Paul Vrenken (1,2), Maria Vertzoni (2), André Dallmann (3)
(1) Bayer AG, Germany, (2) National and Kapodistrian University of Athens, Greece, (3) Bayer HealthCare SAS, France
Replacing animal testing by in silico and in vitro methods is crucial for ethical and sustainable drug development. Physiology Based Biopharmaceutics Modeling (PBBM) offers an alternative approach, integrating absorption modeling, biomimetic dissolution, and PBPK methods to predict oral drug product performance. To date, no open-source PBBM workflow for immediate release formulations of poorly soluble highly permeable active pharmaceutical ingredients (APIs) is available.
Objectives
- Develop an Open Systems Pharmacology (OSP suite V11.2) based framework that can be applied in formulation bridging, dose selection and bioequivalence assessment [1], consisting of the R-shiny based OSP solubility toolbox [2], an extended MoBi® in vitro dissolution model and a custom PK-Sim® version with updated dissolution models and luminal parameters
- Apply the developed framework using vericiguat, a poorly soluble and highly permeable drug, as model API [3] by:
- o integrating information extracted from in vitro solubility and dissolution data into a PBPK model for formulation specific bioavailability prediction
- o assessing absorption related Drug-Drug Interaction (DDI) when vericiguat is co-administered with omeprazole, a proton-pump inhibitor (PPI)
Methods The PBBM framework has a modular set-up, sub-models are activated based on the API properties. For vericiguat, a lipophilic weak base (Log P: 3, pKa: 4.68), the solubility gain per charge (SG) and bile salt partitioning coefficients were optimized with the OSP solubility toolbox using measured solubilities in aqueous buffers, FaSSIF and FeSSIF. These parameters modulate the maximal attainable aqueous solubility and magnitude of micelle partitioning.
The MoBi® extended dissolution equation explicitly models two distinct flows of unbound and micelle bound API to the bulk fluid from the dissolving particle surface. The hydrodynamic conditions, that affect diffusion layer thickness (h) during USP2 dissolution experiments, were accounted for by a (semi)mechanistic model considering vessel fluid volume, paddle speed and fluid viscosity [4]. A formulation specific Product-Particle Size Distribution (P-PSD), for the formulations used in the clinical studies, was fitted to one dissolution experiment [4]. Selecting for most sampling points on the
dissolution slope, to maximize the information gathered. Fitted P-PSDs were then confirmed or rejected by its ability to predict dissolution where the experimental conditions differ.
All solubility and formulation specific parameters were then transferred to PK-Sim® for PBPK modelling. A published PBPK model for vericiguat was used as the base model and intestinal permeability was refitted based on updated solubility parameters [5]. Population simulations were conducted (n=1000) to resemble the clinical trials in the fasted state (Dose range: 1.25-15 mg) and investigate the ability of the model to capture the bioavailability and variability. Additionally, the ability to predict the effect of PPI, omeprazole (40 mg once daily for 5 days), on vericiguat PK was investigated [6].
Results The aqueous solubility-pH profile and FeSSIF solubility were captured well, but FaSSIF solubility was underpredicted. The measured FaSSIF solubility was reached in none of the experiments and replaced with the highest observed concentration, corresponding well with the predicted FaSSIF solubility. The P-PSD approach was successful for all vericiguat dose levels (1.25, 5 and 10 mg), with all but one dissolution profiles being accurately predicted (absolute average fold error: 0.80-1.25), which had an AAFE of 1.30 and deemed adequate.
The geometric means of PK parameters Cmax and AUC0-tlast were captured by the final PBPK model for the eight clinical trials in fasted state. Cmax was slightly overpredicted with an overall mean and absolute prediction error (MPE and MAPE) of 9% and 17%, respectively. For the AUC0-tlast the MPE and MAPE were -4% and 12%. The observed Cmax and AUC0-tlast ratios of PPI versus control treatment were 0.51 and 0.66 and accurately predicted by the model with 0.54 and 0.69, respectively. The tmax in contrast was underpredicted in all simulations.
Conclusion An open-source PBBM workflow was developed and successfully applied for vericiguat when administered in fasted state conditions with and without coadministration of omeprazole. The wider application of the workflow with more APIs and formulation types and following a meal remains to be studied.
References:
[1] https://www.open-systems-pharmacology.org/
[2] https://cran.r-project.org/web/packages/shiny/index.html
[3] Becker C et al. AAPS Open (2022) 8:16
[4] Pepin X et al. J Pharm Sci (2022) 111, 185-196
[5] Frechen S et al. CPT:PSP (2024) 13, 79-92
[6] Boettcher M et al. Clin PK (2020) 59, 1407-1418
Acknowledgement: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 955756
Reference: PAGE 32 (2024) Abstr 11185 [www.page-meeting.org/?abstract=11185]
Poster: Drug/Disease Modelling - Absorption & PBPK