Model-based integrated optimization and evaluation of a multi-step ion exchange chromatography

Using artificial neural networks to accelerate flowsheet optimization for downstream process development

An optimal purification process for biopharmaceutical products is important to meet strict safety regulations, and for economic benefits. To find the global optimum, it is desirable to screen the overall design space. Advanced model-based approaches enable to screen a broad range of the design-space, in contrast to traditional statistical or heuristic-based approaches. Though, chromatographic mechanistic modeling (MM), one of the advanced model-based approaches, can be speed-limiting for flowsheet optimization, which evaluates every purification possibility (e.g., type and order of purification techniques, and their operating conditions). Therefore, we propose to use artificial neural networks (ANNs) during global optimization to select the most optimal flowsheets. So, the number of flowsheets for final local optimization is reduced and consequently the overall optimization time. Employing ANNs during global optimization proved to reduce the number of flowsheets from 15 to only 3. From these three, one flowsheet was optimized locally and similar final results were found when using the global outcome of either the ANN or MM as starting condition. Moreover, the overall flowsheet optimization time was reduced by 50% when using ANNs during global optimization. This approach accelerates the early purification process design; moreover, it is generic, flexible, and regardless of sample material's type.

A comprehensive method for modeling and simulating ion exchange chromatography of complex mixtures

In recent years, many biological-based products have been developed, representing a significant fraction of income in the pharmaceutical market. Ion exchange chromatography is an important downstream step for the purification of target recombinant proteins present in clarified cell extracts, together with many other unknown impurities. This work develops a robust approach to model and simulate the purification of untagged heterologous proteins, so that the improved conditions to carry out an ion exchange chromatography are identified in a rational basis prior to the real purification run itself. Purification of the pneumococcal surface protein A (PspA4Pro) was used as a case study. This protein is produced by recombinant Escherichia coli and is a candidate for the manufacture of improved pneumococcal vaccines. The developed method combined experimental and computational procedures. Different anion exchange operating conditions were mapped in order to gather a broad range of representative experimental data. The equilibrium dispersive and the steric mass action equations were used to model and simulate the process. A training strategy to fit the model and separately describe the elution profiles of PspA4Pro and other proteins of the cell extract was applied. Based on the simulation results, a reduced ionic strength was applied for PspA4Pro elution, leading to increases of 14.9% and 11.5% for PspA4Pro recovery and purity, respectively, compared to the original elution profile. These results showed the potential of this method, which could be further applied to improve the performance of ion exchange chromatography in the purification of other target proteins under real process conditions.

Isotherm model discrimination for multimodal chromatography using mechanistic models derived from high-throughput batch isotherm data

In this work, we have examined an array of isotherm formalisms and characterized them based on their relative complexities and predictive abilities with multimodal chromatography. The set of isotherm models studied were all based on the stoichiometric displacement framework, with considerations for electrostatic interactions, hydrophobic interactions, and thermodynamic activities. Isotherm parameters for each model were first determined through twenty repeated fits to a set of mAb - Capto MMC batch isotherm data spanning a range of loading, ionic strength, and pH as well as a set of mAb - Capto Adhere batch data at constant pH. The batch isotherm data were used in two ways-spanning the full range of loading or consisting of only the high concentration data points. Predictive ability was defined through the model's capacity to capture prominent changes in salt gradient elution behavior with respect to pH for Capto MMC or unique elution patterns and yield losses with respect to gradient slope for Capto Adhere. In both cases, model performance was quantified using a scoring metric based on agreement in peak characteristics for column predictions and accuracy of fit for the batch data. These scores were evaluated for all twenty isotherm fits and their corresponding column predictions, thereby producing a statistical distribution of model performances. Model complexity (number of isotherm parameters) was then considered through use of the Akaike information criterion (AIC) calculated from the score distributions. While model performance for Capto MMC benefitted substantially from removal of low protein concentration data, this was not the case for Capto Adhere; this difference was likely due to the qualitatively different shapes of the isotherms between the two resins. Surprisingly, the top-performing (high accuracy with minimal number of parameters) isotherm model was the same for both resins. The extended steric mass action (SMA) isotherm (containing both protein-salt and protein-protein activity terms) accurately captured both the pH-dependent elution behavior for Capto MMC as well as loss in protein recovery with increasing gradient slope for Capto Adhere. In addition, this isotherm model achieved the highest median score in both resin systems, despite it lacking any explicit hydrophobic stoichiometric terms. The more complex isotherm models, which explicitly accounted for both electrostatic and hydrophobic interaction stoichiometries, were ill-suited for Capto MMC and had lower AIC model likelihoods for Capto Adhere due to their increased complexity. Interestingly, the ability of the extended SMA isotherm to predict the Capto Adhere results was largely due to the protein-salt activity coefficient, as determined via isotherm parameter sensitivity analyses. Further, parametric studies on this parameter demonstrated that it had a major impact on both binding affinity and elution behavior, therein fully capturing the impact of hydrophobic interactions. In summary, we were able to determine the isotherm formalisms most capable of consistently predicting a wide range of column behavior for both a multimodal cation-exchange and multimodal anion-exchange resin with high accuracy, while containing a minimized set of model parameters.

