Background No formal guidelines exist to inform the selection of appropriate equivalence margins or the timing of efficacy assessment for comparative effectiveness trials of biosimilars in rheumatoid arthritis (RA). Studies to date have specified a single time-point for assessment of the primary outcome and an equivalence margin based on experience in placebo-controlled trials of the reference product. Neither of these adequately describes the dynamic early responses to TNF inhibition nor provides an optimally sensitive measure to assess bioequivalence. However, by applying multiple statistical analyses across frequent early time points, a “totality-of-the-evidence” might be generated that could sufficiently demonstrate clinical similarity.
Methods To compare the statistical evaluation of bioequivalence using aggregate information from early time-points to that using data from the primary endpoint at week 16, 189 subjects with active RA on stable doses of oral methotrexate were randomized 2:1 to receive double-blinded treatment with BOW015 or reference infliximab (rIFX) and assessed for efficacy at weeks 0, 2, 6, 14, and 16. The pre-specified primary endpoint was the proportion of subjects achieving an ACR20 response at week 16, as previously reported. In this novel post-hoc analysis, a generalized estimating equation, logistic regression, and an exponential time response model were applied to aggregate data from the early time-points in this trial. Aggregated and integrated analysis were compared to the assessment of bioequivalence using data only from the primary endpoint.
Results At the week 16 primary endpoint, comparable proportions of BOW015- and rIFX treated-patients achieved an ACR20 response, yielding a 95% confidence interval (CI) for the treatment difference of -12.6% to 19.3% by the protocol-specified 2-sided Fisher's Exact test methodology. Analyses to assess the average proportion of ACR20 responders over multiple time-points during the first 16 weeks provide a more stringent assessment of equivalence, with 95% CIs for the treatment difference that fall within a narrower range than that achieved using the Exact test at week 16 (Table 1). Non-linear modeling further confirmed this comparable time-response profile across 16 weeks in agreement with averaged findings.
Conclusions Assessment of therapeutic equivalence by aggregating multiple time-points during the steep phase of the time response curve, using appropriate statistical methods to validate attainment of bioequivalence margins, can provide a more comprehensive assessment of biosimilarity than a measure taken at a single timepoint.
Disclosure of Interest G. Castañeda-Hernández Consultant for: Amgen Inc.; AbbVie Inc.; AstraZeneca; Boehringer Ingelheim GmbH; Epirus Biopharmaceuticals, Inc.; Janssen Biotech, Inc.; Laboratorios Sophia S.A.; Merck Serono; Merck Sharp & Dohme Corp.; Novartis Pharmaceuticals Corp.; Pfizer Inc.; Roche Laboratories, Inc.; Sanofi, and UCB Inc., M. Wyand Shareholder of: Epirus Biopharmaceuticals, Employee of: Epirus Biopharmaceuticals, C. Lassen Shareholder of: Epirus Biopharmaceuticals, Employee of: Epirus Biopharmaceuticals, L. Shneyer Consultant for: Epirus Biopharmaceutical, Merck Pharmaceuticals; J. Kay Grant/research support from: (paid to the University of Massachusetts Medical School): from AbbVie Inc.; Eli Lilly and Company; Pfizer Inc.; and Roche Laboratories, Inc., Consultant for: Amgen Inc.; AbbVie Inc.; AstraZeneca; Boehringer Ingelheim GmbH; Bristol-Myers Squibb Company; Crescendo Bioscience, Inc.; Eli Lilly and Company; Epirus Biopharmaceuticals, Inc.; Genentech Inc.; Hospira, Inc.; Janssen Biotech, Inc.; Merck Sharp & Dohme Corp.; Nippon Kayaku Co., Ltd.; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Regeneron Pharmaceuticals, Inc.; Roche Laboratories, Inc.; Samsung Bioepis; UCB, Inc.