TECHNOLOGY EXPLAINER
Batch and Continuous Culture
— topping up or constantly exchanging, and whether to link the steps
Cell culture runs in one of several modes: batch, where nothing is added; fed-batch, where nutrients are added as the run goes on; and perfusion, where the medium is continuously exchanged. Fed-batch is the mainstream for antibody drugs, but perfusion culture has been reported to reach more than 200 million cells per mL. "Continuous manufacturing", which links perfusion culture directly to purification, is now covered by the international guideline ICH Q13 (adopted in 2022). One of the key components is the hollow-fibre membrane, which keeps the cells in the tank while drawing off only the liquid.
- What an operating mode is (the short version)
- Four modes — batch, fed-batch, perfusion and continuous manufacturing
- Cell density — from tens of millions in fed-batch to 200 million in perfusion
- Our calculation: fed-batch versus perfusion, tank size and productivity
- Cell retention devices — ATF and TFF
- A materials engineer's view (1): a hollow-fibre membrane stops cells but lets antibodies through
- ICH Q13 and where the FDA and Japan stand
- A materials engineer's view (2): a 30-day run is a life test for materials
- What could not be confirmed, and remaining challenges
- Glossary / References / Claim-to-source audit
Sourced = stated in published material or a peer-reviewed paper (link given)
Our calculation = a figure this article derived, with the assumptions spelled out
Not yet confirmed = a concept or outlook with no confirmed track record
Summaries of the modes and readings about materials and processes are marked separately as Commentary.
1. What an operating mode is (the short version)
An operating mode is the choice of what goes in and out during culture, when, and how much. With the same cells and the same bioreactor, the mode changes the cell density reached, the length of a run and the size of tank needed.
- The mainstream: a 2016 life-cycle assessment paper says fed-batch culture became the standard platform for large-scale antibody production because of its ease of scale-up (up to 20,000 L), robustness and high volumetric productivitySourced
- A typical picture: a 2010 review of antibody manufacturing describes typical production cell lines as having a specific productivity of 20 to 70 pg/cell/day and peak viable cell densities of 5 to 30 million cells/mL in 10- to 14-day fed-batch cultureSourced
- The other route: a perfusion culture paper notes that while batch and fed-batch are the most widely used, interest in perfusion culture is growing, citing as advantages the ability to keep a healthy culture going longer at high cell density and the short time the product spends in the tankSourced
Clincke and colleagues (2013), studying high-density perfusion culture, report that when they reached as much as 214 million cells/mL, the limit was set not by cell growth but by the capacity of the membrane to cope with culture fluid made viscous by the high cell density, and by pCO2Sourced. The ceiling on perfusion culture was set not by biology but by engineering — membranes, fluid flow and gas transfer. That is one way to read it (our commentary).
2. Four modes — batch, fed-batch, perfusion and continuous manufacturing
| Mode | What goes in and out | Features confirmed in primary sources |
|---|---|---|
| Batch | Nothing is added during the run (our commentary) | Together with fed-batch, the most widely used mode [Ref. 2] |
| Fed-batch | Concentrated nutrient solution (feed) is added during the run. Volume increases | Media development covers both the basal medium and the concentrated feed [Ref. 1]. Frequent or continuous feeding that keeps glucose and glutamine low can reduce by-products, but has not been favoured in large-scale manufacturing because of operational complexity and validation cost [Ref. 1]. Rising volume changes the mixing conditions, and limiting it with a concentrated feed makes scale-up easier [Ref. 3] |
| Perfusion | Fresh medium goes in continuously, and liquid alone is drawn off while the cells stay in the tank | An example run at perfusion rates (the volume exchanged per day) of 1 to 10 vessel volumes per day. Cell density is stabilised by removing cells (a bleed) [Ref. 2]. Rates of 0.5 to 2 vessel volumes per day have been reported for industrial cell lines [Ref. 5] |
| Continuous manufacturing | Two or more unit operations are linked directly, with material going in and product coming out continuously | Annex III of ICH Q13 shows an example linking a perfusion bioreactor, continuous capture chromatography, virus inactivation, polishing, virus filtration, and concentration and buffer exchange by ultrafiltration [Ref. 4] |
"Features confirmed in primary sources" are all Sourced. The descriptions in the "what goes in and out" column are this article's summary.
