(W0930-03-17) Molecular Structure-Based Predictions of Amorphous Solid Dispersion Formation Are More Accurate When Appropriate Solvent and Faster Solidification Kinetics Are Used

Purpose: The R3m molecular descriptor model can predict successful formation of amorphous solid dispersions (ASDs) in a carrier polymer, polyvinylpyrrolidone vinylacetate (PVPVA), based solely on drug structure. This model was previously developed for melt-quenching (fusion) and rotary-evaporation (solvent-evaporation) methods using 15 drugs at 2 concentrations, where R3m >0.632 values predicted successful ASD formation. Although the prediction accuracy was 100% when the melt-quenching method was used, the prediction accuracy was lower when prepared by rotary-evaporation. Inaccurate predictions with solvent-based preparations were observed for drugs with R3m both above and below 0.632. Previously, successful ASD formation of drugs with R3m < 0.632 was shown to be due to improved mixing efficiency in the presence of solvent. In the current study, we investigate unsuccessful ASD formations for itraconazole (ITZ) prepared by spray-drying. This drug was not fully dispersed in PVPVA when rotary-evaporated from methanol solution. We hypothesize that more rapid solidification during spray-drying from a solvent in which ITZ is more soluble will lead to successful ASD formation of the drug in PVPVA.

Methods: The 3D coordinates of a single molecular conformation of ITZ were generated using the CORINA algorithm, allowing calculation of a single R3m value using E-Dragon (version 5.4, vcclab.org). Mixtures of ITZ-PVPVA at 15% or 75% w/w drug-loading were spray-dried from methanol (ITZ solubility = 0.06 g/100 mL), similar to the previous rotary-evaporation method. For comparison, these same mixtures were spray-dried and rotary-evaporated from dichloromethane (ITZ solubility = 25 g/100 mL). Samples were vacuum-dried and stored in a P<sub>2O₅ desiccator (~0% RH) until characterization. Dispersibility in PVPVA was inferred from powder X-ray diffraction (PXRD), differential scanning calorimetry (DSC), and polarized light hot stage microscopy data.

Results: ITZ (R3m = 0.872) was predicted to form homogeneous ASD with PVPVA as its R3m value was greater than 0.632. Co-solidified mixtures were considered dispersed if characterization by all three methods indicated non-crystalline and homogeneous systems. If one or more techniques revealed either crystallinity or phase separation, samples were classified as not dispersed. The dispersibility results are shown in Table 1. In the previous rotary-evaporation studies, ITZ-PVPVA was observed to be not dispersible at 15% drug loading. At 75% drug loading, ITZ -PVPVA was not rotary-evaporated owing to solvent volume >1.0 L. Similar to the rotary-evaporation method, ITZ did not disperse in PVPVA when spray-dried from methanolic solutions. When the solvent was changed to dichloromethane, the solubility of ITZ improved by >400-fold. Consistent with R3m predictions of dispersability, ITZ-PVPVA formed ASDs when spray-dried at both concentrations, and when rotary-evaporated at 15% drug-loading using dichloromethane. The 75% drug-loaded ITZ-PVPVA mixture, however, remained non-homogeneous when rotary-evaporated from dichloromethane. This was attributed to different solvent drying kinetics between the methods, in addition to different solvents within the same method. The vapor pressure of dichloromethane (boiling point = 39.6°C) was higher than that of methanol (boiling point = 64.7°C). Moreover, the inlet air temperature during spray-drying was 90°C, while the water bath used for rotary-evaporation was maintained at 50°C. In addition to the larger vapor pressure and lower volume of dichloromethane, the higher temperature used for spray-drying resulted in faster solidification when ITZ-PVPVA-dichloromethane solutions were spray-dried relative to rotary-evaporation using methanol. Furthermore, the rate of solvent-evaporation affected the length of time that the ITZ-PVPVA-solvent solutions remained supersaturated, increasing the probability of forming non-homogeneous co-solidified mixtures.

Conclusion: When the solubility of ITZ in the processing solvent improved from 0.06 g/100 mL to 25 g/100 mL, homogeneous ASDs were obtained at lower drug-loading when prepared by either spray-drying or rotary-evaporation. At higher drug-loading, however, the drying kinetics during rotary-evaporation were insufficiently rapid to obtain homogeneous ASDs. Overall, the R3m predictions were accurate for a higher R3m drug when spray-dried using a solvent in which the drug was freely soluble.


Table 1: Dispersibility results for ITZ-PVPVA mixtures with rotary-evaporation and spray-drying methods