579P - Methodological considerations for clinical equivalence of bevacizumab biosimilars

Date 18 December 2016
Event ESMO Asia 2016 Congress
Session Poster lunch
Topics Anticancer Agents
Presenter Dominik Heinzmann
Citation Annals of Oncology (2016) 27 (suppl_9): ix184-ix189. 10.1093/annonc/mdw603
Authors D. Heinzmann
  • Biostatistics Oncology, F.Hoffmann-La Roch, 4074 - Basel/CH



Clinical studies between the biosimilar and reference medicinal product should be conducted to demonstrate comparable clinical efficacy, in adequately powered, randomized, parallel group comparative clinical trial(s). Equivalence margins need to be selected based on statistical and medical reasoning using reference product data, such that potential loss in long-term efficacy is mitigated. This analysis aimed to identify equivalence margins for assessing bevacizumab (bev) biosimilar candidates.


We applied meta-analyses and regression models to randomized bev clinical trial-level data, to assess equivalence margins for bev biosimilar candidates. The indications assessed were ovarian cancer (OC), advanced non-small cell lung cancer (NSCLC), metastatic colorectal cancer (mCRC), metastatic renal cell cancer (mRCC), and metastatic breast cancer (mBC).


Selection of the non-inferiority margin for ORR (i.e. lower margin in an equivalence trial) is provided by the lower bound of the 95% CI of the meta-analysis. Equivalence margins for ORR of 10% (required sample sizes around 900 patients) are associated with large predicted maximum loss (PML) in long-term efficacy, with exception of NSCLC where the predicted loss in PFS may be clinically acceptable (Table). For mCRC, mBC and advanced OC, the predicted loss at an equivalence margin for ORR of 7% is clinically not acceptable, suggesting that for those indications, the margin would need to be lower than 7%, corresponding to a sample size of 1500+.rn

Table: 579P Meta-analytic regression of ORR on long-term efficacy endpoint PFS



The analyses conducted suggest that the magnitude of ORR benefit is heterogeneous across indications and do not provide a large effect size in all indications. For NSCLC, ORR may be considered with an equivalence margin of 10%. When using ORR in mBC, mRCC, equivalence margins would be 

Clinical trial indentification

Not applicable

Legal entity responsible for the study

F. Hoffmann-La Roche Ltd, Basel, Switzerland




D. Heinzmann: Employee of F. Hoffmann-La Roche Ltd