The determination of biosimilarity margin and the assessment of biosimilarity for an -arm parallel design

One of the key issues in biosimilar phase III clinical trials is the determination of biosimilarity margins. Although biological products generally have high product variability, it is not reflected in the methods most commonly used for determining biosimilarity margins. In order to take into account high product variability, in this paper, we propose (Formula presented.) -arm parallel design, which consists of one biosimilar product group and m reference product groups. Because there are m reference product groups, we can estimate the between-batch variation of reference products using a random effect model. A statistical testing procedure appropriate for assessing biosimilarity using (Formula presented.) -arm parallel design is developed based on asymptotic theory. From simulation studies, we conclude that at least three reference product batches are needed for practical use of the proposed method.