Long-Term Physical (In)Stability of Spray-Dried Amorphous Drugs: Relationship with Glass-Forming Ability and...

This study shows the importance of the chosen method for assessing the glass-forming ability (GFA) and glass stability (GS) of a drug compound. Traditionally, GFA and GS are established using in situ melt-quenching in a differential scanning calorimeter. In this study, we included 26 structurally diverse glass-forming drugs (i) to compare the GFA class when the model drugs were produced by spray-drying with that when melt-quenching was used, (ii) to investigate the long-term physical stability of the resulting amorphous solids, and (iii) to investigate the relationship between physicochemical properties and the GFA of spray-dried solids and their long-term physical stability. The spray-dried solids were exposed to dry (<5% RH) and humid (75% RH) conditions for six months at 25 °C. The crystallization of the spray-dried solids under these conditions was monitored using a combination of solid-state characterization techniques including differential scanning calorimetry, Raman spectroscopy, and powder X-ray diffraction. The GFA/GS class assignment for 85% of the model compounds was method-dependent, with significant differences between spray-drying and melt-quenching methods. The long-term physical stability under dry condition of the compounds was predictable from GFA/GS classification and glass transition and crystallization temperatures. However, the stability upon storage at 75% RH could not be predicted from the same data. There was no strong correlation between the physicochemical properties explored and the GFA class or long-term physical stability. However, there was a slight tendency for compounds with a relatively larger molecular weight, higher glass transition temperature, higher crystallization temperature, higher melting point and higher reduced glass transition temperature to have better GFA and better physical stability. In contrast, a high heat of fusion and entropy of fusion seemed to have a negative impact on the GFA and physical stability of our dataset.