QSAR study of lipid peroxidation-inhibition potential of some phenolic antioxidants

Abstract

QSAR models have been developed for lipid peroxidation (LPO)-inhibition potential of a series of phenolic compounds in phosphate-buffered and pre-emulsified linoleic acid systems using the descriptors such as topological indices, topological charge indices and quantum chemical descriptors. In both model systems, a considerably large number (131) of descriptors have been used. The subsequent variable selection is made using standard techniques like stepwise regression and multiple linear regression (MLR) with factor analysis (FA) as the data-preprocessing step for variable selection (FA-MLR). The best equation is obtained from FA-MLR (R² = 0.950, Q² = 0.914) for the first system, and from stepwise regression analysis (R² = 0.960, Q² = 0.949) for the second system. The results reveal that bond dissociation enthalpy (BDE) and maximal electrotopological positive variation (MAXDP) have negative contributions to the LPO-inhibition activity. The best equations satisfy the Fmax criteria suggested by Livingstone and Salt (2005). The derived equations may be helpful in designing novel phenolic antioxidants.