(T1330-01-02) Application of Bottom-Up Absorption Model for Biopharmaceutics Risk Assessment of Preclinical Candidate

Purpose: Physiologically based biopharmaceutics models are well-established in supporting late-stage drug product development.¹ However, their application in early development can be powerful for mechanistic assessment of the absorption potential of preclinical candidates and formulations.^2,3 Moreover, mechanistic absorption models integrate data collected during preclinical development and puts it into a context, allowing robust projections of clinical outcomes.³ The purpose of this work was to demonstrate the application of a pre-FIH (First-in-Human) absorption model built from in vitro measurements of physicochemical properties and validated against preclinical PK data in the rat, dog, and monkey to assess biopharmaceutics risks to exposure. The human absorption model was used to probe for effects with respect to different prandial states, disease-specific gastrointestinal effects, acid reducing agent, dose volume and formulation effects such as particle size, release profile, precipitation kinetics to understand potential risks to absorption or manufacturing.

Methods: A poorly soluble weakly basic investigational drug was considered for this study, referred to as compound A. A mechanistic absorption model was built using GastroPlus (version 9.8, Lancaster, Ca). Models for all species used default physiologies with theoretical absorption scale factor (ASF) model. Aqueous solubility at different pH and in biorelevant media was determined by shake-flask method. pKa and logP were predicted in silico. Precipitation time was considered as 900 s, GastroPlus default. Permeability was determined by MDCK permeability assay and converted to human effective permeability. Disposition and elimination parameters for preclinical models were calculated from IV arm of the preclinical data. Disposition and elimination parameters for human model were calculated by in vitro-in vivo extrapolation. All the parameters used in the model are listed in Figure 1.

Results: The bottom-up absorption model developed for compound A was satisfactorily validated against rat, dog and monkey PK data as shown in Figure 2. The model slightly over predicted rat data, which was attributed to saturation in the disposition at that dose. The human absorption model was built based on in vitro inputs validated across preclinical species and predicted human disposition parameters. The model was first used to predict dose-exposure linearity, shown in Figure 3a, wherein the exposure showed sub-proportionality above 300 mg dose. This result informed the dose-range suitable for Phase 1. Figure 3b shows model-based biopharmaceutics risk assessment for compound A. AUC ratio lower than 1 refers to negative effect whereas ratio higher than 1 refer to positive effect on the exposure for different scenarios. The sensitivity of the exposure to gastrointestinal physiological effects indicates potential risks to absorption due to increase in gastric pH (due to acid reducing agents like proton-pump inhibitors (PPI)), variability in dose volumes and absence of colonic absorption. Interestingly, delayed gastric emptying showed improved exposure, suggesting absorption to be dissolution-limited and higher dissolution in the stomach to likely increase absorption. Probable risk of decrease in the exposure was identified due to larger particle size of the API, highlighting need for particle size control for clinical formulation. A slight negative food effect was identified at higher doses, however, this was within the range of uncertainty in the prediction.

Conclusion: A thorough biopharmaceutics risk assessment approach using mechanistic absorption models early in development is an effective way to increase speed and efficiency of drug development. It also helps in delineating the range of complex absorption phenomena that can be challenging to study. Herein, the pre-FIH absorption model helped demonstrate development risks for compound A and facilitated discussions with the clinical team as clinical studies were being planned. The effect of PPI co-administration was identified as a significant risk to bioavailability, especially if the clinical doses in Ph 1 were to exceed 100 mg. Food effect was low risk, hence, it was recommended to delay food effect study until the pivotal formulation is developed. The team was also informed about uncertainty in the predictions due to uncertainty in the precipitation kinetics in humans.

References: 1. Anand, Om, et al. "The Use of Physiologically Based Pharmacokinetic Analyses—in Biopharmaceutics Applications-Regulatory and Industry Perspectives." Pharmaceutical Research (2022): 1-20.
2. Bhattachar, S.N., Tan, J.S., Stamatis, S.D. and Hilden, J. “Translational Pharmaceutics and Early Phase Clinical Development”. Burger's Medicinal Chemistry and Drug Discovery, D.J. Abraham (Ed.). (2021) 1-34.
3. Parrott, Neil, and Thierry Lave. "Applications of physiologically based absorption models in drug discovery and development." Molecular Pharmaceutics 5.5 (2008): 760-775.

Acknowledgments: Authors would like to thank Lilly Synthetic Molecule Design and Development Management for supporting this work.

Figure 1: Model parameter inputs for compound A



Figure 2: Observed vs. model predicted PK profile of compound A in (a) rat, (b) dog, (c) monkey



Figure 3: (a) Predicted dose-exposure correlation for compound A. (b) Predicted AUC ratio when comparing exposures for different physiological and formulation effects to healthy physiology for compound A.