Key Takeaways
Advances in CHO cell engineering, media, high-density seeding, fed-batch intensification, perfusion, and process control are increasing volumetric productivity in mammalian cell culture.
Process intensification is a continuum that includes optimized fed-batch, non-perfusion seed intensification, N-1 perfusion, hybrid or intermittent approaches, and full production perfusion.
Higher volumetric productivity can reduce the upstream bioreactor volume required for a given output, but downstream bottlenecks can limit facility-level gains.
PAT, 12 control, and model predictive control can support the tighter process understanding and control required as cell densities and productivity increase.
The optimal manufacturing architecture depends on demand, media use, downstream throughput, operational feasibility, product quality, and economics, so maximum volumetric productivity is not always the optimal endpoint.
When Manufacturing Scale Is No Longer Just a Question of Liters
For mammalian cell culture, manufacturing capacity has long been easy to visualize in physical terms: how many bioreactors a facility contains, how large those vessels are, and how many liters of culture they can hold. Those measures still matter, but they increasingly provide an incomplete picture of what a facility can produce. Improvements in Chinese hamster ovary (CHO) cell engineering, cell line development, media, high-density seeding, fed-batch intensification, perfusion, and process control are increasing the amount of useful output that can be generated from a given volume of culture. Across different intensified operating modes, volumetric productivity can vary severalfold even when the same cell line and media platform are used.1,2
That changes the meaning of scale. A process that generates substantially more product per liter per day may require less upstream volume to meet a given production target, and facility models show that intensified or continuous configurations can sometimes translate those gains into smaller overall manufacturing footprints. One recent analysis modeled a facility-footprint reduction of up to 51% for an end-to-end continuous process compared with an optimized fed-batch reference process, although that result reflects a specific modeled configuration rather than an industry-wide expectation.3
The central shift is toward thinking about capacity less as installed volume and more in terms of how much productive capacity can be designed into each liter, how consistently that productivity can be maintained, and whether the rest of the process can handle the resulting throughput. More productive cells, higher-density inoculation, intensified fed-batch, and perfusion can all alter what a given vessel can accomplish, often in combination.2,4,5
Designing More Productivity into the Cell
The productivity equation begins before the production bioreactor. CHO cells remain the dominant host for therapeutic antibodies and other recombinant biopharmaceuticals, but productivity bottlenecks persist, particularly for complex molecules. Cell engineering has progressed from comparatively simple overexpression strategies toward targeted knockouts and broader pathway-directed approaches, including combinatorial engineering of cellular functions associated with secretion, proliferation, apoptosis, and other mechanisms that influence production.1
That work is also becoming more closely connected to the manufacturing process in which the cell line will eventually operate. Conventional cell line development (CLD) workflows have generally selected clones for strong fed-batch performance, but the characteristics that produce the best fed-batch clone do not necessarily yield the best perfusion performer. A perfusion-specific CLD workflow has been shown to benefit from media designed for perfusion, scale-down models that better represent the intended process, and selection criteria that consider growth, cell-specific productivity, and volumetric productivity.6
Intensification is therefore not simply a process development decision that can always be layered onto a cell line after selection. If the intended manufacturing strategy may include perfusion or another highly intensified mode, CLD criteria can influence the performance available later. For drug developers and contract development and manufacturing organizations (CDMOs), this pushes part of the manufacturing strategy earlier in development, when decisions about cell line selection can either preserve or narrow later process options.6
Media and Feeding Are Part of the Productivity Architecture
Cell line selection sets part of the productivity envelope, but realizing that performance depends on the culture environment. Media composition influences CHO cell growth, productivity, and product quality, and contemporary media development is increasingly being informed by systems biology and machine-learning approaches intended to make optimization more predictive.7
The connection to process intensification is especially important. Higher-density cultures place different demands on nutrient availability and metabolic management, so increasing the number of cells at inoculation does not guarantee a durable productivity gain. In an intensified fed-batch platform built around N-1 perfusion, where the seed stage immediately preceding production was operated in perfusion mode, redesigned basal and feed media were critical to maintaining higher viable cell densities and cell-specific productivity across the production run. The platform increased production-bioreactor seeding density to 10–20 × 106 cells/mL and increased already industrially relevant titers by approximately 100% across four monoclonal antibody (mAb)-expressing CHO cell lines during 10- to 14-day runs.4
Substantial gains also remain available within fed-batch itself. By separately manipulating biomass and cell-specific productivity, one study doubled titer relative to its standard 14-day fed-batch process. It also showed that a 5-g/L target could be reached in 10 days with supplementation at medium seeding density, or in seven days with supplementation at high seeding density, compared with 14 days for the standard process.8
Those results illustrate two different uses of higher productivity. One strategy increases output within a given production window, while the other reaches a target output in less time. More productive cells still require media and feeding strategies capable of supporting them, and higher initial biomass creates value only if the process can sustain productive physiology.
