A Time-response Measure to Assess Clinical Equivalence in Rheumatoid Arthritis: an Assessment Using Data From Clinical Trials of Biosimilars. (3rd April 2022)

• Record Type: Journal Article
• Authors: O'Kelly, Michael; Zhang, Aijing; Lipkovich, Ilya; Song, Guochen; Reeve, Russell; Ratitch, Bohdana; Li, Siying; Behnke, Martha; Kay, Jonathan; Chow, Shein-Chung; Baek, Inyoung
• Is Part Of: Statistics in biopharmaceutical research. Volume 14:Number 2(2022)
• Page Start: 204
• Page End: 216
• Publication Date: 2022-04-03
• Subjects: bootstrap -- exponential distribution -- missing data -- time-response; Pharmacy -- Statistical methods -- Periodicals; Pharmaceutical biotechnology -- Statistical methods -- Periodicals; Biopharmaceutics -- Periodicals; Biometry -- Periodicals; Pharmacy -- Statistical methods; Periodicals; 615.190727
• DOI: 10.1080/19466315.2020.1832567
• Languages: English
• ISSNs: 1946-6315

Abstract

ABSTRACT: Because of structural complexity, a "biosimilar" will not be exactly the same as its reference biologic treatment, but is required to be equivalent in all relevant attributes, including efficacy. Therapeutic equivalence is often assessed at a single time point and trajectory up to that time point ignored. This paper describes a measure to assess therapeutic equivalence in rheumatoid arthritis that takes into account both the trajectory and the peak efficacy. This time-response measure is compared with the standard single-time-point measure via simulations based on recent clinical trials of biosimilars. Scenarios can be constructed where the single-time-point measure is more sensitive in detection of non-equivalence, particularly where the time-response curve is not monotone; but for a variety of trajectories the time-response measure has lower Type II error rate (higher power) for a given Type I error rate. Performance is adversely affected by missing data for both measures. A limitation of the time-response measure is that it assumes a two-parameter exponential model for the trajectory of efficacy over time. Results under poor model fit are also presented. Where similarity of clinical outcome over time is a concern, the time-response measure should be considered when comparing a biosimilar and its reference product.