Dissolution mechanisms for the sparingly soluble ionizable compounds in aqueous solution under laminar hydrodynamic conditions

The dissolution of a poorly soluble drug is usually the rate-limiting step in its absorption from the gastrointestinal tract. In this study, the dissolution of two sparingly soluble basic drugs, cinnarizine and papaverine, was observed in a laminar flow channel device described by Shah and Nelson. General dissolution models were developed and solved for ionizable drug dissolution under laminar hydrodynamic conditions.;Two general mechanistic models were developed to simulate ionizable drug dissolution in unbuffered or buffered solutions of various pH values. The first, a kinetic dissolution model, considers reaction, diffusion, and fluid convection in the dissolution process. It uses scaled rate constants to describe the fast equilibrium reactions. The model solution predicts the correct relationship between drug dissolution rate and affecting factors in unbuffered or buffered solutions from pH 2 to 12. However, the equilibrium composition of each component could not be attained because further rate constant increases resulted in computationally instability of the computer program.;The second model, an equilibrium dissolution model, imposed acid-base equilibria differently by using the equilibrium constant expressions in the transport equations directly. This model also includes the simultaneous processes of diffusion, convection and ionic migration. A robust numerical method utilizing the point successive overrelaxation (PSOR) method was developed to solve the model equations for a wide range of medium pH, buffer species and concentration. The results are substantially closer to the experimental data than are those from the kinetic dissolution model because the equilibrium relationships are imposed in this model. The equilibrium dissolution model also gives charge balanced concentration profiles for all species by considering ionic migration.;The third model, consisting of a neural network, was created and applied specifically to model the complex dissolution behavior of drug/buffer co-compressed tablets using the limited training data. The ability to generalize the optimal network to new drugs, buffers and environments was improved by preprocessing and dynamic partitioning the training data. This training protocol was successful in increasing network generalizability without enlarging the training error. The resulting optimal neural network accurately calculated ionizable drug and drug/buffer dissolution rates in different pH solutions.