There are several different alternatives for how an upstream manufacturing process can be designed, but in principle they are all variations of just four principal modes of operation: batch, fed-batch, perfusion, and continuous fermentation. Each of these four process modes has its own advantages. In the early days of biopharma, batch processes prevailed, primarily due to the simplicity of a batch operation. With time, fed-batch has become the dominant mode in both microbial and animal cell biomanufacturing. The main reason is the increased volumetric productivity in fed-batch processes compared with batch, in combination with a relatively straightforward set-up. Perfusion was long regarded a hassle, and this process mode was only used if neither batch nor fed-batch was an option, for example, if the product was easily degradable or toxic, if the volumetric productivity was very low, or if the product had a biologic activity that affected the culture in an unwanted manner.
However, perfusion and perfusion-like processes have recently gained a renewed interest in animal cell bioprocessing for stable products, and an increasing number of biomanufacturers are investigating the possibility of using alternatives to the conventional fed-batch, primarily to increase the overall production capacity and facility output. The continuous fermentation process mode, per the classic definition, is primarily employed for microbial processes, often in research. However, the broader concept of continuous bioprocessing has become increasingly common amongst biomanufacturers. In the latter case, the term “continuous” does not refer to the classic microbial process mode, but is instead used as a broad description of unit operations and processes which are performed in a continuous manner. This is contrary to conventional batch operations where one unit operation is finalized before the next is started. Many industries have already made the transition from batch to continuous manufacturing to increase facility output, minimize downtime, and increase overall efficiency. Examples of products produced in continuous processes include cars, chemicals, stainless steel, and pulp. In this chapter, the focus will be on the process modes per their classic definition, and the next paragraphs are overviews of the four principal upstream modes of operation and some variants thereof.
Batch Culture
A batch operation is performed in a system where all necessary medium components and the inoculum are added at the beginning of the culture. The only material exchange between the process and the exterior after the start time point are gases, solutions for pH control, antifoam (if needed), and potentially process specific additions like inducing agents, viral stocks, and transfection complexes. Dissolved oxygen, pH, and temperature (with the exception of temperaturecontrolled induction and when using biphasic process strategies) are normally held constant during a batch, and the initial medium composition is critical for the outcome of the process. The timing of the harvest will depend on the product and the production kinetics, and the entire culture volume is usually harvested simultaneously.
Batch was the method of choice in the early days of biotechnology, and the simplicity of this set-up provide benefits for very large scale processes and for complex processes, for example, transient transfections and vaccine production with microcarriers. A batch variant is fill-and draw or batch-refill processes. In these processes, a small portion of the production culture is left in the production vessel at harvest, to serve as an inoculum to the next production culture. Besides being used in production, batch is also used for routine maintenance culture, for research and development purposes, and for seedtrain cultures in the scale-up process. As an example, early process development screens are typically performed in batch. However, batch is not an efficient process to maximize volumetric productivity, and in manufacturing other process modes tend to dominate.
Fed-Batch
A fed-batch starts like a batch, with essential medium components and inoculum added in the beginning of the culture. However, a fed-batch includes a gradual feed of highly concentrated, fresh nutrients during the culture to prolong the growth phase and improve productivity. The feed solution can consist of a single nutrient (for example glucose) or multiple nutrients, and a fed-batch process can include simultaneous addition of more than one feed. Such parallel feeding adds complexity but enables a tailored nutrient supplementation, and it can also decrease the risk for precipitation of feed concentrate. Nutrients can be added continuously (continuous feed) or intermittently (bolus feed). A continuous feed profile can be designed in many ways; it can be constant, linearly increasing, exponentially increasing, step-wise increasing, or automatically adjusted based on feed-back from a sensor measuring a culture parameter such as cell concentration, glucose concentration, or bioreactor volume. Not all continuous-feed profiles are easily scalable though, and in very large scale simplicity and robustness might need to be prioritized before sophistication. In these cases, as an example, bolus feeds could be a better option compared with continuous feeds.
The feeding in a fed-batch process will result in an increase in solute particles in the culture and consequently an increase in osmolality. Most cell culture media for animal cells have an osmolality in the range of 260 to 320mOsm/kg to mimic the solute characteristics of serum (290 mOsm/kg). Osmotic pressure is a key parameter in the regulation of nutrient transport across the cell membrane, and a change in osmolality can affect various cellular functions. An elevated osmolality can have positive impacts on the specific productivity of certain cell lines, but hyperosmotic conditions are detrimental for both productivity and growth. A robust clone that tolerates high osmolality can be selected during the screening process in early process development. Cell line engineering to counter osmolality-induced apoptosis and autophagy has also been proposed.
