RESUMO
A computational model for the prediction of solubilizers' effect on drug partitioning has been developed. Membrane/water partitioning was evaluated by means of immobilized artificial membrane (IAM) chromatography. Four solubilizers were used to alter the partitioning in the IAM column. Two types of molecular descriptors were calculated: 2D descriptors using the MOE software and 3D descriptors using the Volsurf software. Structure-property relationships between each of the two types of descriptors and partitioning were established using partial least squares, projection to latent structures (PLS) statistics. Statistically significant relationships between the molecular descriptors and the IAM data were identified. Based on the 2D descriptors structure-property relationships R(2)Y=0. 99 and Q(2)=0.82-0.83 were obtained for some of the solubilizers. The most important descriptor was related to logP. For the Volsurf 3D descriptors models with R(2)Y=0.53-0.64 and Q(2)=0.40-0.54 were obtained using five descriptors. The present study showed that it is possible to predict partitioning of substances in an artificial phospholipid membrane, with or without the use of solubilizers.
Assuntos
Simulação por Computador , Modelos Químicos , Compostos Orgânicos/química , Cromatografia , Solubilidade/efeitos dos fármacos , Relação Estrutura-AtividadeRESUMO
The possibility of developing a quantitative relationship between molecular structure and lymphatic transfer of lipophilic compounds co-administered with a long-chain triglyceride vehicle was examined. Molecular descriptors were calculated using the computer program VolSurf, and lymphatic transfer data were derived from the literature. A significant structure-property relationship was established using partial least squares, projection to latent structures statistics (PLS). R(2)X was 0.77, R(2)Y was 0.83 and the prediction power of Q(2) was 0.73 in the two-component PLS model. A number of descriptors contributed to the prediction leading to a complex model, but the prediction power was improved with the PLS model when compared to the frequently used method by relating logP values (LogKow) with lymphatic transfer.