The use of predictive models to develop chromatography-based purification processes

Chromatography is the workhorse of biopharmaceutical downstream processing because it can selectively enrich a target product while removing impurities from complex feed streams. This is achieved by exploiting differences in molecular properties, such as size, charge and hydrophobicity (alone or in different combinations). Accordingly, many parameters must be tested during process development in order to maximize product purity and recovery, including resin and ligand types, conductivity, pH, gradient profiles, and the sequence of separation operations. The number of possible experimental conditions quickly becomes unmanageable. Although the range of suitable conditions can be narrowed based on experience, the time and cost of the work remain high even when using high-throughput laboratory automation. In contrast, chromatography modeling using inexpensive, parallelized computer hardware can provide expert knowledge, predicting conditions that achieve high purity and efficient recovery. The prediction of suitable conditions in silico reduces the number of empirical tests required and provides in-depth process understanding, which is recommended by regulatory authorities. In this article, we discuss the benefits and specific challenges of chromatography modeling. We describe the experimental characterization of chromatography devices and settings prior to modeling, such as the determination of column porosity. We also consider the challenges that must be overcome when models are set up and calibrated, including the cross-validation and verification of data-driven and hybrid (combined data-driven and mechanistic) models. This review will therefore support researchers intending to establish a chromatography modeling workflow in their laboratory.

Automation of Modeling and Calibration of Integrated Preparative Protein Chromatography Systems

With the increasing global demand for precise and efficient pharmaceuticals and the biopharma industry moving towards Industry 4.0, the need for advanced process integration, automation, and modeling has increased as well. In this work, a method for automatic modeling and calibration of an integrated preparative chromatographic system for pharmaceutical development and production is presented. Based on a user-defined system description, a system model was automatically generated and then calibrated using a sequence of experiments. The system description and model was implemented in the Python-based preparative chromatography control software Orbit.

Smart Process Development: Application of machine-learning and integrated process modeling for inclusion body purification processes

The development of a biopharmaceutical production process usually occurs sequentially, and tedious optimization of each individual unit operation is very time consuming. Here, the conditions established as optimal for one step serve as input for the following step. Yet, this strategy does not consider potential interactions between a priori distant process steps and therefore cannot guarantee for optimal overall process performance. To overcome these limitations, we established a SMART approach to develop and utilize integrated process models using machine learning techniques and genetic algorithms. We evaluated the application of the data-driven models to explore potential efficiency increases and compared them to a conventional development approach for one of our development products. First, we developed a data-driven integrated process model using Gradient Boosting Machines and Gaussian Processes as machine learning techniques and a genetic algorithm as recommendation engine for two downstream unit operations, namely solubilization and refolding. Through projection of the results into our large-scale facility, we predicted a two-fold increase in productivity. Second, we extended the model to a three-step model by including the capture chromatography. Here, depending on the selected baseline-process chosen for comparison, we obtained between 50 and 100% increase in productivity. These data show the successful application of machine learning techniques and optimization algorithms for downstream process development. Finally, our results highlight the importance of considering integrated process models for the whole process chain, including all unit operations.

