Computational prediction of enhanced solubility of poorly aqueous soluble drugs prepared by hot melt method

Solubility is the concentration of a solute in a saturated solution at a given temperature and pressure. Solubility of a drug in aqueous media is a pre-requisite to achieve desired concentration of a drug in the systemic circulation. Low aqueous solubility is a major problem encountered with formulation development of recently designed new chemical entities. Solubility of poorly soluble drugs is enhanced by physical and chemical modifications of drug. Shake flask method is the most commonly used experimental method to determine solubility. However, this method has several limitations. A single solubility experiment can go on for several days and even weeks. Besides this, a large amount of drug is required to carry out the experiment. In order to overcome this and make initial screening easier, computational method can be used to predict solubility. In this study, the solubility of 12 small molecules of BCS class II having a wide range of physicochemical properties were studied to enhance their solubility by hot melt method. Three different grades of PEG (1450, 4000, 8000), PVP K17 and Urea as the hydrophilic carriers was employed for the solubility enhancement. The overall objective of this investigation is to develop a model that could estimate enhanced solubility using physicochemical descriptors. Multiple linear regression (MLR), a statistical tool, was used to generate a equation for the solubility by correlating physicochemical properties of the drug like- molecular size, logP, pKa, HBA, HBD, melting point, polar surface area, and number of rotatable bonds. Solubility enhancement is also influenced by the carrier used, we included the physicochemical properties of the carriers like molecular weight and solubility parameter in the development of the model. MLR analysis model, resulted in an equation, where, Log solubility = 5.982-0.010 MW (drug)-0.452 LogP-0.320 HBA-0.095 ?solubility parameter+0.015 MV. A regression analysis yielded a good fit with a regression value (adjusted R2) of 0.74. The model has been validated by leave one out method. This model has the potential to estimate the solubility of a physically modified drug in screening stages of drug development