(231ao) Application of the COSMO-SAC-Dsp Model in Drug Solubility Prediction

The information of drug solubility is important in drug discovery, development, and manufacturing. Finding the optimal solvent or solvent combinations for desired solubility (i.e., solvent screening) requires making many solubility measurements. Although the techniques of solubility measurement have been improved, it is nonetheless a costly and time-consuming task to measure these data for all drug candidate at all possible operating conditions (temperature, composition of a mixed solvent, etc.). Therefore, it is useful to have a predictive thermodynamic model, which does not rely on experimental data, for the design and development of drugs. The prediction of drug solubility in various pure and mixed solvents is evaluated using the recently COSMO-SAC-dsp model. The solubility data (range from 10^-1 to 10^-6 in mole fraction) of 43 drug compounds in 37 different solvents and their combinations over a temperature range of 273.15 K to 323.15 K (1773 data points) are calculated from the COSMO-SAC-dsp model and compared to experiments. When only the pure drug properties (heat of fusion and melting temperature) are used, the average absolute error of drug solubility prediction from COSMO-SAC-dsp is 0.455 log-unit and is reduced by 4.2 % when comparing with that (0.475 log-unit) of COSMO-SAC (2010) model. When the heat of fusion and melting temperature of the drug are not available, predictions can still be made with a similar accuracy using the solubility data of the drug in other solvent or solvent mixture. Our results show that the COSMO-SAC-dsp model can provide reliable predictions of drug solubility and is a useful tool for drug discovery.