The search for cytotoxic synergy between anticancer agents: a case of Dorothy and the ruby slippers?

In the 1939 movie, "The Wizard of Oz," Dorothy (played by Judy Garland) has many adventures while seeking to return home from the fantasy dream world of Oz. At the end of the movie, she finds out from the good witch of the North that she always had it in her own power to return home by merely clicking together the heels of her magical ruby slippers. The search for synergy, especially with in vitro antiproliferation assays, is reminiscent of Dorothy and the ruby slippers. In some ways, the whole enterprise is like a dream or a fantasy, with thousands of published studies, a myriad of conflicting definitions for interaction terms, many different combined-action assessment procedures, and a wide spectrum of interpretations of similar results. On the other hand, it is often assumed that proper and easy synergy assessment is possible, but that it is necessary for some wizard to tell us the secret. This editorial will briefly discuss two topics: 1) definition of combined-action terms and 2) interpretation and use of a finding of in vitro synergy. A comprehensive critique of synergy-assessment procedures has been published previously by our group (7). A report in this issue of the Journal by Kaufmann et al. (2) will be used to illustrate specific points. When two or more entities (people, organizations, drugs, machine parts, etc.) show synergy, by any definition, it is usually considered to be a positive attribute of the combination. In everyday language, synergy is often used to mean "working together in a cooperative, productive manner." The U.S. Patent Office recognizes synergy as one criterion for the necessary characteristic of "unobviousness" for an invention that is a combination of two or more distinct units. Synergistic combinations are thought to be interesting. They are considered special. If you add synergy to a group, you get a team. Unfortunately, these everyday notions of synergy, which have strong mechanistic connotations, have had a major and often confusing effect on the thinking about drug-drug synergy. Synergy (or antagonism) between two chemical or physical agents is an empirical phenomenon, in which the observed effect of the combination is more (or less) than what would be predicted from good knowledge of the effects of each agent working alone. This general empiric definition of synergy has been interpreted differently by many scientists who have created rival nomenclature schemes and combined-action assessment approaches. A key difference among the rival schemes and approaches is the choice of the null reference for "no interaction" between the agents. The most common null reference models are the Loewe additivity and the Bliss independence models. Loewe additivity would be observed for a drug combined with itself or perhaps for a drug combined with a very close structural analogue that acts at the same target. For Bliss independence, the fractional response caused by a combination of two agents at specific concentrations is equal to the product of the fractional responses of each agent applied alone at the specific concentrations. Different, independent, mutually nonexclusive sites of action for the two agents are implied. In addition, two of the most commonly used synergy assessment approaches for in vitro studies of combinations of anticancer agents, the method of Steel and Peckham (3) and the method of Chou and Talalay (4), each add an extra null reference model to the mix: the Mode II model (similar, but not the same as Loewe additivity) and the mutually nonexclusive model (similar, but not the same as Bliss independence), respectively. [References (7) and (5) include extensive discussions of terms and definitions.] Also, it should be noted that the validity of the derivation of the mutually nonexclusive model of Chou and Talalay (4) and the main formula used to calculate the combination index for the mutually nonexclusive case have been questioned by our group (7). Unambiguous links between empirical findings of synergy (or antagonism) and particular cellular or biochemical mechanisms rarely exist. A major problem with assigning mechanisms to findings of synergy is suggested by the work of Jackson (6). Using computer simulations of simple biochemical pathways, he found that by tweaking the values of enzyme kinetic parameters, or by changing the type of enzyme inhibition (competitive, noncompetitive), or by introducing or removing feed-