Cocrystal Solubility Product Prediction Using an in combo Model and Simulations to Improve Design of Experiments

To predict the aqueous solubility product (K _sp ) and the solubility enhancement of cocrystals (CCs), using an approach based on measured drug and coformer intrinsic solubility (S ₀^API , S ₀^cof ), combined with in silico H-bond descriptors.

A regression model was constructed, assuming that the concentration of the uncharged drug (API) can be nearly equated to drug intrinsic solubility (S ₀^API ) and that the concentration of the uncharged coformer can be estimated from a linear combination of the log of the coformer intrinsic solubility, S ₀^cof , plus in silico H-bond descriptors (Abraham acidities, α, and basicities, β).

The optimal model found for n:1 CCs (-log₁₀ form) is pK _sp  = 1.12 n pS ₀^API  + 1.07 pS ₀^cof  + 1.01 + 0.74 α^API·β^cof - 0.61 β^API; r ²  = 0.95, SD = 0.62, N = 38. In illustrative CC systems with unknown K _sp , predicted K _sp was used in simulation of speciation-pH profiles. The extent and pH dependence of solubility enhancement due to CC formation were examined. Suggestions to improve assay design were made.

The predicted CC K _sp can be used to simulate pH-dependent solution characteristics of saturated systems containing CCs, with the aim of ranking the selection of coformers, and of optimizing the design of experiments.