Calculation of lipophilicity of a large, diverse dataset of anticancer platinum complexes and the relation to cellular uptake.

A quantitative structure--property relationship (QPSR) for the octanol--water partition of platinum complexes was constructed using molecular descriptors derived from density functional (DFT) calculations. A dataset of partition data for 64 complexes, consisting of 43 square-planar platinum(II) and 21 octahedral platinum(IV) complexes, was drawn from literature sources. Not only does this dataset include considerable structural diversity of complexes considered but also a variety of techniques for the measurement of partition coefficients. These data were modeled using descriptors drawn from electrostatic potentials and hardness/softness indices projected onto molecular surfaces. This required initial descriptor selection using a genetic algorithm approach, followed by partial least-squares regression against log Po/w data. In this way, a statistically robust model was constructed, with errors of similar size to the variation in log Po/w from multiple experimental measurements. Implications of lipophilicity for cellular accumulation of Pt-based drugs, and hence for design of new drugs, are discussed, as is the uptake of metabolites of cisplatin.