3. Cell density — from tens of millions in fed-batch to 200 million in perfusion
In perfusion culture, liquid carrying waste products can be drawn off and fresh medium added continuously while the cells stay in the tank. That makes far higher cell densities possible than in fed-batch culture. Clincke and colleagues (2013) ran perfusion culture of CHO cells in a single-use wave-type bioreactor (a 10 L culture bag) connected to a hollow-fibre cell separation deviceSourced.
- With both ATF and TFF, stable culture was achieved at 20 to 35 million cells/mL by removing cellsSourced
- Bleeding held the culture at 90 to 130 million cells/mL, in the growth phase and at high viability, for more than two weeksSourced
- The maximum was 214 million cells/mL with TFF and 132 million cells/mL with ATF. With TFF the limits were the membrane's capacity to cope with the viscous fluid and pCO2; with ATF, the capacity of the vacuum to draw in the viscous fluidSourced
- Above 200 million cells/mL, the pressure in the recirculation loop reached 1 bar and pCO2 an excessive 31 kPa. For this cell line, the authors recommend below 146 million cells/mL on the basis of the theoretical distance between cellsSourced
4. Our calculation: fed-batch versus perfusion, tank size and productivity
Bunnak and colleagues (2016) modelled and compared a fed-batch process and a perfusion process on the assumption of producing 28 kg of antibody a year. Their assumptions (values the paper set as "typically achieved in manufacturing") were as followsSourced.
| Assumption | Fed-batch process | Perfusion process |
|---|---|---|
| Production bioreactor volume | 375 L | 47 L |
| Titre (antibody concentration) | 5 g/L | 2 g/L |
| Length of one culture | 12 days | 30 days (5 days to ramp up) |
| Perfusion rate | — | 2 vessel volumes/day |
| Production bioreactor runs per year | 21 | 9 |
| Purification runs per year | 21 | 53 (pooling 4 days' harvest at a time) |
All Sourced (Bunnak et al. [Ref. 5], Table 1). These are assumptions for a model calculation, not values from any particular plant.
- Fed-batch: 5 g/L in 12 days → 5 ÷ 12 = about 0.42 g/L/day
- Perfusion (during production): liquid at 2 g/L drawn off at 2 vessel volumes a day → 2 × 2 = 4 g/L/day
- Ratio: 4 ÷ 0.42 ≈ about 9.6 times, roughly matching the ratio of production tank volumes, 375 ÷ 47 ≈ about 8.0 times
Reading: perfusion loses to fed-batch on concentration (2 g/L against 5 g/L), but because the liquid passes through the same tank many times, it wins on output per litre of tank. As a result, the model assumes the same annual output from a tank of about one-eighth the volume (our commentary). Clincke and colleagues also cite as advantages of perfusion smaller tanks, less cleaning in place, and the possibility of single-use tanks instead of stainless steelSourced. Assumptions and limits: the five-day perfusion ramp-up, product lost with the bled cells, and purification yields (72% for fed-batch, 76% for perfusion) are ignored.
In the Bunnak assessment, however, the standard perfusion process, pooling four days' harvest for each purification, had almost the same cost of goods as the fed-batch process but used 35% more water and 17% more energy and emitted 17% more CO2. That was because purification ran more than twice as often, and its cleaning (CIP and SIP) accounted for more than 85% of water use. The authors say that extending the pooling period to eight days could make the perfusion process the more environmentally friendly oneSourced. Shrinking the tank shifts the burden downstream if the linked downstream steps then have to run more often (our commentary).
