A fast method to predict solubility of neutral co-crystals in aqueous media

Co-crystals have recently attracted growing interest as enabling formulations for poorly soluble drugs, due to their ability to enhance various physicochemical properties of active pharmaceutical ingredients (APIs). Different co-crystals of the same drug compound are formally regarded as the same active ingredient although they typically have different solubilities. As solubility is one of the key parameters for dissolution profiles, using different co-crystals of the same API can tailor the dissolution behavior. The classical experimental method to determine and compare co-crystal solubility in aqueous media relies on measuring drug and co-former concentrations at the eutectic point. However, these measurements are time- and material- consuming, requiring extensive sampling and sample processing while the eutectic point is being chased. This paper introduces a new and simple approach to predict co-crystal solubility in any medium, including aqueous buffer at physiological pH. The method, based on thermodynamic principles, estimates the solubility product (K_sp) in a target medium by measuring i) the API and coformer concentrations in the solution obtained during cocrystal synthesis and ii) the individual solubilities of the API and coformer in the synthesis solvent and the target medium. From these easily accessible experimental data, K_sp and solubility of the cocrystal in the target medium, such as water, can be derived. The method was validated using two example systems: carbamazepine-benzamide and cilnidipine-p-toluenesulfonamide, which are co-crystals of neutral components. Predicted values aligned well with experimental eutectic point measurements, demonstrating the method's effectiveness even when the two co‑crystals differed in solubility by two orders of magnitude. The approach is advantageous as it minimizes experimental work, and solubility data of the individual components can be collected once and reused for other co-crystals. The prediction method appears suitable for a fast and reasonably accurate estimation of co-crystal solubility during early phase of drug development and solid form selection. The method is currently limited to 1:1 co-crystals of neutral components.