Intensification Is a Continuum
The conventional contrast between fed-batch and perfusion obscures how many intermediate strategies are now available. Process intensification can raise output by increasing inoculation density, sustaining higher biomass, increasing cell-specific productivity, shortening nonproductive portions of the run, extending productive culture, or combining several of those mechanisms. The result is better understood as a continuum than as a binary choice between batch and continuous processing.2,4
One route is to intensify the seed train without using perfusion at all. Enriched batch or fed-batch N-1 cultures have reached final viable cell densities of 22–34 × 106 cells/mL and supported production-reactor inoculation densities of 3–6 × 106 cells/mL. In three mAb processes, those non-perfusion N-1 approaches produced titer and product quality comparable with perfusion-seeded processes and were demonstrated at 500-L and 1,000-L production scales.9
N-1 perfusion can push the concept further. In the intensified platform described above, higher seeding densities allowed the production reactor to enter the run with substantially more biomass already in place. For three products, the intensified process was demonstrated at 500-L scale with quality attributes similar to those from corresponding earlier processes at 1,000-L scale. The production step remained fed-batch; perfusion was used selectively where it created leverage in the seed train.4
Hybrid strategies blur the categories further. Intermittent-perfusion fed-batch has been used to introduce periods of perfusion into an otherwise fed-batch process. Across eight recombinant CHO cell lines, one such approach increased average titer by approximately 45% relative to intensified fed-batch, while an ultra-intensified version using very high initial seeding densities achieved up to about six times the productivity of traditional fed-batch and three times that of intensified fed-batch in the reported experiments.10
These approaches show that process intensification does not require wholesale conversion to continuous production. A drug developer may capture part of the productivity advantage through media optimization, high-density seeding, or selective perfusion while retaining a fed-batch production step. Another process may justify intermittent media exchange or full production perfusion. The relevant question becomes how much intensification a particular molecule, facility, and demand profile actually requires.
What Production Perfusion Changes
Full production perfusion remains one of the most powerful ways to alter the relationship between vessel volume and output. Continuous media exchange can sustain much higher cell densities than conventional fed-batch, maintaining more productive cells within each liter for longer periods. In a controlled comparison using the same CHO cell line and media set, perfusion achieved volumetric productivity of up to 2.29 g/L/day and concentrated fed-batch reached up to 2.04 g/L/day, compared with 0.39–0.49 g/L/day for the fed-batch processes evaluated. Cell-specific productivity was broadly comparable across the operating modes, indicating that much of the volumetric advantage came from the higher cell densities that the intensified processes could maintain.2
High-density perfusion has advanced further, with reported volumetric productivities above 5 g/L/day over extended cultivation periods. Perfusion has also been applied outside the production reactor to generate high-density seed cultures, reinforcing its role as a tool that can be deployed selectively at different points in the process rather than only as a complete manufacturing format.5
The distinction between cell-specific and volumetric productivity is important. A process can raise output by enabling each cell to produce more, by maintaining more producing cells per liter, or by extending the period during which a productive cell population can be sustained. Perfusion particularly affects the latter two variables.2,5
Time also becomes part of the capacity calculation. A process that reaches a required output more quickly changes the amount of product that a given reactor volume can generate over a defined period. Volumetric productivity captures this relationship between output, volume, and time in a way that final titer alone cannot.2,5
The tradeoff is that the surrounding process must support the new level of intensity. Higher-density perfusion requires continuous media exchange, while its greater output places additional demands on process monitoring, product-quality control, and downstream throughput. The upstream reactor may become smaller relative to its output, but the manufacturing challenge shifts toward managing a more productive system.
More Productivity Raises the Value of Process Control
As more output is concentrated into each liter and each production day, control of the culture environment becomes increasingly important to both process performance and product quality. This gives process analytical technology (PAT) a role beyond observing an intensified process: real-time measurements can support active control while the culture is running.