5. Cell retention devices — ATF and TFF
The heart of perfusion culture is the device that keeps cells in the tank and draws off only the liquid. Clincke and colleagues list the options usable in single-use form — centrifugation, TFF, ATF, floating perfusion filters, hydrocyclones and cell settling tanks — and chose TFF and ATF as those that support high cell densities, have moderate equipment cost, can be replaced during culture and can be combined with a wave-type bioreactorSourced.
| Item | TFF (tangential flow filtration) | ATF (alternating tangential flow) |
|---|---|---|
| How the liquid flows | Culture liquid flows along the membrane surface, which helps prevent clogging. Recirculated in one direction by a peristaltic pump | Uses the same membrane as TFF, but a diaphragm pump reverses the flow direction about once a minute. The membrane is backflushed |
| Shear | Uses a peristaltic pump | Less shear than TFF's peristaltic pump |
| Maximum density reached | 214 million cells/mL | 132 million cells/mL |
| What set the limit | Membrane capacity for the viscous fluid, and pCO2 | Capacity of the vacuum to draw in the viscous fluid (pressurising the tank to 0.02 to 0.03 bar restarted the alternating flow) |
| Product passage | Antibody output was comparable, but some of the product was retained by the hollow-fibre membrane, more markedly with TFF | |
All Sourced (Clincke et al. [Ref. 2]). The results come from one study, one cell line and one type of membrane.
Challenges remain at large scale. A 2006 review by Nienow says that at scales such as those above 10,000 L, major questions remain over cell retention and recirculation devices: clogging, throughput, and a different kind of "shear sensitivity"Sourced.
To avoid running short of nutrients and accumulating by-products, Clincke and colleagues ran the perfusion rate in proportion to cell densitySourced. From two operating points in the paper, we calculate the perfusion rate per cell.
- 20 million cells/mL at 1 vessel volume/day: 1 L ÷ (2×10¹⁰ cells) = 50 pL/cell/day
- 200 million cells/mL or more at 10 vessel volumes/day: 10 L ÷ (2×10¹¹ cells) = 50 pL/cell/day
Ten times the cell density means ten times as much medium exchangedOur calculation. High-density perfusion culture comes at the price of heavy medium consumption and more liquid pushed through the membrane (our commentary). Assumptions and limits: the 200 million cells/mL operating point is an approximation using the upper end of the range the paper describes as "above 200 million cells/mL observed for two days at 8 to 10 vessel volumes per day".
6. A materials engineer's view (1): a hollow-fibre membrane stops cells but lets antibodies through
The hollow-fibre membrane used by Clincke and colleagues had 50 polysulfone fibres, a 0.2 µm pore size, a 1 mm inner diameter and a membrane area of 850 cm²Sourced. A 0.2 µm pore is far smaller than a cell, so cells are stopped. Even so, some of the product (antibody) was retained by the membrane, more markedly with TFFSourced — and there lies the membrane design problem.
- What to stop: cells and cell debris
- What to pass: antibody (the product) and spent medium containing waste
- What gets in the way: rising viscosity at high cell density. Above 200 million cells/mL, the pressure in the recirculation loop reached 1 barSourced
Once a layer of cells, debris and protein forms on the membrane surface, the effective pore size shrinks and the membrane presumably starts to hold back antibody as well (our commentary). ATF's backflushing by alternating flow is said to "help prevent clogging"Sourced; the idea is to strip that layer away with the flow.
From the materials side, the question is how to combine the following (our commentary).
- Pore size distribution: stopping cells reliably while keeping antibody transmission high for long periods
- Surface hydrophilicity and low protein adsorption: a surface on which layers form less readily
- Long-term mechanical durability: in ATF the flow reverses about once a minute, for weeks on end
- Manufacture and sterilisation for single use: Clincke and colleagues rinsed the membranes with water and autoclaved them before useSourced
This is an area where the know-how built up in hollow-fibre membranes for water treatment and haemodialysis carries straight over. But the combination of demands — "keep passing the product for weeks, in a viscous cell suspension" — can be read as unique to cell culture.
