According to the WHO, artemisinin-based combination therapies (ACTs) have been integral to the recent reduction in deaths due to Plasmodium falciparum malaria. ACT-resistant strains are an emerging problem and have evolved altered developmental stages, reducing exposure of the most susceptible stages to artemisinin drugs in popular ACTs. Lipophilicity, log Kow, is a guide in understanding and predicting pharmacokinetic properties such as terminal half-life which alters drug exposure. Consistent log Kow values are not necessarily available for artemisinin derivatives designed to extend terminal half-life, increase bioavailability, and reduce neurotoxicity. For other drugs used in ACTs, an assortment of experimental and computational log Kow values are available in the literature and in some cases, do not account for subtle but important differences between closely related structures such as between diastereomers. Quantum chemical methods such as density functional theory (DFT) used with an implicit solvent model allow for consistent comparison of physical properties including log Kow and distinguish between closely related structures. To this end, DFT, B3LYP/6-31G(d), with an implicit solvent model (SMD) was used to compute ΔGowo and ΔGvowo for 1-octanol-water and olive oil-water partitions, respectively, for 21 antimalarial drugs: 12 artemisinin-based, 4 4-aminoquinolines and structurally similar pyronaridine, and 4 amino alcohols. The computed ΔGowo was close to ΔGowo calculated from experimental log Kow values from the literature where available, with a mean signed error of 2.3 kJ/mol and mean unsigned error of 3.7 kJ/mol. The results allow assignment of log Kow for α-and β-diastereomers of arteether, and prediction of log Kow for β-DHA and five experimental drugs. Linear least square analysis of log Kow and log Kvow versus terminal elimination half-life showed strong linear relationships, once the data points for the 4-aminoquinoline drugs, mefloquine and pyronaridine were found to follow their own linear relationship, which is consistent with their different plasma protein binding. The linear relationship between the computed log Kvow and terminal elimination half-life was particularly strong, R2 = 0.99 and F = 467, and can be interpreted in terms of a simple pharmacokinetic model. Terminal elimination half-life for β-DHA and four experimental artemisinin drugs were estimated based on this linear relationship between log Kvow and terminal t1/2. The computed log Kow and log Kvow values for epimers α- and β-DHA and α and β-arteether provide physical data that may be helpful in understanding their different pharmacokinetics and activity based on their different molecular geometries. Relative solubility of quinine and quinidine are found to be sensitive to thermal corrections to enthalpy and to vibrational entropy and do not follow the general trend of longer terminal t1/2 with greater predicted log Kow. Geometric relaxation of α- and β-DHA in solvent and inclusion of thermal correction for enthalpy and entropy results in correct prediction that α-DHA is favored in aqueous environments compared to β-DHA. Predictions made regarding experimental drugs have implications regarding their potential use in response to artemisinin drug-resistant strains.
Bibliographical noteFunding Information:
The authors are grateful to the Louis Stokes North Star STEM Alliance for funding that supported student participation in this research and to the University of Minnesota Morris Chemistry Discipline, Science and Math Division, and Dean’s Office for the sabbatical during which a major part of this work was completed. We also thank Gaussian technical support for help with use of SMD to model 1-octanol and olive oil solvation properties.