Retention Behavior On Immobilized Artificial Membrane Chromatography And Relationship With Drug Absorption And Distribution

The purpose of this study was to investigate the correlation between drug retention behavior on immobilized artificial membrane (IAM) and molecular structure, absorption and distribution of drugs, which was performed by using chemometrics methods.The quantitative structure-retention relationship was based on linear solvation energy relationship (LSER), which is only suitable for small neutral compounds. The retention behavior of ionized compounds was regarded as that of all molecules with charge, so the new variable (molecular average net charge) was introduced in the LSER to get semi-empirical amended LSER, which is suitable to analyze the retention behavior of ionized compounds. Retention of carboxylic compounds is different with other compounds and carboxylic group can reduce the retention significantly. Acidic compounds without carboxylic group had similar retention behavior with same lipophilic neutral compounds. Ionization of basic compounds has different impact on the retention and secondary amine can increase the interaction between basic compounds and IAM. Besides, electrostatic shielding, space matching, et al can influence the retention of basic compounds, which can not be ignored. To get empirical amended LSER with theoretical base, new variable log [1-D(1-f)] was constructed and used as sixth variable and the amended LSER obtained can reflect the instinct character of the IAM system.Artificial neural network and partial least squares regression were used to obtain QSRR according to physicochemical properties, such as clog P, Molecular Weight (MW), Total Surface Area (TSA), et al, of drugs. And external test set was used to verify the prediction ability of the QSRRs. Partial least squares regression was preformed to choose variables and the model was simplified, which was convenient to quantitatively study the impact of the properties on drug retention in IAM.The log k_IAM and larger one of drugs under pH 6.0 and 7.0 conditions were used to predict the intestinal fraction of absorption (FA), and IAM system with pH 6.0 had better prediction ability, which was reasonable because average pH in intestinal fluid is 5.5-6.0. The log k_IAM corrected by MW³ had Sigmoid relationship with FA and can be used to predict and classify the FA of drugs.The log k_IAM of pH 7.0 was linear correlated to drug blood-brain barrier penetration (log BB). The correlation can be improved after log k_IAM was corrected by MW. 3-variable relationship was obtained by using Stepwise regression and hydrogen donor and refractivity have negative coefficients, indicating hydrogen bond donor and molecular volume have negative impact on drug permeation through blood-brain barrier.The log k_IAM under pH 6.0 and 7.0 had no linear relationship with drug skin permeation (log Kp).Δlog k_IAM was obtained from log k_IAM and clog k_IAM based on clog P- log k_IAM relationship for neutral compounds.Δlog k_IAM represents the hydrogen bond, electrostatic and space forces in drug-IAM interaction. Better multiple inear relationship was found between log Kp andΔlog k_IAM(corrected by MW) and clog P under pH 6.0 conditioin than pH 7.0. Besides, PSA and log Kp had sigmoid relationship, which was verified by external test set composed of 14 drugs.