P118: Sample Size Considerations in Biosimilar Pharmacokinetic Studies: A Systematic Review
Poster Presenter
Evgeniia Artamonova
Master student
Technical University of Munich (TUM) Germany
Objectives
This study aims to evaluate the quality of sample size planning in publications of biosimilar confirmatory pharmacokinetic (PK) trials conducted from 2015 to 2025. Additionally, it compares the statistical considerations of the sample size planning with the observed parameters in these studies.
Method
A systematic review was conducted via PubMed for trials demonstrating bioequivalence or biosimilarity (2015–2025). Literature screening was performed using Rayyan. Articles reporting sample size planning were included, and relevant data were extracted and numerically verified.
Results
The search identified 1462 records; 418 confirmatory articles contained sample size information. We focused on the 140 biosimilar studies in this analysis, while the 278 bioequivalence studies will be presented separately. The proportion of studies reporting all key sample size parameters (test/reference ratio (T/R), coefficient of variation (CV), and power) increased over time, from approximately 33.3% in 2015 to 75% in 2025. In most studies, the planned sample size closely matched the number of enrolled participants, and planned dropout rates exceeded observed dropout in 93.2% of studies. Most trials assumed a 5% deviation for the test/reference ratio and 90% statistical power. The Coefficient of Variation (CV) of the PK endpoints Cmax and AUC is a major driver of the sample size. For Cmax, 85.1% of the trials had observed lower CV than what they had planned when designing the trials, while for AUC this was 81.4%, indicating conservative assumptions regarding variability. Smaller discrepancies between planned and observed CV were associated with lower planned CV, subcutaneous administration (vs intravenous) and multiple-dose studies (vs single-dose). A funnel plot of log test/reference ratios against their standard errors showed a symmetrical distribution, indicating consistent effect estimates across studies.
Conclusion
The findings demonstrate continued improvements in the completeness of sample size reporting in biosimilar pharmacokinetic trials and highlight the use of conservative planning assumptions for variability and dropout rates. By systematically comparing planned inputs with observed outcomes across a large body of evidence, this study provides practical benchmarks that can support more informed and transparent design of future biosimilar studies, potentially reducing unnecessary study burden. These insights are relevant for statisticians, clinical pharmacologists, sponsors, and regulators seeking to optimize study efficiency while maintaining confidence in bioequivalence conclusions. Ultimately, this study shows how accumulated empirical evidence may help refine sample size assumptions, support methodological consistency, and contribute to the development of robust and efficient biosimilar programs.
