Increasingly, the drug development pipelines of pharmaceutical companies for new oral delivery compounds contain numerous small molecules that have low aqueous solubility and require bioavailability enhancement to be efficacious. While many bioavailability-enhancing strategies are available to address this need, selecting the best method can be difficult. The goal is to propose the simplest, most cost-effective solution for achieving the target product profile (TPP) and should include a risk assessment that balances the interplay among drug bioperformance in vivo, physical and chemical stability, as well as manufacturability of the drug substance and final drug product. In addition, business drivers, such as the development timeline and projected costs, should be considered. To strike this balance and alleviate these pressure points, drugmakers can leverage amorphous solid dispersions (ASDs) to address poorly bioavailable compounds.

For many formulators, ASDs represent the solution of choice because the technology is widely applicable—spanning the diverse compound-property space from “brick dust” to “greaseball” compounds—and accommodates multiple manufacturing processes (e.g., spray drying and hot melt extrusion).^1,2 A critical aspect of any ASD formulation development program is being able to predict the impact of the amorphous form on drug exposure in vivo relative to the crystalline form. Many simple tools exist for making crude predictions, such as calculating the maximum absorbable dose (MAD) or using the Fraction absorbed Classification System (FaCS) to classify drug absorption as dissolution rate-, solubility-, or permeability-limited. However, these tools provide only a preliminary prediction of in vivo exposure. Absorption modeling can provide much better mechanistic understanding of potential in vivo performance of ASDs using quality inputs generated in vitro.

This article discusses the three key types of in vitro bioperfomance data that absorption models should capture to best predict in vivo performance: (1) drug dissolution; (2) speciation; and (3) precipitation. By carefully generating and incorporating these data into a model, drug developers can mitigate risk, streamline in vivo study design, accelerate the development process and improve formulation decisions, such as lead formulation selection and dosage-form design. 

Differences in ASD dissolution models
ASDs consist of an amorphous drug within a polymer matrix, offering improved dissolution, physical stability and manufacturability.^3,4 As such, the dissolution performance of ASDs typically cannot be described using first-principle models, meaning purely from drug solubility and particle size, because the polymer itself impacts dissolution. In many cases, ASD dissolution performance may depend on pH, complex biorelevant media composition (common for highly lipophilic drugs that have a high affinity for bile salts and lipidic structures in the gastrointestinal (GI) fluid) or drug-to-polymer ratio in the ASD, as it impacts which component—drug, polymer, or both—is primarily responsible for driving dissolution from the ASD.

Accurately capturing ASD dissolution performance requires careful in vitro dissolution test design that considers both the drug and polymer properties with in vivo dose-to-volume ratios. Gastric-to-intestinal transfer dissolution tests are most useful for incorporation into absorption models because they capture dissolution performance in low-pH (stomach) and moderate-pH (intestinal) environments. It is useful to collect in vitro performance data using gastric-transfer methods even for drug compounds that do not have pH-dependent solubility profiles because the ASD matrix polymer may alter dissolution performance or drive physical phenomena, such as particle agglomeration, that may not be captured in other single-medium methodologies.⁵

For ASDs, gastric-transfer in vitro dissolution data is often best captured in an absorption model using a z-factor dissolution model,6 which can describe dissolution performance specific to the ASD of interest (e.g., drug loading and dispersion polymer chemistry), rather than first-principles models based on just drug solubility and particle size with no polymer. The z-factor is a lumped term of constants, diffusivity and true density with one variable—particle radius. A best fit is achieved when the particle radius is optimized to an “effective particle radius.” This radius may not correspond to typical particle size data determined from light scattering techniques, but may be due to physical phenomena, such as particle agglomeration, or for ASDs specifically, binary mixtures of drugs and polymers that have different driving forces for dissolution in a given medium. In the z-factor equation, both drug diffusivity (D) and true density (ρ) remain constant whereas particle radius (r) varies to optimize the z-factor and produce an “effective particle radius” best suited to describe the dissolution data. This model differs from the first-principles models (e.g., Noyes-Whitney), as it allows the incorporation of a unique dissolution rate as a function of pH using a direct fit observed in in vitro dissolution data (see Figure 1).