7. ICH Q13 and where the FDA and Japan stand
(1) The scope of ICH Q13
ICH Q13, "Continuous Manufacturing of Drug Substances and Drug Products", sets its scope as followsSourced.
- It applies to continuous manufacturing of drug substances and drug products for chemical entities and therapeutic proteins, both for new products and for converting existing products from batch to continuous manufacturing
- It can also apply to individual unit operations (for example, perfusion cell culture), but the guideline focuses on the integrated aspects of systems in which two or more unit operations are directly connected
- The batch size in continuous manufacturing can be defined by the quantity of output, the quantity of input, or run time at a defined mass flow rate
- Understanding the residence time distribution (RTD) is considered important for understanding process dynamics
(2) Annex III on therapeutic proteins
- Cell age: bioreactors in continuous manufacturing may run considerably longer than in batch. Limits on in vitro cell age established in batch mode may not apply to continuous runs under different conditionsSourced
- Adventitious agents: because culture runs for long periods and the harvest is processed continuously, the design must show that all the cultures used to make a given drug substance batch are acceptableSourced
- Single-use components: their integrity during use should be assured. Connections (tube welds, connectors) and components may be subject to prolonged use or frequent replacement, and should be assessed as a contamination riskSourced
- Filtration: because filtration runs longer and throughput per area and the number of filter changes may rise, there should be arrangements for changing and integrity-testing filters without stopping the processSourced
- Run time: should take into account the control of adventitious agents and the lifetime of resins and membranesSourced
(3) Uptake in the US and Japan
The US FDA published Q13 as final guidance (from its Center for Drug Evaluation and Research, CDER) in March 2023Sourced. In Japan, the Ministry of Health, Labour and Welfare (MHLW) circulated it as the "Guideline on Continuous Manufacturing of Drug Substances and Drug Products" in a notice from the director of the Pharmaceutical Evaluation Division dated 31 May 2023 (PSEHB/PED Notification No. 0531-1)Sourced.
Material presented by a PMDA reviewer at a training course in September 2022 (explicitly labelled as the speaker's personal view) lists products approved in Japan using continuous manufacturing technology, and all of the listed products were tabletsSourced. No example of a biopharmaceutical approved with continuous manufacturing that links perfusion culture directly to purification could be confirmed in primary sources within this article's research.
8. A materials engineer's view (2): a 30-day run is a life test for materials
In the Bunnak model, one fed-batch culture lasted 12 days and a perfusion culture 30 daysSourced. The time the same culture bags, tubing, sensors and membranes spend in contact with culture fluid becomes 2.5 times longerOur calculation (30 ÷ 12).
What ICH Q13 names and asks to be assessed is precisely the lifetime of materialsSourced.
- Membrane fouling and sensor fouling (stated as added risks of long runs)
- Number of uses of resins, and membrane lifetime (stated as considerations for run time)
- Integrity of single-use connections (prolonged use and frequent replacement)
Summarising the properties this asks of materials gives the following (our commentary).
- Conditions for assessing leachables become tougher. The ICH Q3E draft, covered in our explainer on single-use systems, asks that extraction studies be designed around a worst case that includes "the longest contact time"Sourced. If contact time grows, the assumptions behind the assessment change too
- Repeated mechanical loading. If ATF reverses about once a minute for 30 days, simple arithmetic gives about 43,000 reversalsOur calculation (30 days × 24 hours × 60 minutes). Fatigue of membranes, diaphragms and connections comes into question
- Design for replacement without stopping. As Q13 asks for filters, components whose shape and connection method allow them to be changed and tested without stopping the process become valuable in themselves
Parts that in batch manufacturing only had to "work once" must, in continuous manufacturing, "not degrade for weeks" — and that gap becomes a new specification for materials makers.
9. What could not be confirmed, and remaining challenges
(1) Approvals of continuously manufactured biopharmaceuticals
No biopharmaceutical approved with continuous manufacturing that links perfusion culture directly to purification could be confirmed in primary sources within this article's research, so none is named.
