Statistical methods for assessment of biosimilarity using biomarker data

Abstract

The problem for assessing biosimilarity between biologic products is studied. For approval of follow-on biologic products, the U.S. Food and Drug Administration (FDA) indicated that the follow-on biologic products can be approved under an abbreviated new drug application (ANDA) if the innovator products are approved under a new drug application (NDA). However, for biologic products that are licensed under a BLA, there exists no abbreviated BLA in current Codes of Federal Regulations (CFR). In this case, draft guidance for assessment of biosimilarity is being prepared. As indicated in Chow and Liu (2008), the assessment of bioequivalence for drug products is performed under a so-called fundamental bioequivalence assumption, which uses pharmacokinetic responses as the surrogate endpoint for clinical endpoint for evaluation of the safety and efficacy of the drug products. Following a similar idea, in this article, statistical methods for assessment of biosimilarity between a follow-on biologic product and an innovator product are derived under a fundamental biosimilar assumption and a probability-based criterion for biosimilarity using biomarker data, assuming that the biomarker is predictive of the clinical outcome of the biologic product.