulations suggest that BCS III drugs are better biowaiver candidates than some BCS I drugs because they have a lower risk of bioinequivalence and they are less sensitive to differences in gastric emptying rates and formulation types. BCS I drugs are currently accepted for biowaivers, although a short half-life of elimination and rapid rate of absorption cause a high risk of bioinequivalency. Pharmacokinetic models were constructed and tested to predict in vitro-in vivo correlations, relative bioavailability, risk of bioinequivalency and potential for biowaivers. These models are useful new tools in formulation development and regulatory applications.