The quality implications provide a clear reason for that tighter control. In a perfusion study that varied cell density to represent increasing intensification, significant changes were observed in several glycan populations, supporting the need to assess how higher-intensity operation affects product-quality attributes. Because perfusion harvests product throughout the run, the study also highlighted the importance of relating process parameters and lot allocation to quality variation over time.11
Raman spectroscopy has been used to move glucose management from intermittent manual feeding toward real-time feedback control. In one study involving two CHO processes, Raman-based glucose control improved culture performance and product-quality measures; for one cell line, the controlled process increased overall titer by 25%, while the other showed substantial reductions in glycation with comparable growth and product output.12
An industry-wide assessment of PAT tools ranked in-line Raman spectroscopy, on-line liquid chromatography, and gas analysis among the technologies offering high business value at the production-bioreactor stage. These approaches can provide in-line, on-line, or at-line information that supports real-time process monitoring and control rather than relying solely on later measurements.13
Control strategies are also becoming more predictive. Model 14 (MPC) has been developed to manage interacting culture variables rather than relying only on fixed-parameter control, while machine-learning models have been incorporated into MPC systems that calculate feeding strategies while keeping metabolites and other process variables within defined constraints. These approaches have been evaluated in real-time cell-culture experiments.14,15
The productivity ceiling therefore cannot be pursued independently from control and characterization. A process that sustains more cells and produces more protein per liter must still demonstrate that it can maintain the required product profile under those conditions.
A Smaller Bioreactor Does Not Automatically Mean a Smaller Plant
The facility implications of higher volumetric productivity are compelling, but they require careful interpretation. If a process produces more material from each liter of reactor volume, less upstream volume may be needed to meet a given demand. The 500-L demonstrations of an intensified process corresponding to earlier 1,000-L processes provide a concrete example of that changing relationship at the production-bioreactor level.4
Facility modeling extends the principle beyond the reactor. One end-to-end continuous-manufacturing analysis, using an optimized 15-g/L fed-batch process as its reference, modeled reductions of up to 51% in facility footprint as well as reductions in annual production cost under the scenarios examined. Those figures show what may be possible in a specific continuous architecture, but they should not be treated as universal expectations for process intensification.3
The capacity constraint, however, can move downstream. A fully integrated continuous-process analysis found that increasing upstream production eventually caused multicolumn chromatography operations to become process bottlenecks, with the limiting point depending on chromatography capacity utilization.16
Higher upstream productivity can therefore relocate a capacity constraint rather than eliminate it. If a smaller production reactor generates substantially more mass per unit time, capture and purification operations must be able to absorb that throughput. Otherwise, the benefit of reducing reactor volume can be limited by the next unit operation that reaches its capacity ceiling.16
The real facility opportunity comes from matching high-productivity upstream processing with downstream equipment and operating strategies designed around the new throughput. Advances in cell culture can reduce the liters required upstream, but realizing the full value of that improvement depends on the performance of the entire manufacturing train.
Productivity Does Not Choose the Manufacturing Architecture by Itself
Maximum volumetric productivity is not necessarily the same as optimal manufacturing performance. Economic analyses of batch, hybrid, and continuous processes show that the preferred architecture can change with annual demand, development stage, process configuration, and the weight assigned to operational feasibility.
In one analysis of end-to-end continuous processing, the modeled cost advantage was greater at annual demands of 100–500 kg than at 1–3 metric tons, where multiple parallel continuous trains were required. The economics were sensitive to factors including perfusion rate, productivity, and media cost.17 A process that reduces individual equipment size may therefore still require enough parallel capacity or supporting infrastructure to change the overall business case.
A separate analysis reached different preferred configurations under different conditions. Integrated continuous processing ranked highest for early-phase production and small and medium-sized companies in the modeled scenarios, while a hybrid process combining fed-batch culture with continuous capture and batch polishing became preferable from a cost-of-goods (COG) perspective for commercial production and larger portfolios. When operational feasibility received more weight, the hybrid approach became preferred across company scales.18
Resource use introduces another distinction. In one life cycle comparison, a four-day perfusion-pooling scenario had similar COG to fed-batch but consumed 35% more water, required 17% more energy, and emitted 17% more carbon dioxide. Extending the pooling duration to eight days changed the environmental comparison enough for perfusion to become more favorable in the model.19
Media economics likewise resist simple generalization. In the controlled comparison of intensified modes, perfusion consumed more media but achieved enough additional productivity that media cost per gram of antibody remained comparable with fed-batch. The economic effect depended on how efficiently the additional media was converted into product.2
These comparisons show why process intensity has to be evaluated against the actual manufacturing requirement. The architecture that performs well at modest demand may not remain optimal when multiple parallel trains are needed. A configuration that performs well on COG can rank differently when operational feasibility receives greater weight, and a process with an attractive footprint may not have the lowest environmental burden under every operating assumption.17–19
Productivity, molecule behavior, media demand, downstream throughput, process duration, product quality, facility configuration, and operational complexity all influence whether a more intensified architecture creates value. The productivity leap expands the available manufacturing design space rather than pointing every process toward the same endpoint.