Microbial technologies for biotherapeutics production: Key tools for advanced biopharmaceutical process development and control

Current trends in the biopharmaceutical market such as the diversification of therapies as well as the increasing time-to-market pressure will trigger the rethinking of bioprocess development and production approaches. Thereby, the importance of development time and manufacturing costs will increase, especially for microbial production. In the present review, we investigate three technological approaches which, to our opinion, will play a key role in the future of biopharmaceutical production. The first cornerstone of process development is the generation and effective utilization of platform knowledge. Building processes on well understood microbial and technological platforms allows to accelerate early-stage bioprocess development and to better condense this knowledge into multi-purpose technologies and applicable mathematical models. Second, the application of verified scale down systems and in silico models for process design and characterization will reduce the required number of large scale batches before dossier submission. Third, the broader availability of mathematical process models and the improvement of process analytical technologies will increase the applicability and acceptance of advanced control and process automation in the manufacturing scale. This will reduce process failure rates and subsequently cost of goods. Along these three aspects we give an overview of recently developed key tools and their potential integration into bioprocess development strategies.

In silico process characterization for biopharmaceutical development following the quality by design concept

With the quality by design (QbD) initiative, regulatory authorities demand a consistent drug quality originating from a well-understood manufacturing process. This study demonstrates the application of a previously published mechanistic chromatography model to the in silico process characterization (PCS) of a monoclonal antibody polishing step. The proposed modeling workflow covered the main tasks of traditional PCS studies following the QbD principles, including criticality assessment of 11 process parameters and establishment of their proven acceptable ranges of operation. Analyzing effects of multi-variate sampling of process parameters on the purification outcome allowed identification of the edge-of-failure. Experimental validation of in silico results demanded approximately 75% less experiments compared to a purely wet-lab based PCS study. Stochastic simulation, considering the measured variances of process parameters and loading material composition, was used to estimate the capability of the process to meet the acceptance criteria for critical quality attributes and key performance indicators. The proposed workflow enables the implementation of digital process twins as QbD tool for improved development of biopharmaceutical manufacturing processes.

Analysis of complex protein elution behavior in preparative ion exchange processes using a colloidal particle adsorption model

A fundamental understanding of the protein retention mechanism in preparative ion exchange (IEX) chromatography columns is essential for a model-based process development approach. For the past three decades, the mechanistic description of protein retention has been based predominantly on the steric mass action (SMA) model. In recent years, however, retention profiles of proteins have been reported more frequently for preparative processes that are not consistent with the mechanistic understanding relying on the SMA model. In this work, complex elution behavior of proteins in preparative IEX processes is analyzed using a colloidal particle adsorption (CPA) model. The CPA model is found to be capable of reproducing elution profiles that cannot be described by the traditional SMA model. According to the CPA model, the reported complex behavior can be ascribed to a strong compression and concentration of the elution front in the lower unsaturated part of the chromatography column. As the unsaturated part of the column decreases with increasing protein load density, exceeding a critical load density can lead to the formation of a shoulder in the peak front. The general applicability of the model in describing preparative IEX processes is demonstrated using several industrial case studies including multiple monoclonal antibodies on different IEX adsorber systems. In this context, the work covers both salt controlled and pH-controlled protein elution.

Protein adsorption on ion exchange adsorbers: A comparison of a stoichiometric and non-stoichiometric modeling approach

For mechanistic modeling of ion exchange (IEX) processes, a profound understanding of the adsorption mechanism is important. While the description of protein adsorption in IEX processes has been dominated by stoichiometric models like the steric mass action (SMA) model, discrepancies between experimental data and model results suggest that the conceptually simple stoichiometric description of protein adsorption provides not always an accurate representation of nonlinear adsorption behavior. In this work an alternative colloidal particle adsorption (CPA) model is introduced. Based on the colloidal nature of proteins, the CPA model provides a non-stoichiometric description of electrostatic interactions within IEX columns. Steric hindrance at the adsorber surface is considered by hard-body interactions between proteins using the scaled-particle theory. The model's capability of describing nonlinear protein adsorption is demonstrated by simulating adsorption isotherms of a monoclonal antibody (mAb) over a wide range of ionic strength and pH. A comparison of the CPA model with the SMA model shows comparable model results in the linear adsorption range, but significant differences in the nonlinear adsorption range due to the different mechanistic interpretation of steric hindrance in both models. The results suggest that nonlinear adsorption effects can be overestimated by the stoichiometric formalism of the SMA model and are generally better reproduced by the CPA model.

