(T1530-09-52) Leveraging an Algorithm-Based Digital Platform for Excipient selection facilitating Formulation Development of Poorly Soluble Active Pharmaceutical Ingredients

Purpose: It is estimated that approximately 90% of drug candidates are poorly soluble. APIs that do not dissolve, cannot be absorbed and ultimately, cannot treat. A significant amount of time is devoted during development to optimize the formulation approach, its excipients, and manufacturing technology. This research focused on following a Formulation by Design approach to systematically evaluate the best solubilizing excipients and formulation approach considering API physicochemical properties, differences in solubility parameters and predictions from a virtual formulation assistant ZoomLab^®. This assistant analyzes multiple API and excipient molecular descriptors to feed trained algorithms that propose optimized formulation strategies. Ultimately, the intent is to accelerate development by reducing trial and error in the laboratory.

Methods: Rivaroxaban (RXB) (Teva, Israel), a BCS Class II poorly soluble drug was used as a model drug. The following ZoomLab^® (BASF SE, Germany) apps were utilized: “Developability Classification” for assessing the API’s solubility and permeability; “Decision Support for Solubilization” for identifying the most suitable formulation approach (solid dispersions vs. lipid-based) for RXB; “Solubility Parameter Calculation” for calculating the Hansen Solubility Parameters of RXB; and, “Polymer and Solvent Finder” for ranking predicted solubilizing excipients, polymers and solvents based on solubility parameter differences. Solubility studies using shake flask method: Solubilizing polymers predicted by ZoomLab^® were assessed. Polymer solutions at 2% w/w were saturated with 25 mg each of RXB, and stirred for 72 hrs at 37 ± 1 °C. The solutions were filtered and analyzed using UV at 248 nm. Miscibility studies using Melting Point Depression

Method: Binary Mixtures of Polymers exhibiting the highest solubility using shake-flask method (copovidone (Kollidon^® VA 64) and Soluplus^®) with RXB were prepared. Differential Scanning Calorimetry (DSC) was used to measure the depression in melting point (Tm) and glass transition temperature (Tg) of the binary mixtures compared to the individual materials. The predicted glass transition temperature results using Gordon-Taylor equation (Figure 1 (A)) were compared to the experimental Tg to assess API-polymer miscibility and optimal drug loading in the solid dispersions. Flory-Huggins, on other hand, is used to understand thermodynamic stability by considering both enthalpic and entropic contributions to the miscibility of the drug-polymer system. The melting point data of the binary mixtures were fit to the Flory-Huggins model (Figure 1 (B)) to extract the Flory-Huggins Interaction parameter χ: Rearrangement of equation in Figure 1 (B) into the form y=χ *Φ²polymer yields a convenient form to extract the Flory-Huggins interaction parameter χ directly from the slope—assuming it is temperature independent—where y is presented in Figure 1 (C) The interaction parameter is a quantitative measure of the degree of interaction between the drug and polymer molecules, specifically the excess free energy of mixing. A lower value, particularly if negative, indicates the thermodynamic solubility of the drug within a polymer matrix would be higher than a drug-polymer pair with a more positive interaction parameter.

Results: ZoomLab^® predicted a solubility-limited bioavailability for RXB with an input dose of 20 mg. Based on molecular descriptors, logistic models and comparisons with ten approved “similar” APIs, the model suggested solid dispersions over lipids for a formulation approach, with a maximum achievable drug loading of 35% w/w. Based on recommendations from ZoomLab^® and solubility studies (Figure 2), three polymers were identified as candidates for preparing solid dispersions of RXB using HME: Soluplus^®, povidone K-30, and copovidone. HME requires higher processing temperature to prepare solid dispersion of drugs. Hence, due to the lower degradation temperature of povidone K-30, copovidone and Soluplus^® were chosen for further miscibility studies. RXB- Kollidon^® VA 64 followed the Gordon-Taylor equation for glass transition temperature (Tg), as observed experimentally, whereas RXB-Soluplus^® exhibited a negative deviation from the expected Tg (Figure 3). This indicates higher miscibility in the RXB-Kollidon^® VA 64 binary system compared to the RXB-Soluplus^® system. Furthermore, RXB-Kollidon^® VA 64 showed a lower interaction parameter (χ=0.793, R²=0.922) compared to RXB-Soluplus^® (χ=2.222, R²=0.942), suggesting better interaction and stabilization of the solid dispersion with Kollidon^® VA 64 .  Consequently, a formulation strategy incorporating RXB, Kollidon^® VA 64, and Soluplus^® appears feasible. In this strategy, Soluplus^® provides solubilization, while Kollidon^® VA 64 contributes to kinetic stabilization, playing a crucial role in the formulation's overall performance. The phase diagrams also indicate the polymer and drug HME are largely kinetically stabilized rather than thermodynamically (Figure 3).

Conclusion: This study demonstrated the effectiveness of ZoomLab^® as a tool to accelerate development by reducing laboratory trial-and-error experimentation. It enables a systematic formulation by design approach leveraging the API’s physicochemical properties and molecular descriptors to predict the best solubilizing polymers and adequate formulation approach. By integrating computational predictions with initial experimental data, we were able to gain insights into the stabilization of amorphous solid dispersions. This combined approach aids in understanding drug-polymer interactions while streamlining the formulation development process, ultimately leading to more efficient and robust pharmaceutical development.

Figure 1. (A) Gordon-Taylor Equation (B) Flory- Huggins equation, and (C) Rearranging equation (B) to represent as 'y' in equation y=χ *Φ2polymer

Figure 2. Solubility of RXB in 2 %w/v aqueous solutions of various polymers at 37°C ± 1 °C

Thermodynamic phase diagram of binary mixture of RXB + Kollidon® VA 64/Soluplus®