Correlation of human jejunal permeability (in vivo) of drugs with experimentally and theoretically derived parameters. A multivariate data analysis approach.

The effective permeability (Peff) in the human jejunum (in vivo) of 22 structurally diverse compounds was correlated with both experimentally determined lipophilicity values and calculated molecular descriptors. The permeability data were previously obtained by using a regional in vivo perfusion system in the proximal jejunum in humans as part of constructing a biopharmaceutical classification system for oral immediate-release products. pKa, log P, and, where relevant, log Pion values were determined using the pH-metric technique. On the basis of these experiments, log D values were calculated at pH 5.5, 6.5, and 7.4. Multivariate data analysis was used to derive models that correlate passive intestinal permeability to physicochemical descriptors. The best model obtained, based on 13 passively transcellularly absorbed compounds, used the variables HBD (number of hydrogen bond donors), PSA (polar surface area), and either log D5.5 or log D6.5 (octanol/water distribution coefficient at pH 5.5 and 6.5, respectively). Statistically good models for prediciting human in vivo Peff values were also obtained by using only HBD and PSA or HBD, PSA, and CLOGP. These models can be used to predict passive intestinal membrane diffusion in humans for compounds that fit within the defined property space. We used one of the models obtained above to predict the log Peff values for an external validation set consisting of 34 compounds. A good correlation with the absorption data of these compounds was found.