The concept of automation conjures up images of robots on assembly lines or perhaps automobiles replacing horse-drawn carriages. In both examples, automation provides an ability to work tirelessly, with reproducible high-quality outputs at increased speed. For cell therapy, automation can be used to increase the scale of cell culture operations (e.g., bioreactors replacing flasks) and allow the use of closed systems that can protect cell products from contamination with adventitious agents from the environment or operators themselves. Closed systems also can protect a facility from product-related contaminants or viral reagents and reduce the risk of cross contamination between products made in a shared facility.
Automation can improve analytical testing that supports manufacturing and can facilitate quality control by reducing operator variability and subjective judgment of results (e.g., based on visual interpretation). Automation allows for continuous monitoring of process control parameters that permits validation of cell therapy processes. Once you can reliably monitor a process, you can replicate it to allow scale-out or reproduce the process in other facilities.
Barriers to using automation in cell therapy include availability of equipment from vendors. Cell types and processes are diverse, and technology platforms are lacking because so few cell therapies are yet commercialized. The difficulty in manufacturing cell therapies is another barrier to effective automation, which is in part attributable to the inherent complexity of living cells and challenges associated with “personalized medicines” (e.g., variation in donor cells for autologous therapies). The cost of developing and implementing automation can be high, and introducing it during clinical development or commercialization of cell therapies can result in a comparability risk. Developers of cell therapies therefore need a well–thought-out strategy to properly implement automation into a process, and they need to develop processes that can be automated.
Cell therapy manufacturing in academic laboratories is normally a manual process typically performed by scientists who use their experience to judge when cells have reached confluence and are ready to be passaged. The cell culture process is time consuming. A number of required manual manipulations such as opening bottles and transferring liquids require proper aseptic techniques and controlled environments such as biosafety cabinets. However, even the most careful and skilled scientist can accidentally cause contaminations. The scale of automation-free manufacturing often is limited by the number of flasks that can be handled by an operator.
Even the analytical procedures associated with cell therapy manufacturing are time consuming and highly variable. The manual procedure of counting viable and nonviable cells typically involves loading cells stained with trypan blue onto a hemocytometer, which requires an operator to count cells in marked grids. That procedure is prone to operator variability and sampling errors related to small volumes and dilutions. Operators must decide whether to count cells on the grid borders and whether cells have taken in the dye. Repeatability is also an issue. The same test is performed to support manufacturing and quality control for process monitoring (e.g., of yields), to determine process parameters such as cell density, and for lot-release and stability testing. So the test must be robust to make results comparisons possible between sites and over time.
For cell therapy manufacturing, automation could reduce the number of (or eliminate) open operations to lessen the risk of microbial contamination by using closed systems. Doing so is particularly important for cell therapy products because they cannot be sterile filtered to remove microbial contaminants. An automated process can be performed by operators with less training than academic specialists have. The scale of manufacturing can be increased because the environment in which equipment operates does not require a biosafety cabinet to maintain aseptic boundaries. By removing most direct human interactions, automation facilitates process transfer to multiple manufacturing sites. It also could improve process control and validation to improve quality, reproducibility, and process robustness. For example, the simple process of detaching cells from a substrate can require a physical shock to their culture vessel that induces a wave in the media to detach those cells. It is impossible to produce the same force every time if that physical shock is delivered manually, thus creating a source of process variation.
The ability to automate such processes also could provide higher levels of control with automation-management systems and/or data collection through historian software that could be part of an integrated information technology (IT) system. They could interact with laboratory information management systems (LIMS), allowing for electronic batch and lot-release records to be reviewed efficiently by a quality assurance group for flagging deviations or as part of a program for continued process verification.
For commercial production, an automated cell therapy manufacturing process thus is a highly desirable state. Because changing from manual to automated operations invariably necessitates some level of process change, implementing automation to an existing regulated commercial process may be infeasible (or at best nonoptimal). Concerns could arise over product comparability with material used in pivotal clinical trials on which market approval was based. With no automated equipment for cell therapy processes readily available at the outset of early phase development (for reasons outlined below), developers need a strategy for its implementation during their products’ life cycles.
Implementation of automation technology in a manufacturing process and analytical laboratory requires a strategy that balances
- cost and time needed to implement an effective automation solution
- development stage of a given cell therapy product and its probability of clinical/commercial success
- product comparability as the process is changed, scaled-up/out, or transferred.
