Pharmaceutical products have been with us for many decades. Relatively new, however, are biopharmaceuticals. These are medical drugs produced using living organisms such as microbial cells, mammalian cell lines and plant cell cultures. Each cell is a living factory transforming nutrients into a protein drug product via their metabolic pathways. The bioreactors in which this production takes place are inordinately complex environments, with a nonlinear, dynamic mix of billions of cells and nutrients, vulnerable to temperature change, pH, inhomogeneity and so on.

There is significant interest in improving process control capabilities in bioreactors, one of the main drivers being pressure from the regulatory authorities. The US Food and Drug Administration’s Process Analytical Technology (PAT) initiative is an example of this.

Utilizing ABB’s xPAT product, ABB is collaborating with Irish universities and leading biopharmaceutical players in an initiative funded by Enterprise Ireland to construct models to set up and evaluate the benefits of PAT-enabled, model-based control of a fed-batch mammalian cell culture.

Mammalian cells, particularly Chinese hamster ovary (CHO) cells, and bacterial systems, such as Escherichia coli (E. coli), produce the bulk of the products on the market but alternative systems such as yeast and plant cells are also used.

^{Chinese hamster ovary cell lines are the workhorses of the mammalian cell biopharmaceutical industry. They are used to manufacture a number of licensed therapeutic proteins such as erythropoietin (EPO), CD20, tumor necrosis factor alpha and HER2. They are robust cells, which can easily be adapted to meet the requirements of large-scale protein production. They are easily adapted to grow in suspension at high viable cell densities, which simplifies large-scale culture in stirred tank bioreactors. They are also capable of high levels of protein expression. Also, their DNA is easily modified in order to have the cell line produce the protein product of interest.}

The vast majority of current industrial bioprocesses are operated in batch or fed-batch mode. Fed-batch mode refers to a process whereby a concentrated nutrient feed solution is added either continuously or periodically over the course of a batch and no product is harvested before the endpoint is reached.

The improvement in productivity achieved in fed-batch processes is due mainly to an increase in the integral of viable cells and a resultant increase in volumetric productivity. Fed-batch mode is popular because of reliability, ease of scale-up, significant increase in production levels and ease of process characterization and validation.

In contrast to the chemical and traditional pharmaceutical sectors, process control for bioprocesses is in its infancy due, in part, to the challenges associated with bioreactor control: complex growth media, inadequate measurement of relevant process parameters, the limited and noisy nature of experimental data and difficulties inherent in controlling bioprocesses, which are dynamic, complex and nonlinear.

Process control of bioreactors seeks to influence the individual complex intracellular reactions of billions of cells by controlling their extracellular environment. Traditionally, parameters such as temperature, pH and dissolved oxygen (DO) are measured using in-situ probes and are controlled using PID loops to adjust gas or alkali flows. Control of nutrient levels is still usually manual.

Generally, bolus feeds are introduced at 24-hour intervals based on offline analysis of daily process samples and a priori process knowledge.

There is significant interest in improving the process control capabilities of bioreactors. One of the main drivers for this is pressure from the regulatory authorities. The US Food and Drug Administration (FDA) launched the Process Analytical Technology (PAT) initiative. It defines PAT as a mechanism to design, analyze, and control pharmaceutical manufacturing processes through the measurement of critical process parameters (CPP), which affect critical quality attributes (CQA). The objective is to understand the processes by defining their CPPs, and to monitor them in a timely manner, preferably in-line or online, thus reducing final testing requirements, reject rates and instances of over-processing while enhancing consistency and product quality.

As there is variability both in raw materials and in operation of equipment, a static batch process will produce a variable product. By increasing process understanding and control potential, the PAT initiative aims to design quality into the process, rather than relying on testing the CQAs of the final product, by facilitating a dynamic manufacturing process that can compensate for these underlying variations. PAT has increasingly gained worldwide acceptance as a proven method of ensuring product safety and quality by many industry experts.

