In Silico Prediction of Betulinic Acid Derivatives’ Cytotoxicity: Relationship between Topological Descriptors and CC50 Value

Abstract

Modeling of the relationship between structure and cytotoxicity has helped in the process of designing safer new drug compounds. In this study, modeling was carried out between the structures of 29 betulinic acid derivatives with their cytotoxicity. The modeling is done by using multiple linear regression (MLR) techniques. In the model, an equation is obtained by involving five descriptors and has statistical parameters as r2training of 0.776; Fcal/Ftab of 4.503; r2test of 0.985; r2m of 0.971. The five descriptors involved in the equation are TDB2e (3D topological distance-based autocorrelation-lag 2/weighted by Sanderson electronegativities), TDB9s (3D topological distance-based autocorrelation-lag 9/weighted by I-state), RDF50m (radial distribution function-050/weighted by relative mass), RDF140m (radial distribution function-140/weighted by relative mass), and RDF10s (radial distribution function-010/weighted by relative I-state). The equation could be used to design the new betulinic acid derivatives with lower predicted cytotoxicities regarding the coefficients of the descriptors. In this case, the new substituent is chosen to enhance the value of RDF140m and RDF10s, while also to make the value of TDB23, TDB9s, and RDF50m getting lower, so the CC50 value will rise (the compound become less toxic to the normal cell).