New Tools Facilitate Automated Process Control

June 1, 2020

Volume 33, Issue 6

Page Number: 10–15

Automated sampling systems and integration of PAT are overcoming some of the roadblocks to increased automation of biopharmaceutical manufacturing processes.

Advances in biopharmaceutical manufacturing, such as the shift to flexible, multi-product manufacturing; continuous upstream and downstream processing; and the potential for real-time release, are all enabled by automation and control technology. “The push for faster time to market is leading to more continuous and flexible manufacturing, which requires advanced automation and control for consistent quality,” explains Bob Lenich, director of global life sciences at Emerson. “And the strategy of real-time release requires more comprehensive embedded analytical measurements and advanced controls to ensure quality.”

Although the tools for automation have been available for some time, the process analytical technology (PAT) to provide the data needed to use automated control for biopharmaceutical processes has more recently come into use. Bioreactor manufacturers, such as Pall Life Sciences, Cytiva (formerly GE Healthcare), and Sartorius, are incorporating sensors, PAT, and automation into their manufacturing equipment (1). 

PAT and sensor integration

The increasing availability of PAT tools is promoting greater use of continuous in-line or at-line monitoring, which allows automated control strategies to be implemented, says Melisa Carpio, global technology consultant for cell culture technologies at Sartorius Stedim Biotech. 

Sartorius PAT tools include the new BioPAT Spectro, which uses Raman spectroscopy for process control and monitoring of single-use cell-culture bioreactors, including both the lab-scale Ambr automated micro- and mini-reactors and the company’s Biostat STR single-use production-scale bioreactors. BioPAT Spectro is designed as an open platform and is currently compatible with Raman spectrometers from Kaiser Optical Systems and Tornado Spectral Systems. At the lab scale, data collected from Raman spectroscopy are contextualized with the collected process data and used as a quality-by-design (QbD) tool for building a model, using Sartorius’ Umetrics SIMCA multivariate data analysis (MVDA) software, says Svea Cheeseman, product manager of PAT in cell-culture technologies at Sartorius. 

“Raman spectroscopy is not a direct measurement that comes pre-calibrated and gives you a pre-defined set of parameters. One needs to perform a model-building exercise for every analyte,” Cheeseman explains. “The resulting model that predicts the analyte concentration from the Raman spectra can then be loaded into the Ambr system and used for process monitoring and control in high-throughput process development. The model can also be scaled to the Biostat STR production scale single-use bioreactors, due to identical optical designs of the Raman interfaces.” 

An important part of PAT is integration of the sensors into the process. The optimal design for a single-use bioreactor is an integrated, presterilized, single-use sensor, and these are available for dissolved oxygen, pH, capacitance, and conductivity, says Cheeseman. The BioPAT Spectro Raman sensor is integrated as a single-use, 3D-printed port welded to the bioreactor bag containing a flow cell with two windows. A fiber optic probe is connected to the outside of the port at one end and to the Raman spectrometer at the other to collect Raman spectra. 

For sensors that aren’t available as single-use versions, sterile connectors are used to integrate conventional, multi-use sensors. A multi-use glucose sensor, for example, measures glucose concentration and can be used for feed control with a feedback control loop that activates a pump to add glucose when needed. 

Sartorius’ BioStat STR Generation 3 single-use bioreactor and the company’s BioBrain automation platform were launched in March 2020 (2). A new foam sensor on this bioreactor is a single-use adhesive patch that is attached to the outside of the STR bag and is used to prevent potential batch losses due to filter clogging and pressure build-up due to excess foaming. “There is both a high foam patch with alarming and interlock functionalities as well as a control foam patch with alarming and antifoam agent feed control,” says Cheeseman.

One of the challenges today is integration of single-use instruments into control systems, notes Artur Miguel Arsénio, head of project management, PAT, automation and software at Sartorius Stedim Biotech. He says that vendors are working towards standardized “plug-and-play” communication interfaces, recommended by organizations such as BioPhorum. Standardized, pretested “modular packages” will allow easier and more flexible deployment and changes, suggests Arsénio. 

