Derisking the development of rilzabrutinib, a BTK inhibitor currently in phase III, with a PBPK model in lieu of some clinical studies
Sylvaine Galiègue, Laurent Boulu, Wen Lin, Christine Xu, Sreeraj Macha, Jean-Marie Martinez, David Fabre
sanofi
Objectives: The main objective of this study was to develop and qualify a Physiologically-Based Pharmacokinetics (PBPK) model able to predict the PK of rilzabrutinib, a novel investigational drug currently being evaluated in a phase III study for the treatment of autoimmune and inflammatory diseases. The model was then used to support clinical development and different applications are presented here. PBPK analyses were conducted to predict metabolic interactions via CYP and transporter-mediated interactions. Another example included PBPK analysis for assessing the effect of renal impairment (RI) on the drug PK.
Methods: A PBPK model for rilzabrutinib was built in Simcyp (Certara®) version 20 using a middle-out approach including in vitro data (eg physicochemical properties, protein binding, pH solubility profile, permeability data) and information from clinical studies. A bottom-up approach was driven to build the ADAM model for the absorption part. To this aim, SIVA, the stand-alone tool from Simcyp, was used to analyze in vitro data to provide the solubility and permeability parameters necessary for the simulations. A top-down approach was used to build the disposition part of the model based on clinically observed data. The model was optimized on CYP3A4 Ki and fm based on clinically observed DDI interaction and then verified using the PK results from both arms of the clinical interaction studies. The analyses for RI were performed using Simcyp version 21, newly including the mild renally impaired population.
Results: Optimization of CYP3A4 Ki was given special attention through sensitivity analysis. The final model described the observed rilzabrutinib pharmacokinetics well enabling further applications.
Regarding DDI, simulations were performed using default library compound files within the simulator for a panel of CYP3A4 inhibitors or inducers without any modification. Weak CYP3A4 inhibitor was found to have a limited impact by increasing rilzabrutinib exposure by 1.5-fold and moderate inhibitor increased rilzabrutinib exposure by ~3-fold. Weak CYP3A4 inducer decreased rilzabrutinib exposure by ~30% and moderate, and severe CYP3A4 inducers decreased rilzabrutinib exposure by >60%. Based on the simulations performed with probes and/or substrates for transporters (again using corresponding default library compound files), rilzabrutinib was not predicted to have any relevant effect on P-gp, BCRP, OATP1B1 and OATP1B3 substrates. As per Health authoritie’s guidances, an analysis of the uncertainties on Ki for these transporters up to a 30-fold reduction of in vitro Ki was performed, and the weak impact was still evidenced.
When simulating rilzabrutinib exposure in renally impaired populations, the impact of renal impairment on the exposure was found to be modest. An increase by 1.1- to 1.4-fold in exposure associated with a decrease by 0.8- to 0.7-fold in CL/F were predicted compared to healthy subjects. Mild RI evidenced slightly less impact than moderate and severe RI for which the results were quite similar. Analyzing the simulation outputs provided insights on the underlying mechanisms for renal impairment to impact the drug exposure. Precisely, the decrease in total clearance predicted with RI was likely mainly related to a decrease in CYP3A4 contribution to total clearance caused by a decrease in CYP3A4 enzyme abundance and/or activity in the liver.
Conclusions: The poster presents the development of the PBPK model and its optimization process. The importance of sensitivity analysis on KI was emphasized. In addition, some examples of its application are provided to illustrate the benefit of using PBPK modeling in drug development.
References:
[1] Text for reference 1.
[2] Text for reference 2, etc etc