As yet, off-the-shelf equipment is limited for automating cell therapy processes partly because cell types and processes are diverse and the industry lacks a critical mass to establish platforms. Thus, custom design and manufacture of automation equipment might be needed for a given process. Reliance on custom equipment significantly increases both the capital investment and the time required to design and build that equipment. Also, additional time must be incorporated into development timelines to troubleshoot and integrate novel, custom equipment into good manufacturing practice (GMP) manufacturing facilities. High monetary and timing costs often push implementation of automation forward in development, when manufacturing scales and net-present value of products are high enough to justify such investments (and development cycles tend to be relatively longer). Processes are better understood later in development, allowing a better understanding of what automation must achieve. Processes also become less subject to change in late development. Because automation typically reduces flexibility, this is an important consideration in early development, when processes need to adapt rapidly as new process/product knowledge is gained.
Implementation of automation later in development will necessitate some level of process change, so a company will need a strategy for demonstrating comparability. The mantra in cell therapy manufacturing states that “the product is the process.” That is especially the case with insufficient process characterization data. However, sufficient understanding of both process and product with a strategy to demonstrate comparability enables changes to be made. That said, it is still optimal to design a strategy for commercial manufacturability and automation before initiating clinical trials, enabling preclinical studies to reduce comparability risk.
A company’s ability to implement an automation strategy depends on its degree of process knowledge and in-house understanding of how critical process parameters (CPPs) relate to critical quality attributes (CQAs) under the quality by design (QbD) paradigm. If those are unknowns, it is impossible to establish the parameters of operation or design user requirements. It will be impossible to determine whether parameters need to be monitored or controlled. For example, is it critical that pH be monitored or controlled — or only the concentration of carbon dioxide?
Some cell therapy product developers wrestle with highly variable inputs into their manufacturing processes — such as differences in cell populations among donors in autologous cell therapies or differences among lots of animal-derived growth media — resulting in product variability. Without prerequisite process monitoring (through offline in-process testing or in-line process analytical technology, PAT) or the understanding and ability to adjust a process based on those measurements, dogmatic execution of a fixed process with automated equipment will do nothing to lessen product variability. In fact, it may increase variability. Process understanding is required first to define CPPs and CQAs and determine how they must be monitored and controlled through automation to minimize process and product variability.
Process Understanding and Design Enable Effective Automation: The type of automation used in manufacturing cell therapies depends on the cell types involved and the technology used to isolate, expand, and purify them, as well as the type of final container used. An important step to prepare for process automation is systematic mapping of the process and related knowledge to create a process flow diagram. Process knowledge combines CQAs with a control strategy (process monitoring and controls) based on criticality analysis and data. A complete process flow diagram allows for equipment and process changes to take place over time. Process mapping and knowledge allow developers to define user requirements, which can be used to define what off-the-shelf equipment capabilities are needed. Developers of cell therapies often allow the equipment available to them to define their manufacturing processes because they do not have the data needed to characterize their processes adequately otherwise.
In generating a process map, it is useful to challenge process steps toward understanding whether they are required and how they can be performed to achieve desired outcomes. In research laboratories, for example, passaging of substrate-dependent cells requires a buffer such as phosphate-buffered saline (PBS) to wash those cells and reduce the levels of trypsin-inhibitor–containing serum that is added to detach the cells. Doing so requires multiple steps, and the duration of each step must be tightly controlled. Cells will die if not kept hydrated, and trypsin activity also can damage them. Replacing trypsin with an alternative such as TrypLE recombinant enzyme from Thermo Fisher can allow for longer exposure without reducing cell quality. That can eliminate the PBS washing step entirely because the enzyme is less sensitive to serum levels than trypsin is. So the process could be optimized before automation to allow for fewer steps, thus broadening operating parameters such as time or enzyme activity.
Implementation of closed processing has some limitations, such as small-volume transfers typically performed manually with pipettes and near-complete media exchange usually performed by removing supernatant from cell pellets after centrifugation. Solutions might need to be made at different concentrations to allow for transfer of larger volumes. In addition, the elutriation buffer will be part of the final formulation if a process includes counterflow elutriation that maintains cells in a fluidized bed. So if the product will be cryopreserved, then its cryoprotectant components will need to be concentrated in the cryoprotectant solution and account for salts and sugars in the elutriation buffer to ensure that osmolality is controlled.
The timing of operations is important to control during process scale-up and automation. It is a process parameter that can be easily overlooked but can be critical for in-process stability. GMP processes usually take more time than laboratory processes because they require documentation, scale-up, preparation, and control. Because cells are not in an optimum environment in the downstream cell-recovery part of the process, this may lead to product-quality changes if the timing is not appropriately studied and controlled. Centrifugation is a good example of such a unit operation. It can be performed quickly in a research laboratory setting, but when counterflow elutriation of tangential-flow filtration (TFF) is used, the time required to change buffers or wash cells can increase depending on the scale and technology used.
