Flexible Unified Drug Interaction Models for Estimating a Drug Combination’s Efficacy That Depends on a Specific Variable or Multiple Variables (303196)

Keywords: Drug synergy, Interaction index, Loewe additivity, Bliss independence

When multiple drugs used in combination produce enhanced or reduced effect than that predicted by their individual potencies, the combination is synergistic or antagonistic, respectively. It is measured by the interaction index (tau): tau >1 for synergistic effect and <1 for antagonistic effect. Two popular response surface models representing additivity (tau=1), Loewe additivity and Bliss independence, have been incorporated with tau to test for synergism. However, a drug combination’s effect may depend on some factors. For example, it may be gender-specific. We propose a powerful and flexible unified approach to allow tau to vary across drug dose levels and depends on one or more variables. A set of hierarchical models corresponding to different assumptions about tau are fitted first and then goodness of fit tests are used to test for each specific assumption. Practical strategies for fitting such complex non-linear curves are recommended based on extensive simulation studies. The approach was illustrated through a pre-clinical study aiming to determine and compare the combined anti-inflammatory effect of dexamethasone and pentoxifylline between human newborn and adult blood.