The peak cell concentration and the product yield in a fed-batch varies depending upon the host cell, production clone, culture medium, feeding schedule, and bioreactor control regime. A modern CHO fed-batch process typically takes from 10 to 14 days but longer processes also exist. The peak cell density for CHO cells it is often in the range of 10 to 30 million cells/mL. This is a considerable progress compared with the 1990s when processes peaked at single-digit cell concentrations, and yields were in the low mg/L range for a typical antibody. Processes in clinical stages have now commonly product concentrations in the 3 to 5g/L range, and up to 10 g/L have been reported. Microbial fed-batches are much shorter; for example, a typical E. coli fed-batch might take from 12 to 48 hours depending on the product, whereas a P. pastoris yeast fermentation can take somewhere between 4 and 7 days depending on the protein and the process.
The fed-batch process mode is in many ways a benchmark for biomanufacturing processes, and fed-batch has been an attractive choice for production during the past decades thanks to its robustness and operational simplicity. At the time of writing, most of the biotherapeutics that have received market approval and most products that are currently in the clinical pipeline are produced in fed-batch processes. Even though fed-batch as a concept is mature, and considerable advances in fed-batch operations have been done during the last decades, further improvements can be made. Three considerations are seen below.
● Maximize space time yield (STY). A typical fed-batch for animal cells takes about 2 weeks. If the process can be shortened at the same time as the productivity stays the same, then more batches can be produced per year, and the facility productivity will increase. This is a consideration, for example, in the timing of the harvest. Is it more feasible to harvest early and start the next batch quickly, or harvest late and increase the per batch productivity?
● Minimize cell death. All cells that die during a fed-batch process will remain in the culture medium during the cultivation process and the intracellular components that leak out into the culture will need to be handled in downstream purification. If cell death can be kept to an absolute minimum, then the risk for product degradation decreases, and the downstream process becomes less complex. Cell culture medium formulation, feeding strategy and bioreactor parameter settings are important considerations here.
● Improve product quality. A high volumetric titer might not lead to a high overall process yield if the product aggregates or if it does not meet the CQAs. Controlling product quality through cell line engineering, clone selection, cell culture media formulation, feed strategy, and bioreactor parameter settings is critical for a good overall outcome of the process. For example, the product quality might vary during the course of a fed-batch, and this can lead to product heterogeneity in the harvest.
Finally, even though fed-batch is a very versatile mode of operation, it might not be the optimal choice strictly from a facility utilization and efficiency standpoint. After the completion of each fed-batch comes a non-productive phase with cleaning, cleaning verification, and preparation for the next process. A continuous operation such as perfusion has less down-time and can lead to an increased overall facility utilization, but perfusion has other considerations as seen in the next section.
Perfusion
Perfusion is a process run at constant volume with a continuous feed of fresh culture medium into the bioreactor at the same time as spent medium is removed at an equal rate. A cell retention device keeps the cells in the reactor, whereas all low-molecular components will be removed together with the spent medium. Consequently, the product is continuously harvested, which in practice means that a certain volume of spent medium is collected and then taken to downstream purification. A perfusion process can run for a very long time. Process times of 30 to 90 days are common, but longer processes of more than 6 months also exist [6]. This process mode is not used in microbial fermentation. The reason is that microbes proliferate so quickly that there is no practical need to recirculate biomass, contrary to the slowergrowing animal cell processes.
From a physiological perspective, a perfusion process more closely resembles the natural habitat of cells from multicellular organisms, compared to batch or a fed-batch processes. The continuous flow of culture medium will lead to a metabolic steady-state during perfusion, the culture will have a steady supply of nutrients, and the level of toxic by-products will remain low. This can have a positive impact on cell physiology and productivity, and it enables processes with very high cell concentrations. Many perfusion processes fall within the range of 30 to 80 million cells/mL. If the cell concentration becomes too high, or if the viability drops because of dead cells and debris, a small part of the culture can be bled out to ensure that cell viability stays high.
Perfusion processing can improve productivity compared with batch and fed-batch. First, the high cell concentration in a perfusion process enables a higher volumetric productivity. Second, the improved nutrient availability and steady-state conditions can lead to a higher cell-specific productivity. Third, perfusion can lead to an augmented annual productivity and higher facility utilization because of less downtime between batches. Consequently, if the goal is to produce a pre-determined amount of product annually, then smaller scale bioreactors can be used if the process is run in perfusion mode compared with batch/fed-batch. Commercially approved perfusion processes are run at bioreactor scales from 75 L to 4000L, but it is generally believed that the sweet spot for modern perfusion processes is in the range of 500–1000L or less.
