ABSTRACT
Following our ongoing studies on structure-activity relationship studies of propafenone-type modulators of multidrug resistance, we performed both a Free-Wilson analysis and a combined Hansch/Free-Wilson analysis on a set of 48 compounds using artificial neural networks (ANN). In comparison to classical multiple linear regression (MLR) analysis, the ANN showed equal or even slightly better predictive power in leave one out cross validation procedures and was remarkably superior when performing a leave 8 out cross validation. Additionally, it was possible to train a network using only 14 compounds and to properly predict the MDR-modulating activity of the remaining 34 compounds. In this case, the MLR analysis completely failed due to insufficient number of cases. Attempts to extract informations on which input descriptors are important using a genetic input selection algorithm failed. Best results were obtained using those descriptors which showed highest statistical significance in MLR analyses.
Subject(s)
Daunorubicin/pharmacokinetics , Daunorubicin/toxicity , Drug Resistance, Multiple , Neural Networks, Computer , ATP Binding Cassette Transporter, Subfamily B, Member 1/genetics , ATP Binding Cassette Transporter, Subfamily B, Member 1/metabolism , ATP-Binding Cassette Transporters/genetics , ATP-Binding Cassette Transporters/metabolism , Antineoplastic Agents/chemistry , Antineoplastic Agents/pharmacokinetics , Antineoplastic Agents/toxicity , Cell Membrane/metabolism , Drug Design , Genes, MDR , Humans , Multidrug Resistance-Associated Proteins , Predictive Value of Tests , Regression Analysis , Reproducibility of Results , Structure-Activity Relationship , Tumor Cells, CulturedABSTRACT
A series of 48 propafenone-type modulators of multidrug resistance was synthesized and their P-glycoprotein inhibitory activity was measured using the daunomycin efflux assay. Both a Free-Wilson and a combined Hansch/Free-Wilson analysis were performed using log P, partial log P and molar refraction values as Hansch descriptors. The results of the Free-Wilson analysis show that modifications on the central aromatic ring generally influence pharmacological activity, whereby in almost all cases a decrease in MDR-modulating potency is observed (Q2cv = 0.66). The combined approach results in equations with remarkably higher predictive power (Q2cv = 0.83), specifically molar refractivity shows high significance in all equations derived. This indicates that polar interactions also contribute to protein binding.
