Rescuing poorly soluble drugs: a cheminformatics approach for polymeric micelle-drug formulations

Several drug candidates have failed therapeutic development because of poor aqueous solubility. Polymeric micelle formulations were shown to alleviate this problem for some but not all hydrophobic drug molecules. We have conceived a novel computer-aided strategy for the rational design of polymeric micelle-based delivery systems for poorly soluble drugs. Herein, we have (i) rationally designed a library of poorly soluble and chemically diverse drugs and tested them for loading efficiency (LE) and loading capacity (LC); (ii) developed novel chemical descriptors for polymers and drug-polymer complexes; (iii) generated and interpreted QSPR models for drug loading into polymeric micelle-based delivery systems; (iv) identified, by virtual screening, drugs with poorly aqueous solubility predicted to have either high or low LE and LC; and (vi) experimentally validated model predictions achieving a correct classification rate of ca. 75%. The success of the strategy described herein suggests its broad utility for designing other drug delivery systems.