In-Silico Prediction of Drug Properties in Man Using Preclinical Data and Computer-Assisted Drug Development

Abstract

Computer-assisted drug development (CADD) is a knowledge-based process that integrates newly collected information in a drug/disease-specific knowledge frame and provides a rational basis for decision making during drug development. A mathematical model was developed to predict the dosage regimen appropriate to reach a desired human response for a new central nervous system compound from data collected in preclinical studies. The model incorporates uncertainty on the human pharmacokinetic/pharmacodynamic (PK/PD) relationship scaled from animal data, including the affinity adjustment derived from in-vitro receptor binding studies. Brain receptor occupancy, estimated using positron emission tomography imaging, was used as a surrogate marker of clinical efficacy. The CADD approach was used to design the first-time-in-man and the proof-of-concept study. The expected minimal effective dose in man was estimated by simulating a dose-escalating trial, as the dose necessary to reach and maintain during chronic treatment is a target receptor occupancy of less than 70%, assuming that the therapeutic effect is associated with a receptor occupancy of less than 70% at a given probability level (Prob lt; 70%). Finally, the optimal in-vivo release characteristics from a pharmaceutical formulation were evaluated using a convolution-based model linking the in-vivo delivery rate to the PK/PD response, minimizing an appropriate cost function. Clinical trial simulation was used as a supportive tool for decision making by evaluating different scenarios accounting for the uncertainty in the predicted PK/PD relationship, intersubject variability, and drug potency.