The impact of ASD speciation on drug absorption
Understanding the driving force for dissolution is critical for capturing ASD bioperformance in an absorption model. When ASDs dissolve, a wide array of drug-containing species can form that may contribute to in vivo absorption. Common species include freely dissolved drug (neutral and ionized forms); drug bound to bile-salt micelles and other lipidic structures; and, in some cases, nano-sized species arising from amorphous-phase separation once the amorphous solubility of the drug has been reached in GI fluid.^3,7,8 Concentration-vs-time profiles for each drug species—unbound, micelle-bound, and nanospecies—are typically quantified in vitro through bile-salt partitioning experiments with centrifugation and subsequent ultraviolet (UV) analysis (i.e., liquid chromatography). For many lipophilic Biopharmaceutical Classification System (BCS) Class II drugs—a common drug class for which ASDs typically show benefit—the rate-limiting steps to absorption are solubility and drug transport across the unstirred water layer (UWL) in the small intestine. Any drug species that are small and abundant enough, such as micelle-bound drug and nanospecies, can contribute to drug absorption by enhancing the dissolution rate and acting as a drug “shuttle” to enhance transport of the drug across the UWL. As a result, these drug species provide a reservoir at the epithelium surface for drug absorption^3,7,9 (see Figure 2).

Capturing the impact of ASD drug speciation on absorption in silico can be accomplished in two different ways. First, dissolution rate enhancement arising from bile-salt micelles or amorphous nanospecies can be captured with the z-factor model, as long as the in vitro data is biorelevant in terms of composition and dose-to-volume ratio. Secondly, for drugs that are limited by the UWL in vivo, typically drugs with permeability values > 2 to 3 x 10-4 cm/s, the impact of drug species like micelle-bound drugs or nanospecies can be captured by modifying the effective permeability used for simulation. This operates under the assumption that the UWL permeability is much lower than the epithelium permeability, which is a common characteristic for BCS Class II drugs.^10,11 This method, which is known as the nano-modified permeability method, in essence captures these effects by accounting for the overall size and abundance of each drug-containing species relative to unbound drug.¹² As a result, an increase to the effective permeability is captured for simulation and compared to a standard baseline permeability to provide a range of anticipated in vivo exposure. This method is most useful for capturing the impacts of food effects driven by increased bile-salt solubilization in the fed state and for drugs that have very low aqueous solubility (<10 µg/mL), but form a large amount of drug nanospecies upon ASD dissolution. As Figure 3 shows, the nano-modified model predicted in vivo performance much more accurately.

Precipitation events in ASD models in low- and moderate-pH environments
Typically, the most difficult phenomenon to capture in an ASD absorption model is precipitation, particularly crystallization, because it is often difficult to estimate in vitro compared to the in vivo situation. Additionally, multiple precipitation events from ASDs can occur down the GI tract as a function of pH, fluid volumes and components in GI fluids that attenuate supersaturation (e.g., bile-salt micelles or other lipidic components). These events include drugs precipitating to a drug-rich amorphous phase and crystallizing to a lower-energy drug form with a lower solubility, either simultaneously or sequentially. Many commercial absorption modeling software options can accommodate these scenarios, allowing for multiple precipitation events or solubilities of different drug forms (see Figure 4).

Amorphous phase separation as a key indicator of the amorphous solubility
Amorphous phase separation commonly occurs after dissolution or when the ASD enters a low-solubility environment, such as the duodenum, from a high-solubility environment, such as the stomach, (i.e., a basic drug) and is a key indicator of the amorphous solubility. This transition represents the point at which drug is saturated in the aqueous phase and begins to form a second drug-rich phase that remains amorphous, but is not freely dissolved in solution. For basic compounds, this event usually occurs rapidly when the drug encounters the higher pH of the intestine. This process should be reflected accurately in the absorption model with a rapid mean precipitation time. Although precipitation is rapid, it decreases to the maximum solubility attainable from the amorphous form as a function pH and GI fluid composition, and is often much higher than the corresponding solubility of the crystalline drug form. For neutral or acidic compounds, amorphous phase separation can occur upon dissolution of the ASD, or in the event of decreasing GI fluid volume—e.g., water absorption.

The importance of ASD crystallization
Crystallization is another important phenomenon to capture because it defines the rate at which the absorption benefit provided by the amorphous drug form will diminish over time. This can be measured in vitro with a test that is biorelevant in terms of fluid composition, pH, dose to volume ratio and transit time. Crystallization rates are often best captured in in vitro systems that provide biorelevant transit of drug, such as controlled transfer dissolution systems or dissolution permeation systems with an absorption compartment.^14,15 Removing drug from solution or transiting drug over time while dissolution is occurring better depicts the extent of drug supersaturation and, as a result, crystallization rate.

