Statistical considerations in biosimilar clinical efficacy trials with asymmetrical margins

Development of biosimilars to innovative therapeutic biologics promises reduction of healthcare cost and therefore will provide patients worldwide greater access to effective treatments. Because of the differences in raw materials or manufacturing processes, 'equivalence' of bioavailability between a biosimilar and the reference biologic is generally regarded as insufficient, and thus, clinical trials providing efficacy and safety data are often required by regulatory agencies. The traditional non-inferiority trial design may not be accepted for establishing biosimilarity in order to avoid superior efficacy with additional safety (e.g., immunogenicity) risks. On the other hand, the bioequivalence trial design, which is used in the generic paradigm for the evaluation of bioavailability of generic chemical drugs, is not appropriate for evaluating clinical efficacy because the equivalence margins are generally too wide and not justified on statistical or clinical grounds. Motivated by the World Health Organization guideline and the newly released Food and Drug Administration draft guideline on biosimilars, we propose a biosimilarity trial design for evaluating clinical efficacy. The design uses a non-inferiority margin and an asymmetrical non-superiority margin for statistical inference. The independent choice of both margins provides the scientific foundation for drawing clinical efficacy conclusions while maintaining the logical consistency of the inference. The design also has a higher statistical power than a naïve equivalence trial design.