Modeling the impact of amino acid substitution in a monoclonal antibody on cation exchange chromatography

A vital part of biopharmaceutical research is decision making around which lead candidate should be progressed in early-phase development. When multiple antibody candidates show similar biological activity, developability aspects are taken into account to ease the challenges of manufacturing the potential drug candidate. While current strategies for developability assessment mainly focus on drug product stability, only limited information is available on how antibody candidates with minimal differences in their primary structure behave during downstream processing. With increasing time-to-market pressure and an abundance of monoclonal antibodies (mAbs) in development pipelines, developability assessments should also consider the ability of mAbs to integrate into the downstream platform. This study investigates the influence of amino acid substitutions in the complementarity-determining region (CDR) of a full-length IgG1 mAb on the elution behavior in preparative cation exchange chromatography. Single amino acid substitutions within the investigated mAb resulted in an additional positive charge in the light chain (L) and heavy chain (H) CDR, respectively. The mAb variants showed an increased retention volume in linear gradient elution compared with the wild-type antibody. Furthermore, the substitution of tryptophan with lysine in the H-CDR3 increased charge heterogeneity of the product. A multiscale in silico analysis, consisting of homology modeling, protein surface analysis, and mechanistic chromatography modeling increased understanding of the adsorption mechanism. The results reveal the potential effects of lead optimization during antibody drug discovery on downstream processing.

Model-based selection of the degree of cross-linking of cation exchanger resins for an optimised separation of monosaccharides

One of the main steps in designing preparative chromatographic separation units is the selection of a well-performing adsorbent. This is often based on expert knowledge or based on case studies of preselected adsorbents. Therefore, the selection is usually limited in terms of an optimised choice. In this contribution, a model-based optimisation of the selection of an adsorbent on the basis of correlations between structural adsorbent properties with model parameters of a transport dispersive model is proposed. Model parameters of glucose and xylose for five cation exchanger resins with varying degree of cross-linking are experimentally determined in a sequential approach. Void fractions and particle porosities were determined by pulse experiments with different tracers. Single-component isotherms were determined threefold via breakthrough curves with concentrations of up to 250 g l^-1 at 60 °C. Mass transfer coefficients were determined by batch experiments. Correlations between the degree of cross-linking of the resins and the Henry coefficients as well as the mass transfer coefficients were derived and applied in an optimisation case study. Based on the derived mathematical formula, the process performance of experimentally not investigated resins were predicted. Further, the selection of a resin for a preparative monosaccharide separation was included into optimisation algorithms.

Hybrid optimization of preparative chromatography for a ternary monoclonal antibody mixture

In the purification of monoclonal antibodies, ion-exchange chromatography is typically used among the polishing steps to reduce the amount of product-related impurities such as aggregates and fragments, whilst simultaneously reducing HCP, residual Protein A and potential toxins and viruses. When the product-related impurities are difficult to separate from the products, the optimization of these chromatographic steps can be complex and laborious. In this paper, we optimize the polishing chromatography of a monoclonal antibody from a challenging ternary feed mixture by introducing a hybrid approach of the simplex method and a form of local optimization. To maximize the productivity of this preparative bind-and-elute cation-exchange chromatography, wide ranges of the three critical operational parameters-column loading, the initial salt concentration, and gradient slope-had to be considered. The hybrid optimization approach is shown to be extremely effective in dealing with this complex separation that was subject to multiple constraints based on yield, purity, and product breakthrough. Furthermore, it enabled the generation of a large knowledge space that was subsequently used to study the sensitivity of the objective function. Increased design space understanding was gained through the application of Monte Carlo simulations. Hence, this work proposes a powerful hybrid optimization method, applied to an industrially relevant process development challenge. The properties of this approach and the results and insights gained, make it perfectly suited for the rapid development of biotechnological unit operations during early-stage bioprocess development.