Designing Capacity Around Productivity
The widening range of viable operating modes changes the capacity equation. Installed reactor volume still matters, but it reveals less about potential output when the productivity of those liters can vary so widely. The challenge becomes selecting the combination of cell line performance, media strategy, process intensity, and control appropriate for the product, then designing the rest of the manufacturing train to support the resulting throughput.
For drug developers and CDMOs, the central capacity question is becoming how much reliable, quality-controlled productivity can be designed into each liter, and how effectively the surrounding process can convert that productivity into released drug substance.
The next leap in mammalian cell culture may ultimately be measured less by the size of the vessels being installed than by how much those vessels can do.
References
1. Schulz, Alexandra, et al. “Systematic review and data-driven insights into CHO cell engineering for next-generation antibody production.” mAbs. 18: 2615475 (2026).
2. Xu, Sen, et al. “Bioreactor productivity and media cost comparison for different intensified cell culture processes.” Biotechnology Progress. 33: 867–878 (2017).
3. Partopour, Behnam, and David Pollard. “Advancing biopharmaceutical manufacturing: economic and sustainability assessment of end-to-end continuous production of monoclonal antibodies.” Trends in Biotechnology. 43: 462–475 (2025).
4. Xu, Jianlin, et al. “Development of an intensified fed-batch production platform with doubled titers using N-1 perfusion seed for cell culture manufacturing.” Bioresources and Bioprocessing. 7: 17 (2020).
5. Wong, H Edward, et al. “From chemostats to high-density perfusion: the progression of continuous mammalian cell cultivation.” Journal of Chemical Technology & Biotechnology. 97: 2297–2304 (2022).
6. Balassi, Vincent, et al. “Improving outcomes in intensified processing via optimization of the cell line development workflow.” Biotechnology Progress. 41: e70003 (2025).
7. Kavoni, Hossein, et al. “Recent advances in culture medium design for enhanced production of monoclonal antibodies in CHO cells: A comparative study of machine learning and systems biology approaches.” Biotechnology Advances. 78: 108480 (2025).
8. Mahé, Amélie, et al. “Exploring the limits of conventional small-scale CHO fed-batch for accelerated on demand monoclonal antibody production.” Bioprocess and Biosystems Engineering. 45: 297–307 (2022).
9. Yongky, Andrew, et al. “Process intensification in fed-batch production bioreactors using non-perfusion seed cultures.” mAbs. 11: 1502–1514 (2019).
10. Xiang, Shaoxun, et al. “Developing an ultra-intensified fed-batch cell culture process with greatly improved performance and productivity.” Biotechnology and Bioengineering. 121: 696–709 (2024).
11. Liang, George, et al. “Effects of process intensification on homogeneity of an IgG1:κ monoclonal antibody during perfusion culture.” Applied Microbiology and Biotechnology. 108: 274 (2024).
12. Gibbons, Luke, et al. “An assessment of the impact of Raman based glucose feedback control on CHO cell bioreactor process development.” Biotechnology Progress. 39: e3371 (2023).
13. Gillespie, Christopher, et al. “Systematic assessment of process analytical technologies for biologics.” Biotechnology and Bioengineering. 119: 423–434 (2022).
14. Jones, Jonathan, et al. “Improved control strategies for the environment within cell culture bioreactors.” Food and Bioproducts Processing. 138: 209–220 (2023).
15. Rashedi, Mohammad, et al. “Machine learning-based model predictive controller design for cell culture processes.” Biotechnology and Bioengineering. 120: 2144–2159 (2023).
16. Ding, Chaoying, et al. “Process design of a fully integrated continuous biopharmaceutical process using economic and ecological impact assessment.” Biotechnology and Bioengineering. 119: 3567–3583 (2022).
17. Mahal, Hanna, Harvey Branton, and Suzanne S Farid. “End-to-end continuous bioprocessing: Impact on facility design, cost of goods, and cost of development for monoclonal antibodies.” Biotechnology and Bioengineering. 118: 3468–3485 (2021).
18. Pollock, James, et al. “Integrated continuous bioprocessing: Economic, operational, and environmental feasibility for clinical and commercial antibody manufacture.” Biotechnology Progress. 33: 854–866 (2017).
19. Bunnak, Phumthep, et al. “Life-cycle and cost of goods assessment of fed-batch and perfusion-based manufacturing processes for mAbs.” Biotechnology Progress. 32: 1324–1335 (2016).
