Statistical Challenges Arising in Tech Transfers of Lyophilized Biologics and Improvements Made Using a Finite-Sample Chebyshev Inequality and a Lognormal Sum Approximation

• Activity Number: 411
• Type: Topic Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Lisa Bernstein*+
• Companies: Genentech
• Keywords: manufacturing ; lyophilized ; lognormal ; biologics ; Chebyshev inequality ; moisture

Abstract

Manufacturers of protein and antibody-drug conjugate pharmaceuticals must maintain a sufficient inventory to meet patients' needs. Lyophilizing material into dried 'cakes' helps maintain the supply by reducing degradation. However, to leverage the benefits of lyophilization, manufacturers must demonstrate that the freeze-drying process itself preserves product quality and uniformity. In particular, they must show that the process produces material that consistently meets residual moisture specifications. Validation of the lyophilization process requires analysis of moisture data consistent with regulatory guidance and good statistical practice. This talk will describe challenges, both practical and regulatory, that arise in this context, including limited replication at the highest level, destructive testing, and hard-to-characterize variability. It will be shown how a finite-sample variant of Chebyshev's inequality [1] was used to handle one such challenge, and how another was addressed with an analytic approximation from the engineering literature [2]. Ref: [1] J Saw, et al, Amer. Statistician, 38:2, 130-132 (1984) [2] L Fenton, IRE Transact Comm Systems, 8: 1, 57-67 (1960)