 

Sampling systems

Automated bioreactor monitoring is one of the keys to collecting the data needed for process understanding and process control, but this is not a simple task. Tools have been developed, however, to automatically collect, transport, and analyze samples. Lonza’s Modular Automated Sampling Technology (MAST) system takes samples from bioreactors while maintaining sterility and reliably moves the samples to analyzers for automated analysis. The fully automated system “allows real-time monitoring and control that was never possible before,” says Clint Pepper, director of MAST products at Lonza. He explains that an event-based sampling approach allows process control systems to “collect data at precise points in the manufacturing process such as inoculation, induction, or when a critical parameter unexpectedly goes out of range.” While some in the biopharma may be skeptical of autosamplers, Pepper says the MAST system was designed to overcome issues of sterility failures and lack of reliability that limited earlier autosamplers. The system is currently in use in laboratory environments, where it enables dynamic experimental designs to speed the development process, Pepper says. However, it was “designed from the ground up to be used in the [good manufacturing practice] GMP environment” and the company expects to launch a 21 Code of Federal Regulations Part 11-compliant version of the software in 2021. Lonza also plans to validate the system for specific quality control testing applications. 

Fialab’s inline sequential injection analysis (SIA) instrumentation automates the ultra-high-performance liquid chromatography (UHPLC) or mass spectrometry (MS) pretreatment of samples for labor-intensive, complex analysis, such as peptide mapping or glycan profiling. “The SIA system does-in a very small box-what a scientist in a lab would do: the SIA draws in reagents/sample, mixes, and holds as needed to prepare the samples for LC analysis. The sample never leaves the capillary tubing,” explains Daniel Hasle, account manager for pharmaceutical PAT at FIAlab.

Fialab’s sample preparation system is part of an experimental setup at Rutgers University in the Department of Chemical and Biochemical Engineering in an FDA-funded project that is using a fully autonomous system developed by Rutgers for sampling and sample preparation for an on-line LC–MS system to monitor the monoclonal antibody (mAb) glycolization profile. “Current process performance understanding is hindered by lack of robust online tools and lengthy assays, which results in difficulty running processes in a real-time fashion, which leads to inconsistent product quality,” said the researchers in a poster presentation (3). Experiments showed that the on-line LC/MS system provides the same sensitivity as traditional methods. In addition, the researchers used an in-situ Raman probe with a Kaiser Raman spectrometer for real-time metabolite monitoring to develop a calibration model for real-time measurement of glucose and lactate concentrations, and they are working on models for other metabolites. The model will be used to develop a feedback process control loop for the bioreactor (3).

Automation and digital twins

Although biopharmaceutical manufacturing overall lags behind other types of manufacturing processes in its digital maturity, early adopters of Industry 4.0 technologies are moving to automation and connectivity. 

“[Automation] is a priority area where our customers see opportunity to remove some of the bottlenecks they have today and get more out of their existing facilities,” says Darren Verlenden, head of bioprocessing at MilliporeSigma.

Although automation solutions are available and accepted by regulatory bodies, they are often adopted from other industries, resulting in a high level of customization, says Verlenden. He notes that modular, configurable solutions developed specifically for the bioprocessing industry can help to reduce implementation time and cost. As an example, he points to MilliporeSigma’s launch at the end of April 2020 of the Bio4C Software Suite that combines the company’s existing processing technologies with software, automation, and analytics as a digital ecosystem that is part of the company’s BioContinuum Platform. Verlenden explains that, in regard to digital maturity, these systems help break down digital silos and move toward connected plants with full process control. 

The next step of the digital maturity model (4) is an adaptive plant with predictive process control. Achieving this level will require collaboration, says Verlenden. “Suppliers will need to develop the right technologies to achieve this vision in partnership with the industry and implement them in a way that meets the regulatory requirements of bioprocessing,” he explains. 

 

Digital twins, which are process models used for simulating a process and predicting what will occur in different scenarios, are key to adaptive plants. Sanofi received the International Society for Pharmaceutical Engineering (ISPE) Facility of the Year Award in the Facility of the Future category in 2020 for its digitally enabled continuous biomanufacturing facility in Framingham, MA, which opened in October 2019 (5). The facility has digitally connected equipment, and simulations enable flexible production (6) because digital twins can be used to evaluate process changes. As another example, Cytiva has been using digital twins to predict and optimize cell-culture processes (7).

Process Systems Enterprise’s (PSE) digital twin software for biopharmaceutical processes is being used by early adopters for both digital design (e.g., increased R&D efficiency, tech transfer, scale-up, investigating batch to continuous conversion) and digital operations (e.g., soft sensing, advanced process control, control system tuning), says Edd Close, principal consultant at PSE. 

