Since Loewe described chemical synergism and additivism in 1928, multicomponent therapies whose efficacy exceeds that of their individual constituents have arisen by both design and happenstance, and are now standard treatments for cancer and infectious diseases. Synergistic combinations of drugs already in the clinic might therefore be a rich source of potential therapies, and a relatively unexploited one, due in part to the daunting number of combinations that can be generated. Writing in the Proceedings of the National Academy of Sciences, Borisy and colleagues now describe a systematic high-throughput approach to mining this resource.

Dubbed 'combination high-throughput screening' (cHTS), the new technology lies at the heart of the authors' company CombinatoRx, and can be applied to the analysis of two-component and higher-order combinations. The first step is to categorize compounds on the basis of individual activity. Each active compound is then paired with every other compound — both active and inactive — at six different concentrations, and assayed again. By contrast, inactive single entities are pooled in groups of four before testing. Active 'pools' are then deconvoluted to identify the specific pair that confers activity — a more efficient strategy made possible by the lack of overlapping activities and the relatively low frequency at which active combinations form from inactive single agents.

Borisy et al. used their cHTS system to find pairs of drugs that are more efficacious than their individual constituents against three disease-causing entities. From a total of about 120,000 combinations, 22 pairs with antifungal activity against fluconazole-resistant Candida albicans were identified. Six of these contained a known antifungal and a non-antifungal agent, with the remaining sixteen comprising two non-antifungals — none of the pairs comprised two known antifungal agents. Twenty-six combinations that suppress production of the immunostimulatory cytokine tumour-necrosis factor-α in human blood cells were also discovered.

In their final assay, the authors identified thirteen new synergistic pairs that inhibit the proliferation of tumour cells. When tested in a mouse model of human lung carcinoma, the combination of the antipsychotic chlorpromazine and the antiprotozoal pentamidine was more effective than the anticancer drug paclitaxel, and did not cause the side effects that commonly result from standard chemotherapy. So it seems that, at least in drug discovery, the whole can be greater than the sum of its parts.