Biosimilar Approvals Streamlined With Advanced Statistics Amidst Differing Regulatory Requirements

The FDA and European Medicines Agency (EMA) mandate high similarity between biosimilars and reference products, but their regulatory processes differ, especially with multiple reference products. | Image Credit: thien - stock.adobe.com

Innovative statistical methods, such as simultaneous CIs and multiplicity-adjusted TOST (MATOST) were explored by researchers to address challenges posed by differing FDA and European Medicines Agency (EMA) regulatory requirements for biosimilar approvals, according to a review published in Biologics: Targets and Therapy.¹

These differences have prompted research into innovative statistical methods for more efficient and accurate biosimilar assessments. When a biosimilar has both US and EU reference products, sponsors must conduct 3 bridging studies comparing the biosimilar product with the US-licensed reference product, the biosimilar product to the EU-approved reference product, and the US-licensed reference productto the EU-approved one.

Despite Europe's greater biosimilar adoption, regulatory disparities and complex US approval processes hinder development, underscoring the need for global harmonization.²

The FDA and EMA also differ on foreign-approved comparators; the EMA permits them (with caveats), while the FDA requires 3-way comparisons.¹ Researchers criticize the conventional 3-way comparison for neglecting variability, underutilizing data, and creating multiple testing issues. This study investigated 2 alternatives, simultaneous confidence intervals and MATOST.

Regulatory Framework for Biosimilar Approval

Both agencies have slightly different definitions for biosimilars. The FDA defines biosimilars as biological products highly similar to the reference product, with no clinically meaningful differences between safety, purity, and potency compared with the reference product. However, the EMA defines biosimilars as biological medicinal products containing the same active substance as the reference product and similar in quality characteristics, biological activity, safety, and efficacy.

The FDA's stepwise approach assesses analytical similarity, investigating the structural and functional characteristics of the biosimilar product based on critical quality attributes at various stages of the manufacturing process. The stepwise approach includes a pharmacokinetics (PK) and pharmacodynamics (PD) similarity assessment, demonstrating PK and PD through human clinical pharmacology studies after passing animal toxicity studies, and a clinical similarity assessment, gathering information on immunogenicity, clinical efficacy, and safety.

Additionally, the main purpose of the 3-way comparison is to ensure the drug substance or product meets product specifications, to demonstrate biosimilarity between the test product and the US-licensed product, and to establish a bridge justifying the use of clinical data generated using the EU-approved reference product as the comparator. Strict requirements on biosimilar products are necessary to ensure the efficacy and safety of biosimilars on the market.

Challenges in Biosimilar Assessment

The drug sponsor should provide 3-way bridging evidence when using a non–US-licensed product in a regulatory submission for a biosimilar product. This evidence justifies the use of comparative data from the non–US-licensed product in assessing biosimilarity and bridging to the US-licensed reference product.

Disadvantages, such as underutilizing data collected from the entire study, can impair conventional pairwise comparisons, potentially biasing and misleading results. Since pairwise comparisons do not use the same reference product in each comparison, and each comparison may use a different equivalence acceptance criterion margin, the precision of each equivalence test varies. Multiple comparisons will inflate the type 1 error rate unless some significance-level adjustment approaches are considered. Overall, these pairwise comparisons cannot distinguish the relationship among the biosimilar product, the US-licensed reference, and the EU-approved reference.

Without loss of generality, the US-licensed reference product and the EU-approved reference product are considered as the 2 reference products, with the US-licensed reference as the primary reference.

Comparative Analysis of Statistical Approaches

The simultaneous CI method in a parallel study design used fiducial probabilities to perform pairwise equivalence tests. The simulation study used 4 methods for pairwise comparison and included 3 simultaneous CI methods with different assumptions and 1 conventional approach. The results found all approaches can maintain desired power when the sample size per arm is larger than 20. The simultaneous CI method under equal variability has the best performance among all the methods and can achieve similar statistical power to the conventional approach. The study found that the simultaneous CI method works well in parallel study design.

Among simulation results, the Holm and Bonferroni method were very conservative, and TOST was more likely to fail to reject the null hypothesis to control type 1 error inflation. In a crossover design with multiple formulations, especially within the Williams design, the MATOST method is not favorable since it may lead to an unrealistically large sample size, though it can control the type 1 error rate.

Addressing Challenges in Equivalence Testing

The FDA requires sponsors submitting biosimilar products for regulatory approval to conduct a 3-way pairwise comparison when both US-licensed and EU-approved reference products exist. Determining equivalence margins in the equivalence tests is crucial. Using sample estimation in the hypothesis setting is unnecessary because it does not consider the variability of the estimation.

Generalizing the CI method will allow researchers to determine equivalence margins by sampling standard deviations and considering the variability of the estimation. Researchers need to propose another method for controlling type 1 error inflation specifically for crossover designs to reduce the large sample size required for conducting a biosimilar study.

“For future research, it is possible to consider further utilization of fiducial inference under crossover designs and generalize the simultaneous CI method...,” the authors concluded. “Generalizing the simultaneous CI method can help researchers in determining equivalence margins by sampling standard deviations and taking the variability of estimation into consideration.”

References

1. Pong A, Chow S, Chow SC. Comparison of innovative and conventional methods in biosimilar bridging studies with multiple references. Biol.: Targets Ther. 2024;18:377-387. doi:10.2147/btt.s470182
2. Jeremias S. Improving biosimilar access through global regulatory convergence. The Center for Biosimilars^®. January 15, 2025. Accessed February 24, 2025. https://www.centerforbiosimilars.com/view/improving-biosimilar-access-through-global-regulatory-convergence