Simultaneous confidence interval methods for analytical similarity assessment.

Area of Science:

• Biopharmaceutical Science
• Statistical Modeling
• Regulatory Science

Background:

• Analytical similarity assessment is crucial for biosimilar drug development.
• Current methods often rely on pairwise comparisons, which can be problematic with multiple reference products.
• Ensuring statistical similarity of key quality attributes between test and reference products is a regulatory requirement.

Purpose of the Study:

• To propose a novel simultaneous confidence approach for analytical similarity assessment of biosimilar products.
• To provide an alternative to the traditional pairwise comparison method, particularly for cases with multiple references.
• To evaluate the performance of the proposed method against the pairwise method through simulation studies.

Main Methods:

• Development of a simultaneous confidence approach based on fiducial inference theory.
• Proposal of three versions of simultaneous confidence intervals, considering different population variance assumptions.
• Extensive simulation studies to compare the proposed method with the pairwise comparison method.

Main Results:

• The proposed simultaneous confidence approach demonstrated comparable or superior performance to the pairwise method in simulation studies.
• The study highlighted potential concerns and limitations associated with the exclusive use of the pairwise comparison method.
• The fiducial inference-based approach provides a robust alternative for assessing similarity with multiple references.

Conclusions:

• The simultaneous confidence approach offers a statistically sound and potentially more robust method for biosimilar analytical similarity assessment.
• This method addresses limitations of pairwise comparisons, especially in complex scenarios involving multiple reference products.
• The findings support the adoption of advanced statistical methods to enhance biosimilar regulatory science and product quality assurance.