(199c) An Integrated Modeling and Experimental Framework for Predicting the N-Linked Glycosylation of Monoclonal Antibodies

An integrated modeling and experimental framework for predicting the N-linked glycosylation of monoclonal antibodies

Ioscani Jiménez del Val and Cleo Kontoravdi

Department of Chemical Engineering and Chemical Technology, Imperial College London, South Kensington Campus, SW7 2AZ

Development and approval of pharmaceutical products is an extremely costly and time-consuming process which has resulted in unsustainable expense for healthcare providers and patients. In order to address this critical issue, regulatory agencies and industry experts have jointly encouraged implementation of the Quality by Design (QbD) paradigm for the development of all new drug products in the pipeline, including therapeutic proteins. Adoption of QbD principles in pharmaceutical product development is expected to reduce not only regulatory approval time and costs, but also to encourage innovation by building processes around the mechanistic relationships between their inputs and end product quality. Monoclonal antibodies (mAbs) are currently the highest-selling products of the biopharmaceutical industry and are expected to make up half of the top -ten selling pharmaceutical products by 2016 [1]. All approved mAbs contain N-linked carbohydrates on their crystallizable fragment (Fc), the distribution of which greatly impacts the safety and efficacy of these therapeutic glycoproteins and thus defines it as a critical quality attribute (CQA) of mAbs under the QbD scope.

Based on QbD guidelines, the purpose of this work has been to generate an integrated modeling and experimental framework to mechanistically and quantitatively tie process conditions –namely, nutrient availability- with the glycosylation-associated quality of mAbs. The integrated framework consists of a dynamic mathematical model that mechanistically couples mAb N-linked glycosylation with cellular metabolism and an experimental methodology that measures the key variables required for parameter estimation and model validation. In order to mechanistically link cellular metabolism with mAb glycosylation, the pathway for nucleotide sugar donors (NSDs) –which are metabolic products directly derived from glucose and glutamine and intrinsic co-substrates for the glycosylation reactions- has been modeled. A Monte Carlo-based global sensitivity analysis method [2] was then used both to reduce the NSD model and to determine which of its parameters are crucial for accurate representation of the system. Experimentally, batch cultures were performed with hybridoma (CRL-1606 from ATCC), GS-NS0 and GS-CHO cell lines. Typical data was collected from these batch cultures and included viable cell density, glucose, glutamine, lactate, and ammonia concentrations, as well as antibody titre. Intracellular samples were produced by quenching with ice-cold 0.9% NaCl solution and extracting with perchloric acid. The intracellular samples were then assayed for glucose and glutamine using fluorescent kits and used for NSD quantification using anion exchange HPLC. The HPLC method was based on a previously reported one [3], but optimized to resolve eight nucleotide sugars and four nucleotides in 30 minutes. Glycan analysis was performed at all time-points using MALDI-TOF mass spectrometry. The generated data was used to determine the unknown parameters of the model for all three cell lines. Subsequently, fed-batch validation experiments, where glucose and glutamine were added throughout culture, were performed in order to assess model and parameter validity. For all three cell lines tested, the results show accurate representation of the dependence between extracellular glucose and glutamine availability, intracellular NSD concentrations and mAb Fc glycosylation profile.

Our results suggest that this combined modelling and experimental framework is an effective way of mechanistically linking readily-measurable process inputs with critical quality attributes of biopharmaceuticals as underlined in the QbD methodology. Furthermore, the framework outlined here has the potential of aiding the development of novel processes for the manufacture of therapeutic glycoproteins under the Quality by Design paradigm in the future.

1. EvaluatePharma, World Preview 2016 "Beyond the Patent Cliff" (2011).

2. Kucherenko S, et al. Reliab. Eng. Syst. Safe 94 (7) (2009) 1135-1148.

3. Tomiya N, et al. Analytical Biochemistry 293(1) (2001) 129-137.