Where is the right balance in designing an optimal phase III biosimilar trial?
*Yulan Li, Novartis  Yodit Seifu, Novartis Pharmaceuticals Corporation 

Keywords: non-inferiority, biosimiarity, asymmetric margins

Development of biosimilars to originator therapeutic biologics (reference biologic) promises the reduction of health care cost, providing patients worldwide with greater access to effective treatments. The demonstration of high similarity between the biosimilar and the reference biologic at the pharmaceutical/technical level is critical for the development of biosimilars (EMA draft guidance, 2013) (FDA draft guidance, 2012). A high level of analytical and functional similarity of the biosimilar and the reference biologic products serves as the foundation of the biosimilar drug development program. Similarity is to be further demonstrated through a focused nonclinical development program. Finally, any remaining residual uncertainty will be addressed by a tailored clinical program. In this talk, we share our thoughts on several statistical related issues (design types, choice of NI margin, type I error rate, etc) as well as the right balance between scientific rigor and feasibility in optimizing biosimilar clinical programs to address remaining residual uncertainty. In our opinion, fundamental principles used for setting the non-inferiority margin should be appropriately applied in the context of biosimilars due to unique features associated with the biosimilar clinical trials (choice of disease model, choice of primary endpoint, statistical objectives, relevance of historical data for determining the NI margins, etc). We also share our recent statistical methodology discussions with the EMA Innovative Task Force (ITF) on biosimilar clinical trial designs. Equivalence designs (with lower and upper margins) are usually preferred for comparison of efficacy and safety of the biosimilar and reference products. Non-inferiority design (requiring one margin) may be appropriate, if appropriately justified. Applying a non-inferiority design allows confirmation of similarity without the burden of testing something that is implausible – particularly when even a higher dose of the originator in the setting would not result in improvement. These considerations allow for a reasonably high ability to rule out pre-specified clinically meaningful differences between the two products, should they exist without ignoring feasibility of the trial conduct, thus providing patients with more affordable treatments in a timely fashion.