Reliability of the Hansen solubility parameters as co-crystal formation prediction tool

Pharmaceutical co-crystals present an opportunity to improve the solubility of conventional active pharmaceutical ingredients (APIs). Despite advances in co-crystal screening, the rational design of even the chemically simplest co-crystals remains challenging. Hansen solubility parameters (HSPs) have previously been used as a tool to predict co-crystal formation using only the chemical structure. The aim of this study was to validate the use of HSPs as a tool to predict co-crystal formation, analyse its limitations and examine the previously set Δδ inclusion cut-off value. A total of 109 co-formers of carbamazepine, caffeine and theophylline were used as a training set. Sixteen different descriptors were examined. An additional 72 co-formers of piroxicam and nicotinamide were used to test the methods and new cut-off values. The established cut-off value (8.18 MP_a^0.5) despite being similar to the previously reported value (7 MP_a^0.5), offered no real advantage over the previously reported value. Our results suggest the use of the modified radius (Ra) method of calculating the solubility difference, which had higher sensitivity of 90% compared to 86% for the previously reported method and cut-off value to indicate co-crystal formation as well as a lower miss and false omission rates.