Setting up a digital twin for a particular process involves training the model using bioprocess data, explains Klaus Mauch, CEO of Insilico Biotechnology. In addition, “Machine learning is used to capture the dynamics of the cell-culture process and also enables integration of factors that cannot be represented mechanistically, for instance because their impact is not entirely known,” explains Mauch. Insilico Biotechnology is collaborating with the Laboratory of Systems Theory and Automatic Control at the Institute for Automation Engineering of the Otto von Guericke University in Magdeburg, Germany on model predictive control of the production of mAbs in mammalian cell-culture processes using digital twins. 

A challenge for widespread use of these new technologies is that digital technology changes at a much faster pace than the pace of adoption of new technology in biopharma manufacturing, concludes Verlenden. Change is happening, however. “Around 90% of the customers we speak with realize that they will need to upgrade their process automation to control, analyze, and optimize next-gen processes in the next five years,” says Verlenden.

 

Remote monitoring

Automating processes and allowing monitoring from remote locations outside the cleanroom benefits biopharma manufacturing by keeping people-and their inherent contamination risk-out of aseptic processing areas. “Providing access to both the system and process information from a remote location while maintaining system security has been a key focus of many groups in recent years,” says Bob Lenich, director of global life sciences at Emerson. “Incorporating data diodes that offer secure, one-way delivery of information without exposing the process to outside risk is a growing trend. Process engineers can see what’s going on in real-time and do their job without physically being on the plant floor.”

Digital twins are another technology that allows virtual engineering. “We’ve seen an increase in the use of high-fidelity digital twins to help test planned changes or designs for automation and to train operators so that they don’t have to assemble somewhere in a facility,” notes Lenich.

Virtual connections, such as smart glasses, enhance the ability of operators inside a cleanroom to interact with teams, experts, and vendors outside of the cleanroom during critical moments, says Angelo Stracquatanio, CEO and cofounder of Apprentice. The company’s Tandem platform, for example, uses hands-free, cleanroom-compliant, augmented reality (AR) smart glasses and accessories as tools that allow remote troubleshooting, facility tours, acceptance testing, and technical transfer activities, as well as immersive training in simulated AR environments. 

And as more commercial manufacturing is moving to 24/7 production, remote technical support is becoming more commonly used to provide incident resolution for manufacturing information technology applications, adds Wendy Reddington, UK general manager for Zenith Technologies, a Cognizant company. A remote engineering team can either securely log into the site application to troubleshoot and resolve incidents directly, or those on location can be guided through the repair process by the remote team, she explains.

During the travel and “social distancing” restrictions caused by the COVID-19 pandemic, online tools have been seeing more widespread use. “Analytics, accessible remotely and on mobile devices- especially for areas such as equipment performance monitoring-allows organizations to keep a close eye on their operations. We have seen this play a pivotal role recently in operations where there is limited access to a site or actual operations,” notes Lenich. 

“We started to see the demand for intelligent manufacturing platforms increase over the past year, but this recent pandemic has really accelerated the interest and understanding in its value. I think it pushed organizations to focus on long-term solution implementations that can survive global disruptions,” concludes Stracquatanio.

References

1. A. Shanley,BioPharm Intl. 33 (4) 14-17 (2020).
2. Sartorius, “Sartorius Launches BIOSTAT STR Generation 3 with BIOBRAIN-Making Biopharmaceutical Development and Production Faster, Simpler, and Safer,” Press Release, March 9, 2020.
3. V. Chopda et al., “Integration of Process Analytics for Real-Time Monitoring of Critical Quality Parameters and Attributes for Continuous Biomanufacturing,” Poster at IFPAC (North Bethesda, MD, 2020).
4. BioPhorum, “A Best Practice Guide to Using the BioPhorum Digital Plant Maturity Model and Assessment Tool” (May 2018).
5. Sanofi, “Sanofi receives 2020 International Society for Pharmaceutical Engineering (ISPE) Facility of the Year Award,” Press Release, April 9, 2020.
6. Sanofi, “Factory of the Future,” Sanofi.com, April 8, 2020.
7. J. Markarian, Pharm. Tech. 43 (3) 16-21 (2019). 

Article Details

BioPharm International
Vol. 33, No. 6
June 2020
Pages: 10–15

Citation

When referring to this article, please cite it as J. Markarian, "New Tools Facilitate Automated Process Control," BioPharm International 33 (6) 2020.

download issueDownload Issue : BioPharm International-06-01-2020