Multivariate Mixed-Effects Kinetic Models for Multiple Correlated Quality Attributes From Accelerated Stability Studies

Abstract

Stability study is a critical component for the submission and market authorization of a new drug or biological product. Long-term stability studies are required to establish the stability profile and shelf life of the drug product. Accelerated stability studies may provide insight into degradation pathways and help expedite the development of formulation and packaging. Accelerated stability studies are also useful to identify the stability-indicating attributes and appropriate assay methods for measuring degradation. Selection and specifications of critical quality attributes are imperative for ensuring the quality, safety, and efficacy of the drug. For individual attributes, Arrhenius equation has been used to combine accelerated stability data with long-term stability data to predict the degradation rates under long-term storage conditions. In practice, multiple stability-indicating critical attributes are available. The multiple attributes are intrinsically correlated because they are all indicators of drug quality. We propose a multivariate mixed-effects kinetic model, which can both combine the accelerated and long-term stability data using Arrhenius equation and incorporate lot-to-lot variability and between-attribute correlations with random effects. Simulation studies show that the multivariate modeling of the correlated attributes may provide more accurate estimates of degradation rates. The proposed model may provide insight on how to select the stability-indicating attributes. The stability data of three quality attributes for a therapeutic protein are analyzed for illustration.