Multiplicity Adjustment in Bioequivalence Using Two One-Sided Tests (TOST)

• Activity Number: 179
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Steven Hua*+ and Siyan Xu and Ronald Menton
• Companies: Pfizer Research and Boston University and Pfizer Inc.
• Keywords: Multiplicity ; Bioequivalence ; TOST ; FWER ; Closure test principle ; Sequential adaptive

Abstract

When assessing bioequivalence (BE) of two drugs, it's a common practice that two pharmacokinetic parameters AUC and Cmax are either marginally tested against two one-sided null hypotheses using Schuirmann's two one-sided tests (TOST) or to show that 90% confidence intervals of the ratio in each PK parameter (of two drugs) is comprised in the 0.80-1.25 range. But less discussed is the fact that, as both PK parameters are measured on each subject, there are 4 one-sided null hypotheses (two for each parameter), without a proper multiplicity adjustment, the family-wise error rate which is the probability of incorrectly rejecting at least one null hypothesis in a family of multiple tests, alpha_F, could otherwise rise above that of the nominal type I error rate (alpha). This paper proposes two approaches using TOST, one is to apply the closure test principle so that BE can be declared with FWER controlled at alpha_F=alpha, the other is a sequential procedure to allow control of FWER by pre-specifying the significant level on AUC and setting the significant level for Cmax adaptively after testing of AUC. The methods are illustrated using data from a crossover PK bioavailability study.