Quantitative estimation of intra-subject variability in bioequivalence studies of generic solid oral dosage forms by multiple regression analysis

Abstract Intra-subject variability (CVintra), which determines the 90% confidence intervals and the number of subjects needed for assessment in bioequivalence studies, is generally investigated by using pilot study Results. However, conducting pilot studies greatly affects the speed and cost of drug development. In this study, we performed multiple regression analysis of the major factors that predict the extent of the CVintra of pharmacokinetic parameters by using 4 quantitative variables (drug solubility, variability of the peak drug concentration [Cmax] or area under the drug plasma concentration versus time curve [AUC], bioavailability, and elimination half-life), which were obtained from a database of the results of bioequivalence studies conducted by Nihon Generic Co., Ltd. A total of 77 studies [34 components] were included. From this analysis, we confirmed that multiple regression equations can be used to estimate 3 kinds of %CVintra values with high correlation coefficients (r = 0.8971 for the Cmax of all drug types, r = 0.9739 for the AUC of renal excretion-type drugs, and r = 0.5368 for the AUC of non-renal excretion-type drugs). When the predicted equations were applied to newly planned bioequivalence studies (2 studies [2 components]) without pilot studies, it was verified that the %CVintra of the Cmax and AUC could be estimated with a predicted value of ±5. Further, both formulations were found to be bioequivalent in term of the Cmax and AUC, with 90% confidence intervals of sufficient power.