An occasional challenge in the form selection of novel solid oral therapeutics is when the drug substance persistently crystallizes as a solvated crystal structure rather than an anhydrous free form necessary for final drug product formulation. The stability of anhydrous forms relative to solvates can be promoted by elevating the temperature of the crystallization above the coexistence temperature of the two crystals. The challenge is to find a temperature where the anhydrous form is most stable yet still below the boiling point of that particular solvent.
Here, we present a procedure to rigorously determine the “desolvation window” for a given solvate-anhydrous pair using a combined computational and experimental approach. The solubilities of the solvate and anhydrous form are measured at ambient temperature to determine their relative stability. Then, theoretical estimates of the desolvation entropy are used to extrapolate the stability to higher temperature and determine the coexistence point relative to the solvent boiling point. We demonstrate this approach on the marketed oncology drug Axitinib and compare the predicted desolvation window for multiple solvates to the one determined through rigorous experimentation including high temperature solubility measurements and high temperature competition slurries. Finally, we discuss various ways that this predictive workflow could be utilized without complete data such as in compounds where no anhydrous form is known experimentally.
