Statistical Assessment Of Biosimilarity Based On Modified Relative Distance

Recently more and more innovative (brand-name) biological products are going off patent protection, the assessment of biosimilarity between a biosimilar product (or follow-on biological) and an innovative biological product has received considerable attention. For assessment of biosimilarity, the standard methods for assessment of bioequivalence for generic (small molecule) drug products may not be appropriately applicable because of fundamental differences between a small molecule (chemical) drug product and a biologic (large molecule) product. For approval of biosimilar products, some corresponding regulatory agencies has published several product-specific concept papers as regulatory guidelines for approval pathway of biosimilar products. However, the assessment of biosimilarity is still in primary stages, further research and explore may be needed.So far, the most commonly used design for assessing biosimilarity between the biosimilar product and the innovative biological product is a three-arm balanced design, and biosimilarity is assessed by the ralative distance of the two discrepancies, one from the biosimilar product and the innovative product, and the other from the different batches of the innovative product Therefore, the biosimilar products have greater inherent variability, the method based on the relative distance only focus on the mean differences and suppose a same fixed variance of the biosimilar products and the innovative products. In this way, we may come to an inaccurate conclusion. To cure the above problem, this paper presents the modified relative distance, which add the differences of the standard deviation into the relative distance to assess the biosimilarity between the biosimilar products and the innovative products. As the distribution of the statistic modified relative distance is very complicated, we choose bootstrap method to conduct the statistical analysis.Contents:1. Compare the type-I error and the power of the bootstrap and the ratio estimator method based on the relative distance; Test the type-I error and the power using bootstrap based on the modified relative distance;2. Check the impact on the type-I error and the power when the biosimilar limits are different.Results:1. Bootstrap method performs fairly with the ratio estimator method on the type-I error and the power based on the relative distance. Bootstrap method is able to control the type-I error and maintain the power at a relatively high degree.2. A relatively strict biosimilarity limit leads to higher type-I error and lower power which suggests that a loose biosimilarity limit may be a better choice.Conclusion:It is the right way to assess the biosimilarity between the biosimilar products and the innovative products using bootstrap method based on the modified relative distance.