(2) Cell retention devices at large scale
This article has not confirmed any primary source showing how far the large-scale challenges Nienow's review pointed out (clogging, throughput, shear)Sourced have since been solved.
(3) Reducing the environmental impact of perfusion processes
The finding that extending the pooling period for purification to eight days could make the perfusion process the more environmentally friendly one comes from a sensitivity analysis of the Bunnak modelSourced; this article has not confirmed any verification in a real plantNot yet confirmed.
(4) Titre and cell density values depend on each study's conditions
The values in Figs. 2 and 3 come from different studies, cell lines and devices, or are model assumptions. They do not generalise about which mode is better.
- Fed-batch culture is the standard for large-scale antibody production, with 10 to 14 days and 5 to 30 million cells/mL described as typicalSourced
- Perfusion culture has reached 214 million cells/mL, with the limits set by membrane capacity, viscosity and pCO2Sourced
- Perfusion loses on concentration but makes about 9.6 times more per litre of tank. The model assumes the same annual output from a tank about one-eighth the sizeOur calculation
- If downstream steps run more often, the water and energy for cleaning shift thereSourced
- ICH Q13 was adopted in 2022, and taken up by the US in March 2023 and by Japan in May 2023Sourced
- Continuous manufacturing is a production mode that tests the "lifetime" of membranes, resins and single-use components (our commentary)
10. Glossary
- Batch culture
- A mode in which medium is added at the start, nothing is added during culture, and everything is harvested at the end.
- Fed-batch culture
- A mode in which concentrated nutrient solution is added during the run. The volume increases.
- Perfusion culture
- A mode in which medium is continuously exchanged while the cells stay in the tank.
- Continuous manufacturing
- A production mode that links two or more unit operations directly, with material going in and product coming out continuously.
- Perfusion rate (vessel volumes/day)
- The volume of medium exchanged per day, expressed as a multiple of the tank's liquid volume.
- Bleed
- Removing part of the culture, cells included, to keep cell density constant.
- TFF
- Tangential flow filtration. Liquid flows along the membrane surface, filtering while limiting clogging.
- ATF
- Alternating tangential flow. A diaphragm pump reverses the flow direction back and forth, backflushing the membrane.
- Hollow-fibre membrane
- A membrane in the form of thin hollow fibres, used in bundles, with liquid passing through the walls.
- Titre
- The concentration of product (such as antibody) in the culture fluid, expressed in g/L.
- Specific productivity
- The amount of product one cell makes per day (pg/cell/day).
- Residence time distribution (RTD)
- The distribution of times material takes to pass through a process. A basis for controlling continuous manufacturing.
- In vitro cell age
- Culture time or number of generations since thawing from the cell bank. Controlled by setting an upper limit.
- ICH Q13
- The international guideline on continuous manufacturing of drug substances and drug products (Step 4 in 2022).
