Classifying drugs

CAMBRIDGE, Mass.—Multidrug treatments are becoming increasingly prevalent but little effort has gone into determining whether the drugs will interact with each other in unexpected ways. Researchers at Harvard University therefore examined a series of antibiotics for their impact pair-wise on bacteria and classified the drugs into distinct families. They presented their work in Nature Genetics (2006, 38, 489-494).

 

The scientists used a bioluminescence assay of E. coli growth rates to measure the impact of pairs of antibiotics at sublethal doses. They categorized the pairings as additive, synergistic, antagonistic buffering or antagonistic suppression, and then used the Prism algorithm to classify the drugs into subgroups. They found that most drugs classified easily into interaction groups. Interestingly, compound similarities within each subgroup were based on cellular function rather than chemical structure.

 

They also found that drugs they could not subgroup often impacted more than one biological pathway, while drugs that classified into their own group often exhibited novel mechanisms-of-action. These last two points offer possible insights into the battle to combat drug resistance in microbes or cancer cells.