The perfusion equipment and instrumentation is more complex compared with the instrumentation required for batch and fed-batch processes. A major difference for perfusion is the need for a retention device to prevent the cells from being washed out from the bioreactor. The retention device can be filter-based or gravity-based, and it can be integrated in the bioreactor or externally placed. External devices enable replacement in the event of a failure. This is not possible in the case of an internal device such a spin filter. If an internal spin filter fouls, then the whole production run needs to be terminated.
The perfusion rate depends on the nutrient requirements of the culture, which subsequently is related to the cell concentration and the nutrient composition of the culture medium. The perfusion rate is typically measured in reactor volumes (RV) per day and/or cell-specific perfusion rate (CSPR) pL/cell/day. Processes with non-optimized culture medium will need a high perfusion rate to avoid nutrient limitation in the culture, and this puts a lot of strain on the liquid handling logistics. In those cases, medium optimization during the process development phase can save both time and cost. A good starting point for CSPR is 50–100 pL/cell/day, and in the final process, a perfusion rate of 1–2 RV/day is considered a benchmark. Some existing processes have a much higher perfusion rate, with 5 RV/day or even higher. These high volumes can bring logistical problems and added complexity related to the large liquid volumes that need to be prepared. In addition, the downstream purification can be more challenging if the product is very diluted.
Historically, perfusion was only used when the product was easily degradable (coagulation factors), if the product activity decreased upon long residence time in the bioreactor (enzymes), if the product had any undesired metabolic effect on the production culture (growth factors, toxins), or if the volumetric titer was very low (antibodies). Fed-batch was in most cases considered a more feasible option because of the less complex set-up and easier validation compared to perfusion. For the approved biotherapeutics manufactured using animal cell culture in 2015, fewer than 10% were estimated to be manufactured in a perfusion process. However, the perception of perfusion started to change in the mid-2000s when perfusion and perfusion-like processes were mentioned as alternatives to traditional fed-batch processes. Underlying reasons for this change were developments in cell culture media composition and cell line engineering, which enabled intensified perfusion processes with higher volumetric productivity and harvest with more concentrated product. In addition, stirred-tank singleuse bioreactors with volumes up to 2000 L were now available, with advantages that included an increased flexibility and a decreased upfront investment. Other important changes were developments in the retention technology area and better in- process controls, both aspects contributing to a more stable perfusion process. Taken together, this opened the attractive possibility of designing flexible, highly productive perfusion processes in single-use bioreactors 10 to 30 times smaller than the conventional stainless steel bioreactors used to produce the same amount of product in fed-batches. Companies such as Genzyme, Bayer, Janssen, Biomarin, and Merck Serono have successfully implemented perfusion or perfusion-like processes and lead the industry in its adoption.
Without a doubt, continuous operations like perfusion can offer advantages compared to batch-wise processes, including a higher volumetric productivity and better utilization from an overall facility standpoint. In a real scenario, the comparison might be rather complex, and the complete life cycle of a product, from the development stage to large-scale manufacturing, should be considered. The benefits of continuous operations caused many other industries to move away from batch processes a long time ago. However, the robustness and high productivity of modern bioprocess fed-batch operations, the large amount of existing knowledge about fed-batch, and the current biomanufacturing infrastructure that is designed for fed-batch all contribute to maintaining the current status quo where fed-batch is the preferred choice for most bioprocesses. As demands for efficiency increase, continuous operations will likely become more common in the future. However, the opportunities associated with perfusion need to be carefully weighed against the risks and potential pitfalls.
Continuous Fermentation
Continuous fermentation is a microbial process with a constant flow of culture medium through the reactor. The main difference compared with an animal cell perfusion process is that no device prevents the biomass from staying in the culture vessel in a continuous fermentation.1 The volume in a continuous fermentation is usually constant in industrial applications, but it can fluctuate in specific processes such as waste water treatment. The concept of continuous fermentation processes is closely linked to the chemostat, where one nutrient is growth limiting and used to determine the growth rate. However, there are several other, less common ways by which a continuous fermentation can be controlled: through constant pH (pH-auxostat), constant optical density (turbidostat), and constant substrate (nutristat).
Continuous fermentation starts as a batch process. At a certain point, for example, when the culture reaches the exponential growth phase, or when the culture becomes substrate limited, a feed with fresh growth medium is started, and an equal volume of culture broth is removed. Continuous fermentation is a superior tool in research, but the number of industrial applications is limited. Reasons for this include an increased risk for contamination, risk for genetic drift in the culture, and difficulties to control the process.
