Yumi Cleary (1), Valerie Cosson (1), Stephen Fowler (1), Gerard Greig (1), Elena Guerini (1), Navita Mallalieu (2) and Neil Parrott (1)
(1) Roche Innovation Center Basel, Switzerland, (2) Roche Innovation Center, NYC, USA
Objectives: Basimglurant is a small molecule currently developed for major depressive disorder. It is eliminated by hepatic metabolism through CYP enzymes 1A2 and 3A4. Both PBPK and PPK models were developed. The ability of the PBPK model to predict between-subject variability was assessed by comparison to the variability estimated through the PPK model. Subsequently, utility of the PBPK model to assist in sample size calculations for a drug-drug interaction (DDI) study with ketoconazole was investigated.
Methods: A PBPK model was developed based on physicochemical properties, in vitro and clinical PK data using SimCYP version 12.1. A PPK model was developed based on 3762 basimglurant plasma concentrations obtained from 164 individuals using NONMEM (Version 7, level 1, double precision). Simulations for 0.5 mg basimglurant once daily for 14 days were performed with the PBPK and the PPK models for 1000 virtual individuals with exactly the same distributions of demographics. The median and 5th-95th percentiles of simulated basimglurant plasma concentration profiles as well as distributions of simulated Cmax and AUCτ at steady state were compared between the models. Subsequently, a ketoconazole DDI trial with the number of subjects varying between 3 to 60 was simulated with 10 replications. The median and range of the median AUC ratios of basimglurant in the absence and presence of ketoconazole were examined in relation to the number of subjects in each trial.
Results: The median and 5th-95th percentiles of basimglurant plasma concentrations simulated with the PBPK model as well as the distributions of the predicted Cmax and AUCτ showed good agreement with the PPK model (median Cmax is 3.09 vs. 3.31 ng/mL, and median AUCτ is 40.6 ng*h/mL vs. 43.1 ng*h/mL between PBPK vs. PPK model simulations, respectively). The range of the median AUC ratios of basimglurant in trials with 3 to 10 subjects was larger and fluctuated more widely than in the trials with more than 10 subjects. Constant median and range of median AUC ratios were seen in trials with 14 to 60 individuals per trial indicating that a sample size of 14 would be sufficient to investigate the DDI with reasonable confidence.
Conclusions: The SimCYP predicted between-subject variability of basimglurant was in accordance with the estimates made with a PPK model and the utility of the PBPK model to determine an appropriate sample size for a DDI study with ketoconazole was shown.
Reference: PAGE 25 () Abstr 5782 [www.page-meeting.org/?abstract=5782]
Poster: Drug/Disease modeling - Absorption & PBPK