11. References
- Li F, et al. (Genentech) "Cell culture processes for monoclonal antibody production", mAbs 2(5):466-479 (2010) — pmc.ncbi.nlm.nih.gov
- Clincke MF, et al. (KTH Royal Institute of Technology, Sweden) "Very high density of CHO cells in perfusion by ATF or TFF in WAVE bioreactor. Part I. Effect of the cell density on the process", Biotechnology Progress 29(3):754-767 (2013) — pmc.ncbi.nlm.nih.gov
- Lemire L, et al. (Polytechnique Montréal / National Research Council Canada) "Scale-up of a monoclonal antibody CHO fed-batch production in stirred tank bioreactors: Effect of hydrodynamic conditions and feeding regimen", Biotechnology Progress (January-February 2026 issue, doi:10.1002/btpr.70073) — pmc.ncbi.nlm.nih.gov
- ICH "Continuous Manufacturing of Drug Substances and Drug Products Q13", final version (adopted 16 November 2022) — database.ich.org
- Bunnak P, et al. (University College London and others) "Life-cycle and cost of goods assessment of fed-batch and perfusion-based manufacturing processes for mAbs", Biotechnology Progress 32(5):1324-1335 (2016) — pmc.ncbi.nlm.nih.gov
- U.S. FDA "Q13 Continuous Manufacturing of Drug Substances and Drug Products", guidance page (March 2023, final) — fda.gov
- Director, Pharmaceutical Evaluation Division, Pharmaceutical Safety and Environmental Health Bureau, Ministry of Health, Labour and Welfare "Guideline on Continuous Manufacturing of Drug Substances and Drug Products", PSEHB/PED Notification No. 0531-1 (31 May 2023, posted by PMDA; in Japanese) — pmda.go.jp
- PMDA (Kyoko Sakurai, Office of New Drug IV) "Points to note in the regulations surrounding continuous manufacturing of biopharmaceuticals", Regulatory Science Expert Training material (9 September 2022; explicitly the speaker's personal view; in Japanese) — pmda.go.jp
- Nienow AW "Reactor engineering in large scale animal cell culture", Cytotechnology 50:9-33 (2006) — pmc.ncbi.nlm.nih.gov
- ICH "Guideline for Extractables and Leachables Q3E", draft (Step 2, 1 August 2025) — database.ich.org
12. Claim-to-source audit
| Claim in the text | Basis | Label |
|---|---|---|
| That typical production cell lines have a specific productivity of 20 to 70 pg/cell/day and peak viable cell densities of 5 to 30 million cells/mL in 10- to 14-day fed-batch culture; that media development consists of a basal medium and a concentrated feed; and that frequent or continuous feeding can reduce by-products but has not been favoured in large-scale manufacturing because of operational complexity and validation cost | Li et al. 2010 review — Reference 1 https://pmc.ncbi.nlm.nih.gov/articles/PMC2958569/ | Sourced |
| That batch and fed-batch are the most widely used and interest in perfusion is growing; the advantages of perfusion (longer healthy culture, high density, short product residence in the tank, smaller tanks with less cleaning in place, and the possibility of single-use tanks); the set-up of a 10 L culture bag with hollow-fibre membranes; perfusion rates of 1 to 10 vessel volumes/day and stabilisation by bleeding; stable at 20 to 35 million cells/mL, held at 90 to 130 million cells/mL for more than two weeks, maxima of 214 million (TFF) and 132 million (ATF) cells/mL and the limiting factors for each; 8 to 10 vessel volumes/day, a pressure of 1 bar and pCO2 of 31 kPa above 200 million cells/mL; the recommendation of below 146 million cells/mL; the options for single-use cell separation devices and why TFF and ATF were chosen; how TFF and ATF work (tangential flow, diaphragm pump, a cycle of about one minute, backflushing, lower shear); the membrane specifications (50 polysulfone fibres, 0.2 µm, 1 mm inner diameter, 60 cm, 850 cm², rinsed with water and autoclaved); that pressurising the tank to 0.02 to 0.03 bar restarted alternating flow in ATF; that part of the product was retained by the membrane, more markedly with TFF; and that the perfusion rate was scaled to cell density | Clincke et al. 2013 original paper — Reference 2 https://pmc.ncbi.nlm.nih.gov/articles/PMC3752962/ | Sourced |
| That in fed-batch culture rising volume changes mixing conditions, and that limiting the volume rise with a concentrated feed makes scale-up easier | Lemire et al. 2026 — Reference 3 https://pmc.ncbi.nlm.nih.gov/articles/PMC12908111/ | Sourced |