Crystallization rate is often the most uncertain parameter in an ASD absorption model—even when generated using the most-accurate in vitro systems—because it is difficult to perfectly capture this stochastic event as it occurs in a complex environment in vivo. The best approach is to perform a parameter sensitivity analysis (PSA) on precipitation rate to capture a range in expected in vivo performance, or to optimize precipitation time against observed data, assuming a high degree of confidence in other parameters such as solubility and permeability, to establish a framework for making future in vivo predictions.

Conclusions
Predicting the in vivo performance from an ASD formulation can be one of the most challenging aspects of formulation development given the interplay of dissolution, drug speciation and precipitation. ASDs are and will remain an important area of research over the coming years, given their proven ability to improve the oral absorption of poorly soluble drugs. Establishing a robust strategy for predicting in vivo exposure from ASDs will be a key focus area among researchers and formulators moving forward. Pharma & biotech companies developing low-solubility molecules may benefit from working with contract development and manufacturing organizations (CDMOs) that can bring to bear resources in this research & development area. 

Note: All Figures are located in the image slider at the top of the article.

References 

1. Wilson, M., et al., Hot-melt extrusion technology and pharmaceutical application. Therapeutic Delivery, 2012. 3(6): p. 787-797.
2. Dobry, D., et al., A Model-Based Methodology for Spray-Drying Process Development. Journal of pharmaceutical innovation, 2009. 4: p. 133-142.
3. Friesen, D.T., et al., Hydroxypropyl Methylcellulose Acetate Succinate-Based Spray-Dried Dispersions: An Overview. Molecular Pharmaceutics, 2008. 5(6): p. 1003-1019.
4. Mudie, D.M., et al., A novel architecture for achieving high drug loading in amorphous spray dried dispersion tablets. International journal of pharmaceutics: X, 2020. 2: p. 100042-100042.
5. Stewart, A., et al., Mechanistic Study of Belinostat Oral Absorption From Spray-Dried Dispersions. Journal of Pharmaceutical Sciences, 2019. 108(1): p. 326-336.
6. Takano, R., et al., Oral Absorption of Poorly Water-Soluble Drugs: Computer Simulation of Fraction Absorbed in Humans from a Miniscale Dissolution Test. Pharmaceutical Research, 2006. 23(6): p. 1144-1156.
7. Taylor, L.S. and G.G.Z. Zhang, Physical chemistry of supersaturated solutions and implications for oral absorption. Advanced Drug Delivery Reviews, 2016. 101: p. 122-142.
8. Wilson, V., et al., Relationship between amorphous solid dispersion in vivo absorption and in vitro dissolution: phase behavior during dissolution, speciation, and membrane mass transport. Journal of Controlled Release, 2018. 292: p. 172-182.
9. Stewart, A.M., et al., Impact of Drug-Rich Colloids of Itraconazole and HPMCAS on Membrane Flux in Vitro and Oral Bioavailability in Rats. Molecular Pharmaceutics, 2017. 14(7): p. 2437-2449.
10. Sugano, K., Fraction of a dose absorbed estimation for structurally diverse low solubility compounds. International Journal of Pharmaceutics, 2011. 405(1): p. 79-89.
11. Sugano, K., Theoretical Framework III: Biological Membrane Permeation. Biopharmaceutics Modeling and Simulations, 2012: p. 64-121.
12. Stewart, A.M. and M.E. Grass, Practical Approach to Modeling the Impact of Amorphous Drug Nanoparticles on the Oral Absorption of Poorly Soluble Drugs. Molecular Pharmaceutics, 2020. 17(1): p. 180-189
13. Kesisoglou, F., et al., Effect of Amorphous Nanoparticle Size on Bioavailability of Anacetrapib in Dogs. Journal of Pharmaceutical Sciences, 2019. 108(9): p. 2917-2925.
14. Xu, H., et al., In vitro characterization of ritonavir formulations and correlation to in vivo performance in dogs. European Journal of Pharmaceutical Sciences, 2018. 115: p. 286-295.
15. Polster, C.S., et al., Mechanism for Enhanced Absorption of a Solid Dispersion Formulation of LY2300559 Using the Artificial Stomach Duodenum Model. Molecular Pharmaceutics, 2015. 12(4): p. 1131-1140.