Abstract

The objectives of this study were 1) to develop and verify mechanistic absorption models to predict the pharmacokinetics (PK) of two immediate-release weak base drug (WBD) products (Drug X and Drug Y) with a long time to reach maximum plasma concentration (T_max) (e.g., T_max>6 h.), and 2) to understand what are the key input parameters for these models. For each drug product, the absorption model was initially constructed using measured drug physicochemical parameters in the drug product (e.g., particle size, solubility, permeability) while other drug parameters (e.g., precipitation time, FuEn (unbound fraction in enterocytes)) and physiological parameters (e.g., stomach pH and gastric transit time) were kept as default or as predicted values using the Gastroplus (version 9.0) built-in functions. Literature has shown that lysosome could extensively sequester lipophilic WBDs. Following sensitivity analyses, FuEn was optimized to reflect potential lysosomal trapping of the WBDs. The optimized model was then used to predict the drug PK for different doses, different drug particle sizes in the drug product or administration conditions with regard to food to evaluate the model performance. Predictions were considered acceptable if the predicted maximum concentration (C_max), T_max and area under the concentration-time curve (AUC_inf) were within ± 1.5 folds of the observed data. Absorption models without the lysosomal trapping component optimization could not adequately capture the C_max and T_max following single oral dosing of Drug X and Drug Y, although with a good prediction of AUC_inf (Figure 1). Sensitivity analyses revealed that varying stomach transit time, small intestinal transit time, precipitation time or permeability did not significantly impact the prediction of the absorption phase (e.g., C_max) and T_max. However, the long T_max (8 hours and 6 hours) of Drug X and Drug Y were well captured after considering lysosomal trapping in the model, by optimizing the FuEn to 1% and 4% from 100% (default software parameter) for Drug X and Drug Y, respectively (Figure 1). Furthermore, the optimized model achieved a good prediction of the drug PK for different doses, particle sizes of drug product or administration conditions with regard to food with the predicted C_max, T_max and AUC_infwithin ± 1.5 folds of the observed data. Our data suggested that incorporating lysosome trapping via optimizing parameter FuEn into the mechanistic absorption model was critical for a better PK profile prediction for two immediate-release WBDs that exhibit a long T_max.