Higher Order Structure Methods for Similarity Assessment

Abstract

Guidance from the EMA and FDA suggest a stepwise approach for assessing biosimilarity, an approach that leverages both structural and functional characterization of the biosimilar product to define appropriately-sized non-clinical and clinical studies. Because higher order structure (HOS) dictates protein function and stability, HOS is a key product quality attribute for which demonstration of analytical similarity is essential; by extension, characterization of the HOS of a protein biosimilar can aid in reducing residual uncertainty and informing appropriately sized non-clinical and clinical studies. A review of seven biosimilar advisory committee briefing documents showed a wide range of diversity in HOS methods utilized for similarity assessment to date. No correlation was observed between the types of methods selected, the number of methods, the number of reference product lots characterized, or the subsequent non-clinical or clinical study designs. The diversity in method selection appears to arise from two factors: the range of opinions across the industry on the ability of HOS methods to inform technical decisions, and the regulatory risk tolerance of different organizations. These two factors inform an organization’s overall HOS similarity strategy, and each organization must balance speed, sensitivity, specificity, and cost to select the HOS characterization methods it applies to the similarity exercise. We recommend a quantitative approach for HOS method selection and analytical similarity study design. Qualifying HOS methods provides a quantitative measure of method sensitivity and specificity to better inform a method ranking process from which appropriate methods may be selected. These methods should then be applied with appropriate lot selection and with a sufficient number of lots, emphasizing trends over time in the reference product material. Quantitative assessment of method sensitivity and specificity combined with appropriate lot selection provides objective measures to reduce residual uncertainty and better inform the analytical similarity conclusion.