(T0930-10-54) Evaluation of PBPK Modeling in Predicting the Distribution and PK of Targeted Protein Degraders in Preclinical Species

Purpose: Targeted protein degraders (TPDs) are a novel class of therapeutics to achieve degradation of the protein of interest through inducing proximity with inherent proteolysis machinery. The special design of most TPDs results in overall unfavorable physicochemical properties according to the Lipinski’s rule-of-5 for druggability. As a result, accurate prediction of in vivo pharmacokinetic parameters for TPDs remains challenging. Physiologically based pharmacokinetics (PBPK) modeling harnesses the physiological features of animals and humans and has been widely utilized in PK prediction. However, its applications for TPDs are still publicly scarce. The purpose of this study is hence to explore the usefulness of PBPK modeling in TPDs PK prediction in animal species, with in-depth analysis of the impact from key physiochemical and ADME parameters. Specifically, the accurate prediction of volume of distribution at the steady state (Vss) is prioritized, because Vss is not only required to fit the frequently observed distribution phase in PK curves of TPDs, but also critical to predict the half-life and determine the appropriate dosing regimen.

Methods: We selected 3 TPDs prepared through BeiGene’s preoperatory Chimeric Degradation Activation Compounds (CDAC^TM) platform for exploration for validation using GastroPlus® (V9.7). The selected TPDs possess diverse physiochemical and ADME properties, including lipophilicity, pKa and unbound fraction in the plasma (Table 1). As Vdss is highly dependent on tissue distribution, we compared the observed Vdss with predicted values based on 5 built-in tissue partitioning methods, as well as selection of perfusion limited distribution versus permeability limited distribution model (the latter is recommended by GSK in modeling a RIPK2 TPD¹). As lipophilicity is a major determining factor for tissue distribution, we also compared the prediction results using experimentally measured LogD (pH 7.4) and software predicted LogP based on the chemical structure of TPDs. With the identification of most appropriate tissue partitioning methods and successful prediction of Vdss of all 3 TPDs, we next performed sensitivity analysis on critical physicochemical and ADME parameters to highlight the necessity for accurate measurements.

Results: Among all 5 methods to calculate tissue/plasma partition coefficients, the Lukacova (with lysosome) method provided reasonable prediction (1-5.7-fold difference compared with observed Vdss for 3 TPDs in all species), and consistently outperformed other 4 methods. The Poulin&Theil-Extracellular method significantly underestimated Vdss in all cases. In terms of the distribution model, permeability limited distribution failed to capture the elimination phase of the PK curve. In addition, in our cases Vss could be predicted with higher accuracy using the in-silico predicted LogP values compared with experimentally measured LogD (pH 7.4). Based on results of the sensitivity analyses, LogP and plasma protein binding were the most sensitive parameters affecting Vss prediction (Table 2), in alignment with opinion from the industry². In our cases, the blood/plasma ratio did not significantly influence Vss prediction (mostly within 3-fold difference at tested levels).

Conclusion: Based on our knowledge, this work is the first to comprehensively evaluate PBPK options and parameter sensitivities in the Vss prediction for TPDs. We have identified that Lukacova (with lysosome) method is most appropriate to describe plasma/tissue partitioning for TPDs, and the perfusion limited tissue distribution model works better than permeability limited model. In terms of physicochemical and ADME parameters, in-silico predicted LogP can be used to predict Vss with reasonable accuracy, and accurate measurement of unbound fraction in plasma is required, especially for TPDs that are highly bound to plasma protein. Even though the selected compounds attempted to cover a broad range of physicochemical/ADME properties of TPDs, the observations and conclusion are based on 3 molecules only. Further investigation and validation using a larger panel of compounds with various molecular design is warranted.

References: 1. Harling, John D., Paul Scott-Stevens, and Lu Gaohua. "Developing pharmacokinetic/pharmacodynamic relationships with PROTACs." (2020).
2. Volak, Laurie P., et al. "Industry perspective on the pharmacokinetic and ADME characterization of heterobifunctional protein degraders." Drug metabolism and disposition (2023).

Table 1: physicochemical and ADME parameters of selected TPDs

Table 2: sensitivity analysis of fu%, blood/plasma ratio and LogP on Vss prediction