Manufacturing process design can use the principle of design-to-value, which can reduce lot failures by challenging whether processes are overdesigned. In the design and implementation of automation, a staged approach can be beneficial. That allows staged investment in equipment and learning about the equipment and process over time. It is also important to consider the amount of technical complexity involved and a technology’s ability to scale up or out. That will determine the level of investment required and your ability to implement automation during clinical development while minimizing the risk of differences in product made before and after that change.
Implementation of Automated Equipment
Development of an automated manufacturing process requires the work of a crossfunctional team experienced in GMP manufacturing, bioprocessing, quality assurance, validation, regulatory, facilities, and equipment engineering. A multidisciplinary team is needed to ensure that a process and associated equipment provide a solution that will be efficient and can be implemented in a GMP facility. Issues such as health and safety, cleaning, particulate formation, sensor controls, connectivity to automation systems, operator interfaces, and facility utilities are important to capture in user requirements documents. That helps a company ensure that the factory acceptance and site acceptance testing will be straightforward.
A company’s choice of technology can depend on the process duration and scale as well as the type of operation. For a small-scale process completed within days, it may be suitable to have one piece of equipment that combines several unit operations. At larger scales, however, it may be easier to have modular equipment, with each module designed to complete a defined unit operation and linked to the others by transfer of materials in a closed process. That can provide flexibility in scale-up and may allow for the same equipment to be used in different processes.
One of the team’s first actions should be to determine what defines a lot, which can drive the scale of equipment as well as the process itself. For example, technology selected for expanding cells can be determined based on appropriate lot size and the number of cells required for each lot. That is based on cells per dose and the number of doses required at commercial scale.
Here’s another example: A justification can be made that vials cryopreserved in multiple programmable, validated cryofreezers be combined into a single lot if they are all cryopreserved at the same time. If vials are frozen at different times when there are too many to fill just one programmable cryofreezer, then variability within that lot would create sublots. In this case, the lot/sublot definition can help define the scale of a unit operation and thus the controls that need to be implemented.
The frequency that a process is performed in commercial operation also is an important consideration. If cell therapy manufacturing is performed in campaigns that occur infrequently, then it is important to consider operator refresher training as well as the effects on equipment from a preventative-maintenance perspective. In our experience, cell counting equipment needs to be used routinely to ensure that it performs as expected both for calibration and in use.
An automated campaign can use a lot of single-use disposable tubing sets and bags. It is important that those consumables be appropriate for use to ensure the quality of cell therapy products. Key considerations include the level of particulates, the potential of leachates and extractables, and the integrity of seals and connections. For the latter, it is preferable to use overmolded connections wherever possible instead of barb connectors and ties. The presence of leachates can vary from lot to lot depending on the batch of resin, associated stabilizers, and amount of sterilizing radiation (1). It is therefore important to purchase consumables from vendors with quality systems that address those issues. When using closed systems and automation, it may be useful to dilute reagents (making and storing them ready for use) and to package them in bags or containers with tubing that can be welded. Stability and sterility of media and reagents in such containers need to be determined.
Even when equipment is commercially available for automating processes such as cell counting, it is still incumbent on a cell therapy developer to demonstrate that such equipment is fit for purpose through installation, operational, and performance qualification (IQ, OQ, and PQ). The equipment supplier might provide instructions for use, but the cell therapy manufacturer must demonstrate that set points and potential ranges of parameters or settings it uses are appropriate for its own product and process. For example, automated cell counting equipment must follow ICH Q2 suggestions for determining accuracy, precision, repeatability, specificity, detection limit, quantitation limit, and linearity range for process intermediates in different matrices as well as the final product (2). Again, the time required for sample preparation and testing needs to be controlled.
A Coming Convergence?
As the field of cell therapy matures, its degree of manufacturing automation will naturally increase to allow more product to be made both cost effectively and high in quality. Because of the current diversity of products in development, no single clear platform has been developed for cell processing. But as cell therapies reach commercialization, commonalities may arise that allow for convergence to a platform automation that will allow for greater access and reduced costs of these next-generation biological products.
1 Lindskog E, et al. Pharma Implementation of Raw Material Control Strategies in the Manufacture of Single-Use Bioprocessing Containers. BioPharm Int. 28(1) 2015.
2 ICH Q2: Validation of Analytical Procedures: Text and Methodology. US Fed. Reg. 1 March 1995: 11260; www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q2_R1/Step4/Q2_R1__Guideline.pdf.
Corresponding author Ian Harris is product development team leader in cell therapy, and Francis Meacle is director of cell therapy manufacturing; and Donald Powers is senior manager of vaccines and advanced therapies supply chain at Janssen Research and Development, 1400 McKean Road, Spring House, PA 19477; firstname.lastname@example.org.