Switching Design for Assessment of Interchangeability in Biosimilar Studies.

Background: In biosimilar studies, assessing the switchability and interchangeability of biosimilars with their reference products is essential for ensuring reliable clinical evaluation. This study explores optimal trial design strategies incorporating balanced and uniform structures to enhance statistical efficiency in treatment effect under a carryover setting. Methods: Using a linear mixed-effect model for log-transformed responses, we conducted a theoretical variance-based evaluation of all possible two-treatment switching designs in three-period and four-period crossover trials, considering settings with and without carryover effects. A total of 247 distinct three-period designs and 65,519 distinct four-period designs were enumerated and classified according to structural properties, with particular attention to those incorporating a non-switching arm (NSA). Results: SBUwP-NSA (Strongly Balanced Uniform-within-Period designs with a Non-Switching Arm) consistently achieved the minimum variance for treatment effect estimation in both carryover and no-carryover settings. In the absence of carryover effects, UwP-NSA (Uniform-within-Period designs with a Non-Switching Arm) attained equivalent efficiency. In contrast, commonly used dedicated switching designs exhibited substantially lower relative efficiency, achieving as little as 50-55% of the efficiency of the optimal designs, depending on carryover assumptions. Conclusions: This comprehensive theoretical evaluation demonstrates that incorporating strong balance and uniformity properties can yield substantial efficiency gains in switching studies. The results provide quantitative guidance for selecting efficient crossover designs, enabling improved estimation precision while maintaining practical relevance for interchangeability and switching assessments in biosimilar research.

Duke Scholars

Author Shein-Chung Chow Biostatistics & Bioinformatics, Division of Biostatistics