Downstream Processing Technologies/Capturing and Final Purification : Opportunities for Innovation, Change, and Improvement. A Review of Downstream Processing Developments in Protein Purification

Increased pressure on upstream processes to maximize productivity has been crowned with great success, although at the cost of shifting the bottleneck to purification. As drivers were economical, focus is on now on debottlenecking downstream processes as the main drivers of high manufacturing cost. Devising a holistically efficient and economical process remains a key challenge. Traditional and emerging protein purification strategies with particular emphasis on methodologies implemented for the production of recombinant proteins of biopharmaceutical importance are reviewed. The breadth of innovation is addressed, as well as the challenges the industry faces today, with an eye to remaining impartial, fair, and balanced. In addition, the scope encompasses both chromatographic and non-chromatographic separations directed at the purification of proteins, with a strong emphasis on antibodies. Complete solutions such as integrated USP/DSP strategies (i.e., continuous processing) are discussed as well as gains in data quantity and quality arising from automation and high-throughput screening (HTS). Best practices and advantages through design of experiments (DOE) to access a complex design space such as multi-modal chromatography are reviewed with an outlook on potential future trends. A discussion of single-use technology, its impact and opportunities for further growth, and the exciting developments in modeling and simulation of DSP rounds out the overview. Lastly, emerging trends such as 3D printing and nanotechnology are covered. Graphical Abstract Workflow of high-throughput screening, design of experiments, and high-throughput analytics to understand design space and design space boundaries quickly. (Reproduced with permission from Gregory Barker, Process Development, Bristol-Myers Squibb).

Prediction uncertainty assessment of chromatography models using Bayesian inference

Mechanistic modeling of chromatography has been around in academia for decades and has gained increased support in pharmaceutical companies in recent years. Despite the large number of published successful applications, process development in the pharmaceutical industry today still does not fully benefit from a systematic mechanistic model-based approach. The hesitation on the part of industry to systematically apply mechanistic models can often be attributed to the absence of a general approach for determining if a model is qualified to support decision making in process development. In this work a Bayesian framework for the calibration and quality assessment of mechanistic chromatography models is introduced. Bayesian Markov Chain Monte Carlo is used to assess parameter uncertainty by generating samples from the parameter posterior distribution. Once the parameter posterior distribution has been estimated, it can be used to propagate the parameter uncertainty to model predictions, allowing a prediction-based uncertainty assessment of the model. The benefit of this uncertainty assessment is demonstrated using the example of a mechanistic model describing the separation of an antibody from its impurities on a strong cation exchanger. The mechanistic model was calibrated at moderate column load density and used to make extrapolations at high load conditions. Using the Bayesian framework, it could be shown that despite significant parameter uncertainty, the model can extrapolate beyond observed process conditions with high accuracy and is qualified to support process development.

Water on hydrophobic surfaces: Mechanistic modeling of hydrophobic interaction chromatography

Mechanistic models are successfully used for protein purification process development as shown for ion-exchange column chromatography (IEX). Modeling and simulation of hydrophobic interaction chromatography (HIC) in the column mode has been seldom reported. As a combination of these two techniques is often encountered in biopharmaceutical purification steps, accurate modeling of protein adsorption in HIC is a core issue for applying holistic model-based process development, especially in the light of the Quality by Design (QbD) approach. In this work, a new mechanistic isotherm model for HIC is derived by consideration of an equilibrium between well-ordered water molecules and bulk-like ordered water molecules on the hydrophobic surfaces of protein and ligand. The model's capability of describing column chromatography experiments is demonstrated with glucose oxidase, bovine serum albumin (BSA), and lysozyme on Capto™ Phenyl (high sub) as model system. After model calibration from chromatograms of bind-and-elute experiments, results were validated with batch isotherms and prediction of further gradient elution chromatograms.

