A study on antimalarial artemisinin derivatives using MEP maps and multivariate QSAR.

Artemisinin and some derivatives with activity against D-6 strains of Plasmodium falciparum were studied. Molecular electrostatic potential (MEP) maps were used in an attempt to identify key features of the compounds that are necessary for their activities, and then use those to propose new artemisinin derivatives. The partial least squares (PLS) method was then used to generate a predictive model. The PLS model with three latent variables explaining 88.9% of total variance, with Q((2)) = 0.839 and R(2) = 0.935, was obtained for 15/6 compounds in the training/external validation set. For construction of the model, the most important descriptors were the highest occupied molecular orbital (HOMO) energy, atomic charges on the atoms O1 (Q(1)) and C3 (Q(3)), molecular volume (VOL), and hydrophilic index (HYF). From a set of 20 proposed artemisinin derivatives, one new compound (39) with higher antimalarial activity than the molecules initially studied was predicted. Synthesis of these new derivatives may follow the results of the MEP maps studied and the PLS modeling.