(T0930-04-20) Hansen Solubility Parameters Estimates of Amorphous Dispersion Behavior Are Accurate in Some Cases but Often Inconclusive

Purpose: Selecting polymers for amorphous solid dispersions can be enabled by structure-based predictions. This study evaluates the utility of Hansen Solubility Parameters for identifying dispersion carriers based on the structure of the polymer and active pharmaceutical ingredient (API) molecules. Although solubility parameters have been extensively used to predict API-polymer miscibility, miscibility does not always lead to stable amorphous dispersions. To better understand the limits of solubility parameters in predicting dispersion behavior, this study tests the hypothesis that solubility parameter predictions of miscibility will agree with observed behavior.

Methods: Twelve API-polymer combinations (4 APIs and 3 interrelated polymers) were prepared using four APIs included cimetidine (CIM), sulfanilamide (SUL), indomethacin (IMC), and ketoconazole (KTZ), and three interrelated polymers (polyvinylpyrrolidone-PVP, polyvinyl acetate-PVAc, and the co-polymer PVPva). Melt-quench preparations of the API-polymer mixtures were prepared at 75% w/w API-loading. Dispersibility (formation of single homogeneous amorphous phase) of each API in each polymer was inferred from powder X-ray diffraction, differential scanning calorimetry, and polarized light hot stage microscopy. Co-solidified mixtures were considered dispersed if all methods indicated non-crystalline and disperse systems. If one or more techniques revealed either crystallinity or phase separation, samples were classified as not dispersed in the respective polymers. The Hansen partial solubility parameters reflecting dispersion interactions (δ_d), polar interactions (δ_p), and hydrogen bonds (δ_h), were calculated by the Van Krevelen and Hoftyzer method using group contributions for each force type. Fedor’s molar volume (M_v) for each moiety was adopted from the respective group contributions. The total solubility parameter (δ_T) was calculated as per the following equation:
δ_T=√((δ_p)² + (δ_d)² + (δ_h)²)
A homogeneous dispersion was considered likely if the difference between the total solubility parameters of each ingredient (∆δ_{T,(API-polymer)}) was < 2 (J/cm³)^1/2, unlikely if ∆δ_{T,(API-polymer)}>10 (J/cm³)^1/2, and uncertain if ∆δ_{T,(API-polymer)} ranged between 2–10 (J/cm³)^1/2. The experimental observations of dispersion behavior were compared to predictions made using the Hansen Solubility Parameters.

Results: The results for all twelve mixtures are provided in Tables 1 and 2. The ability of an API to form a homogeneous phase with a polymer is determined by the features of the API and the polymer at the molecular level. The functional groups and their accessibility within the molecular structure determines how one molecule will interact with another. It was expected that the Hansen solubility parameters would successfully predict dispersion formation as it captures the molecular features that favor molecular interactions between API and polymer. SUL had the highest δ_T=35.93 (J/cm³)^1/2, while the other API and polymers had much lower values, ranging from 21.50 to 26.28 (J/cm³)^1/2. Based on the results, CIM, IMC and KTZ were expected to be miscible and form stable dispersions with PVP, having ∆δ_T,(API-PVP) < 2 (J/cm³)^1/2). The only other API-polymer combination having a clear indication of miscibility was IMC in PVPva. The ∆δ_{T,(API-polymer)} for SUL and all three polymers predicted its immiscibility with each. In PVAc, a difference of 4.08 (J/cm³)^1/2) was dispersible (KTZ), but a difference of 3.34 (J/cm³)^1/2) was not (IMC). In PVPva, the API with the smallest differences (IMC and KTZ) were dispersible, while the highest differences (CIM and SUL) were not. However, the miscible API having the largest difference (KTZ, ∆δ_{T,(API-PVPva)}=2.65 (J/cm³)^1/2 was very similar to the immiscible API with smallest difference (CIM, ∆δ_{T,(API-PVPva)}=3.16 (J/cm³)^1/2. The predictions were most accurate in PVP and least accurate in PVac. Dispersion in PVP tends to be driven by hydrogen bond formation between the drug and the polymer. The vinyl acetate monomer is a weaker hydrogen bond acceptor, so dispersions in this polymer are less dependent on the formation of specific interactions. The molecule’s mobility seems to have played an important role in API-polymer mixtures where Hansen solubility parameters showed indefinitive predictions.

Conclusion: Although the solubility parameter approach did not misclassify the dispersion behavior, it also did not produce definite predictions in almost half of the API-polymer combinations. API-polymer combinations that were very similar (∆δ_{T,(API-polymer)} < 2 (J/cm³)^1/2) or very different (∆δ_{T,(API-polymer)}>10 (J/cm³)^1/2) were reliably predicted by solubility parameters, whereas uncertainty in the behavior of 5 combinations with intermediate differences led to inconclusive predictions. The results suggest that solubility parameters are most effective in systems where the outcome is dependent on the formation of specific intermolecular interactions but less useful otherwise. While we have studied only 12 API-polymer combinations, and further study is needed, it is clear that solubility parameters have value. However, this approach is insufficient for many drugs. An alternative or complimentary approach to predict dispersibility may be useful.


Table 1: List of solubility parameters for APIs and polymers for API-polymer systems.


Table 2: Dispersibility predictions by solubility parameter method and experimental outcomes in all three polymers.