Distinguishing rational from irrational applications of pharmacogenetic synergies from the bench to clinical trials.

Single therapeutic agents very often fail in unselected patients. It is therefore commonplace to combine an agent specifically with a selected patient subgroup or with another agent. To support such efforts, it is useful to clarify the distinctions between the terms and the mathematical models used in analyzing combinations. To incorporate molecular disease classifications, the familiar concept of the therapeutic window is modified to define a pharmacogenetic window, which is an unambiguous numerical measure of the magnitude of interaction produced by a combination, and to define a test of pharmacogenetic synergy. In contrast, certain common comparative methods, such as vertical windows (comparing effects at a given dose) and animal models of mutational targets may be dominated by undesirable features. Although this discussion is oriented towards cancer therapy, an extension of these concepts to other comparative biologic assays is feasible and advisable.