Predicting the Long-Term Stability of Biologics with Short-Term Data.

Abstract

Understanding the long-term stability of biologics is crucial to ensure safe, effective, and cost-efficient life-saving therapeutics. Current industry and regulatory practices require arduous real-time data collection over three years; thus, reducing this bottleneck while still ensuring product quality would enhance the speed of medicine to patients. We developed a parallel-pathway kinetic model, combined with Monte Carlo simulations for prediction intervals, to predict the long-term (2+ years) stability of biotherapeutic critical quality attributes (aggregates, fragments, charge variants, purity, and potency) with short-term (3-6 months) data from intended, accelerated, and stressed temperatures. We rigorously validated the model with 18 biotherapeutic drug products, composed of IgG1 and IgG4 monoclonal antibodies, antibody-drug conjugates, dual protein coformulations, and a fusion protein, including high concentration (≥100 mg/mL) formulations, in liquid and lyophilized presentations. For each drug product, we accurately predicted the long-term trends of multiple quality attributes using just 6 months of data. Further, we demonstrated superior stability prediction via our methods compared with industry-standard linear regression methods. The robust and repeatable results of this work across an unprecedented suite of 18 biotherapeutic compounds suggest that kinetic models with Monte Carlo simulation can predict the long-term stability of biologics with short-term data.