The Use of Bayesian Hierarchical Logistic Regression in the Development of a Modular Viral Inactivation Claim

Area of Science:

• Biopharmaceutical Manufacturing
• Viral Clearance
• Statistical Modeling

Background:

• Low pH inactivation is a standard method for enveloped virus removal in biopharmaceutical production.
• Current statistical methods for viral inactivation claims often lack probabilistic characterization of log10 reduction values (LRVs).
• Existing analyses struggle to incorporate historical data and right-censored data effectively.

Purpose of the Study:

• To introduce a novel statistical approach for evaluating low pH viral inactivation.
• To enable probabilistic characterization of future experimental LRVs.
• To support the development of modular viral clearance claims for biopharmaceutical manufacturing.

Main Methods:

• Bayesian hierarchical logistic regression modeling was employed.
• The model accommodates historical data from diverse experiments.
• The method handles right-censored data, a common challenge in inactivation studies.

Main Results:

• The Bayesian approach allows for probability statements on successful viral inactivation based on process parameters.
• This facilitates the creation of modular claims for viral clearance, ensuring a critical LRV.
• The risk-based approach, combined with descriptive statistics, aids in coherent decision-making for LRV claims.

Conclusions:

• Bayesian modeling offers a more robust statistical framework for viral inactivation validation.
• This methodology enhances the reliability of modular viral clearance claims.
• The approach supports efficient and scientifically sound biopharmaceutical development.