(M0930-10-69) Mechanistic Modeling of In Vitro Biopharmaceutic Experiments for a Weakly Dibasic Drug for Incorporation into PBPK Models

Purpose: In the late stages of formulation development, drug product batch dissolution data is used to support crucial activities like adherence to Quality by Design and setting clinically relevant specifications for the dissolution of the commercial product or the particle size of the drug substance. New translational modeling methodologies need to be developed for the use of quantitative biopharmaceutic principles in medicinal product development and approval, and physiologically based biopharmaceutics models (PBBM) are progressing significantly in this domain. When using PBBM to stimulate product dissolution in vivo, the effects of surface pH and other diffusion-relevant species should be considered. However, most commercial in-silico PBBM software tools don't handle solubility and dissolution in a way that considers some key factors. When solutes diffusing away from a solid surface during dissolution react with medium components, through an acid-base reaction, the surface pH near the solid surface differs from the bulk pH. The dissolution rate of the solid is determined by the surface drug solubility, which is controlled by the surface pH, as well as the diffusion coefficients of all species present. Entrectinib is an example of a weak diprotic base, and thus, the pH at the surface of dissolving solid particle could be much higher than the pH of the bulk dissolution medium. This means that the rate of dissolution in vitro or in vivo would be slower than expected based on the bulk pH. The purpose of this study was to establish a stepwise biopharmaceutical modeling framework for BCS class II, weakly basic drugs, in which the relevant biopharmaceutical parameters are estimated from in vitro data and are subsequently integrated into PBBM to predict the in vivo exposure of drug products.

Methods: Entrectinib solubility in aqueous buffer solution and biorelevant media (simulated gastric fluid, SGF, fasted state simulated intestinal fluid, FaSSIF, and fed state simulated intestinal fluid, FeSSIF) were used as reported [1]. The reported pKa values of entrectinib were 2.54 and 7.54 respectively, however we noticed that the Henderson-Hasselbalch equation did not adequately describe the measured solubilities, thus we corrected the pKa's and solubility factors (SF1 and SF2). The solubility model was then extended to the bile micelle-mediated component using biorelevant media. Bile micelle partition coefficients (K_m:w) were then fitted in the model to accurately predict the solubilities. A dissolution study of 200 mg entrectinib was reported in a USP II apparatus with 0.0001 N HCl, 500 mL, initial pH 4, and a paddle speed of 75 rpm [1]. Mass transport models were used to estimate the surface pH in unbuffered media by expanding on Mooney et al.'s classical methods with a numerical solution that led to a fourth-degree polynomial equation [2]. Bulk pH was estimated using charge-balance equations. These equations are solved using MATLAB R2022b software. The estimated surface pH value was incorporated into the built-in fluid dynamics model in the Simcyp In vitro Analysis toolkit (SIVA, Version 4; Certara, Simcyp, UK) to deconvolute in vitro dissolution data. The estimated parameters from in vitro data were then convoluted and translated in the Simcyp software to develop mechanistic PBBM.

Results: The aqueous solubility model was validated using the data over a wide pH range, from 1 to 10. Figure 1a, depicts the reported versus predicted entrectinib solubility in different aqueous media. The aqueous solubility modeling confirmed the estimated pKa’s were 2.54 and 5.3 and the solubility factors were 5950 and 101.08, respectively. Using these parameters as inputs, the solubility model was then expanded to the bile micelle-mediated component using biorelevant media. Fitted coefficients values were 5.119 and 5.617 for neutral and ionized species, respectively. The model predictability was then also confirmed against SGF, FaSSIF and FeSSIF solubilities (Figure 1a). Simulation results indicated that the bulk pH rises from pH 4 to pH 4.90 (Figure 2b) after dissolution of 200 mg entrectinib in 0.0001 N HCl, pH 4, while the surface pH rises from pH 4 to pH 4.64 (Figure 1b). This increase is attributed to the basic entrectinib's self-buffering effect and is close to the experimental bulk pH measured after dissolution (pH 5.2), which is limiting dissolution. Moreover, if the medium pH is kept constant at pH 4 (assuming no self-buffering effect), the extent of dissolution is clearly over predicted (Figure 2a).

Conclusion: Aqueous solubility modeling assisted in confirming the fundamental physicochemical parameters of the weakly basic drug molecule. The solubility model performance was then also confirmed against biorelevant media. The effect of pH resulted in solubility-limited dissolution of entrectinib. A sequential, stepwise approach is used to estimate the biopharmaceutical parameters that are subsequently integrated into PBBM, i.e., the aqueous solubility, biorelevant solubility, particle surface pH and dissolution processes. Such approach may be helpful to guide the formulation strategy selection for weakly basic compounds, such as the inclusion of acidulants in the formulation order to increase bioavailability.

References: 1. Parrott, N.; Stillhart, C.; Lindenberg, M.; Wagner, B.; Kowalski, K.; Guerini, E.; Djebli, N.; Meneses-Lorente, G. Physiologically Based Absorption Modelling to Explore the Impact of Food and Gastric PH Changes on the Pharmacokinetics of Entrectinib. AAPS J. 2020, 22, 1–13.
2. Mooney, K.G.; Mintun, M.A.; Himmelstein, K.J.; Stella, V.J. Dissolution Kinetics of Carboxylic Acids I: Effect of PH under Unbuffered Conditions. J. Pharm. Sci. 1981, 70, 13–22.

Figure 1: (a) Entrectinib aqueous and biorelevant solubility modeling, (b) Surface pH of entrectinib in different aqueous pH media

Figure 2: Simulated and reported dissolution profiles of entrectinib 200 mg capsule formulation, (a) not considering self-buffering effect (b) considering self-buffering effect using simulated bulk pH