Associated benefits of greater process understanding, apart from improved product quality, include faster process optimization and speed to market, improved product titers, decreased process variability, shorter cycle times and reduced waste.

Facilitating the PAT initiative is the increased availability of online measurements. In order to effectively control a parameter, it must first be measured. Spectroscopic techniques such as near infrared (NIR), mid-IR and Raman are online monitoring tools for nutrient and metabolite concentrations. Automated multifunction analyzers such as the Nova or YSI systems use a combination of enzymatic, amperometric, potentiometric and coulter counter or CCD camera analyzers to perform simultaneous quantification of nutrient and metabolite concentrations, cell density and viability, dissolved oxygen and carbon dioxide, and pH at-line.

Automated at-line analysis must also be able to withdraw an automated, sterile sample from the bioreactor. There are a number of systems capable of delivering this requirement currently being developed and brought to market. Closedloop feedback control based on at-line samples is possible if the sample interval is less than 25 percent of the dominant system response time.

A project to investigate the potential of the bio-application of PAT has been funded by Enterprise Ireland and led by Professor Brian Glennon. The School of Chemical and Bioprocess Engineering at University College Dublin, research groups at UCD, Dublin City University, the Tyndall Institute at University College Cork and the National Institute for Bioprocessing Research and Training (NIBRT) have been collaborating with ABB to set up and evaluate the benefits of PAT-enabled, model-based control of a fedbatch mammalian cell culture.

A number of Irish-based multinationals such as Pfizer, Eli Lilly, Jannsen Biologics and Merck as well as a group of indigenous SMEs including BioUETIKON, Technopath and BioImages, among others, form an industrial advisory board that meets quarterly to comment on and guide the research strands. The project is based on an ABB Extended Automation System 800xA control system and an xPAT system comprised of FTSW800 analyzer controllers and data management system.

An industry-specific application built on the System 800xA infrastructure, xPAT (eXtended PAT) is a next-generation PAT solution that harnesses the System 800xA operations and engineering environment and integration capability to provide significant improvements in the overall process and end-product quality. It provides life sciences users with a single system to access and examine online, real-time process data directly from the manufacturing operation.

The configurable Windows-based system collects data from ABB and/or third-party vendors’ analytical instruments and analyzes the data to determine the actual condition of the process. It then passes the resulting information to the ABB or third-party control system, and to other applications that support the drug manufacturing process.

ABB is providing the engineering services required to install and configure the system at UCD. A CHO cell line is being used as the model system because it is the most common industrial expression system. A number of PAT technologies both in-line and at-line – such as mid-IR and Raman spectroscopies for the monitoring of substrate and metabolite concentrations and flow cytometery, Canty imaging systems and the Beckman Coulter Vi-Cell for the determination of many cell parameters such as cell density, viability and cell cycle – have been evaluated and developed for the process under investigation.

Researchers at the Tyndall Institute and at UCD develop a sample valve assembly that is capable of taking an automated, sterile sample from the reactor to facilitate at-line analysis. A first-principle, semi-empirical model describing the biomass, substrate and metabolite trajectories is utilized in an MPC framework in order to control the feed rate of nutrients to the reactor.

Experimental work aimed to implement, optimize and evaluate the practicalities and benefits of installing a PATenabled advanced control system for a mammalian cell process. This work has shown a 15 percent increase in the maximum viable cell density achieved in the MPC-controlled fed-batch bioreactor when compared with standard bolus fed-batch bioreactor runs, a significant increase. An increase in viable cell density means that there are a greater number of production units within the bioreactor and so more of the biopharmaceutical is produced.

A control system such as the xPAT system, capable of integrating a wide variety of PAT instruments and managing the data that they produce, opens up the possibility of implementing advanced model-based control strategies. These advanced strategies can increase productivity and process robustness, as well as decrease process variation, all extremely desirable outcomes in the biopharmaceutical industry.