Downstream process development strategies for effective bioprocesses: Trends, progress, and combinatorial approaches

The biopharmaceutical industry is at a turning point moving toward a more customized and patient-oriented medicine (precision medicine). Straightforward routines such as the antibody platform process are extended to production processes for a new portfolio of molecules. As a consequence, individual and tailored productions require generic approaches for a fast and dedicated purification process development. In this article, different effective strategies in biopharmaceutical purification process development are reviewed that can analogously be used for the new generation of antibodies. Conventional approaches based on heuristics and high-throughput process development are discussed and compared to modern technologies such as multivariate calibration and mechanistic modeling tools. Such approaches constitute a good foundation for fast and effective process development for new products and processes, but their full potential becomes obvious in a correlated combination. Thus, different combinatorial approaches are presented, which might become future directions in the biopharmaceutical industry.

Application of spectral deconvolution and inverse mechanistic modelling as a tool for root cause investigation in protein chromatography

In chromatographic protein purification, process variations, aging of columns, or processing errors can lead to deviations of the expected elution behavior of product and contaminants and can result in a decreased pool purity or yield. A different elution behavior of all or several involved species leads to a deviating chromatogram. The causes for deviations are however hard to identify by visual inspection and complicate the correction of a problem in the next cycle or batch. To overcome this issue, a tool for root cause investigation in protein chromatography was developed. The tool combines a spectral deconvolution with inverse mechanistic modelling. Mid-UV spectral data and Partial Least Squares Regression were first applied to deconvolute peaks to obtain the individual elution profiles of co-eluting proteins. The individual elution profiles were subsequently used to identify errors in process parameters by curve fitting to a mechanistic chromatography model. The functionality of the tool for root cause investigation was successfully demonstrated in a model protein study with lysozyme, cytochrome c, and ribonuclease A. Deviating chromatograms were generated by deliberately caused errors in the process parameters flow rate and sodium-ion concentration in loading and elution buffer according to a design of experiments. The actual values of the three process parameters and, thus, the causes of the deviations were estimated with errors of less than 4.4%. Consequently, the established tool for root cause investigation is a valuable approach to rapidly identify process variations, aging of columns, or processing errors. This might help to minimize batch rejections or contribute to an increased productivity.

Modeling and simulation of anion-exchange membrane chromatography for purification of Sf9 insect cell-derived virus-like particles

Recombinant protein-based virus-like particles (VLPs) are steadily gaining in importance as innovative vaccines against cancer and infectious diseases. Multiple VLPs are currently evaluated in clinical phases requiring a straightforward and rational process design. To date, there is no generic platform process available for the purification of VLPs. In order to accelerate and simplify VLP downstream processing, there is a demand for novel development approaches, technologies, and purification tools. Membrane adsorbers have been identified as promising stationary phases for the processing of bionanoparticles due to their large pore sizes. In this work, we present the potential of two strategies for designing VLP processes following the basic tenet of 'quality by design': High-throughput experimentation and process modeling of an anion-exchange membrane capture step. Automated membrane screenings allowed the identification of optimal VLP binding conditions yielding a dynamic binding capacity of 5.7 mg/mL for human B19 parvovirus-like particles derived from Spodoptera frugiperda Sf9 insect cells. A mechanistic approach was implemented for radial ion-exchange membrane chromatography using the lumped-rate model and stoichiometric displacement model for the in silico optimization of a VLP capture step. For the first time, process modeling enabled the in silico design of a selective, robust and scalable process with minimal experimental effort for a complex VLP feedstock. The optimized anion-exchange membrane chromatography process resulted in a protein purity of 81.5%, a DNA clearance of 99.2%, and a VLP recovery of 59%.

Next generation vaccines and vectors: Designing downstream processes for recombinant protein-based virus-like particles

In recent years, the development of novel recombinant virus-like particles (VLPs) has been generating new perspectives for the prevention of untreated and arising infectious diseases. However, cost-reduction and acceleration of manufacturing processes for VLP-based vaccines or vectors are key challenges for the global health system. In particular, the design of rapid and cost-efficient purification processes is a critical bottleneck. In this review, we describe and evaluate new concepts, development strategies and unit operations for the downstream processing of VLPs. A special focus is placed on purity requirements and current trends, as well as chances and limitations of novel technologies. The discussed methods and case studies demonstrate the advances and remaining challenges in both rational process development and purification tools for large biomolecules. The potential of a new era of VLP-based products is highlighted by the progress of various VLPs in clinical phases.