Geometric mean ratio-dependent scaled bioequivalence limits with leveling-off properties

In this study, novel approaches for the design of bioequivalence (BE) limits are developed. The new BE limits scale with intrasubject variability but only until a geometric mean ratio (GMR)-dependent plateau value and combine the classic (0.80-1.25) and expanded (0.70-1.43) BE limits into a single criterion. Plots of the extreme GMR values accepted as a function of coefficient of variation (CV) have a convex shape, similar to the classic unscaled 0.80-1.25 limits. The performance of the novel approaches in comparison to the classic unscaled 0.80-1.25 limits as well as the two expanded BE limits, i.e., 0.70-1.43 and 0.75-1.33 was assessed using simulated data. Two-period crossover BE investigations with 12, 24 or 36 subjects were simulated with assumptions of CV 10%, 20%, 30% or 40%. At low CV values, the performance of the novel BE limits is almost identical to the 0.80-1.25 criterion. On the contrary, the expanded BE limits are very permissive even at high GMR values. For high CV% values (30% and 40%), the new BE limits show a much greater probability of declaring BE when GMR = 1 in comparison to the classic 0.80-1.25 limits. In addition, when the drug products differ more than 25%, the new BE limits show much lower percentage of acceptance than the expanded 0.70-1.43 limits. One of the major advantages of the new BE limits is their gradual expansion with variability until a GMR-dependent plateau value. Finally, the continuity and leveling-off properties of the new BE limits make them suitable for the assessment of BE studies, irrespective of the level of variability encountered. (c) 2005 Elsevier B.V. All rights reserved.