Abstract
INTRODUCTION AND OBJECTIVE: Physiologically based pharmacokinetic (PBPK) models are increasingly used to predict food effect (FE) but model parameterization is challenged by in vitro-in vivo (IVIV) disconnect and/or parameter nonidentifiability. To overcome these issues, we propose a simplified PBPK model, in which all solubility-driven processes are lumped into a single parameter, solubility, which is optimized against observed concentration-time data. METHODS: A set of commercially available biopharmaceutical classification system (BCS) II/IV compounds was selected to measure the solubility in a fasted state simulated intestinal fluid (FaSSIF) medium. The compounds were ranked from the lowest to the highest dose-adjusted FaSSIF solubility (FaSSIF/D) value and subdivided into three areas based on an upper and a lower limit: drugs with FaSSIF/D > upper limit having no FE, drugs with FaSSIF/D < lower limit having FE, and drugs between the limits said to be in the sensitivity range (SR), for which we tested the hypothesis that solubility-limited absorption (SLA) identified by simplified PBPK model can reliably predict positive FE if their exposures are not impacted by gut efflux or gut metabolism. RESULTS: We demonstrate, using a subset of drugs within SR for which PBPK models were available, that drugs with SLA exhibited a positive FE, while those with no SLA did not show FE. CONCLUSIONS: This proposal allows for a reliable binary prediction of FE to enable timely decisions on the need for pilot FE studies as well as the timing of pivotal FE studies.