| That Q13 reached Step 2 on 27 July 2021 and was adopted at Step 4 on 16 November 2022; its scope (chemical entities and therapeutic proteins, new products and conversions); that it can apply to unit operations such as perfusion cell culture but focuses on direct connection of two or more; the definition of batch size (output, input, run time); RTD; and in Annex III, the example system, cell age, adventitious agents, integrity of single-use components and assessment of connections, filter changes and integrity testing, consideration of resin and membrane lifetime in run time, and the risks of long runs including membrane and sensor fouling | ICH Q13 final version — Reference 4 https://database.ich.org/sites/default/files/ICH_Q13_Step4_Guideline_2022_1116.pdf | Sourced |
| That fed-batch became the standard because of scalability up to 20,000 L, robustness and high volumetric productivity; that 0.5 to 2 vessel volumes/day have been reported for industrial cell lines; the model assumptions (375 L and 47 L, 5 g/L and 2 g/L, 12 and 30 days, 5-day ramp-up, 2 vessel volumes/day, 21 and 9 runs, 21 and 53 purifications, purification yields of 72% and 76%); that perfusion with 4-day pooling had almost the same cost of goods but used 35% more water and 17% more energy and emitted 17% more CO2; that CIP and SIP were more than 85% of water use; and that extending to 8 days could make perfusion more environmentally friendly | Bunnak et al. 2016 — Reference 5 https://pmc.ncbi.nlm.nih.gov/articles/PMC5082523/ | Sourced |
| That the FDA published Q13 as final guidance (CDER) in March 2023 | FDA guidance page — Reference 6 https://www.fda.gov/regulatory-information/search-fda-guidance-documents/q13-continuous-manufacturing-drug-substances-and-drug-products | Sourced |
| That the MHLW circulated the "Guideline on Continuous Manufacturing of Drug Substances and Drug Products" by a division director's notice dated 31 May 2023 (PSEHB/PED Notification No. 0531-1) | MHLW notice (posted by PMDA) — Reference 7 https://www.pmda.go.jp/files/000252834.pdf | Sourced |
| That the products listed in the September 2022 training material as approved in Japan using continuous manufacturing technology were all tablets (material explicitly labelled as the speaker's personal view) | PMDA training material 2022 — Reference 8 https://www.pmda.go.jp/files/000248673.pdf | Sourced |
| That at scales above 10,000 L, major questions remain over clogging, throughput and shear sensitivity in cell retention and recirculation devices | Nienow 2006 review (abstract) — Reference 9 https://pmc.ncbi.nlm.nih.gov/articles/PMC3476006/ | Sourced |
| That extraction studies should be designed around a worst case including the longest contact time | ICH Q3E draft — Reference 10 https://database.ich.org/sites/default/files/ICH_Q3E_EWG_Step2_DraftGuideline_2025_0704.pdf | Sourced |
| Output per litre of tank per day (fed-batch about 0.42 g, perfusion 4 g, about 9.6 times); tank volume ratio about 8.0 times; perfusion rate per cell of 50 pL/cell/day (two points); contact time 2.5 times; about 43,000 ATF reversals (30 days) | Our calculation, using the Bunnak model's assumptions, the Clincke operating points and a cycle of about one minute, and ignoring ramp-up, bleed losses and purification yield | Our calculation |
| Verification in a real plant of the environmental improvement from extending the pooling period to 8 days | A result of the Bunnak model's sensitivity analysis; this article has not confirmed verification in a real planthttps://pmc.ncbi.nlm.nih.gov/articles/PMC5082523/ | Not yet confirmed |
| Biopharmaceuticals approved with continuous manufacturing linking perfusion culture directly to purification; the current state of large-scale cell retention challenges | Not stated because they could not be confirmed in primary sources within this article's research (commentary) | Commentary |
| The definition of batch culture (nothing added during the run); the summary of what goes in and out in the four modes; the reading that the perfusion ceiling was set by engineering; the reading that perfusion loses on concentration but wins on output per tank; the framing that more downstream runs shift the burden; the idea that a layer on the membrane surface starts to hold back antibody, and the summary of properties needed in membrane materials; the framing that continuous manufacturing tests material lifetime; and the reading that hollow-fibre know-how from water treatment and dialysis carries over | This article's summary and commentary based on published content. Not views expressed by the papers or institutions | Commentary |
| That Figs. 1 to 4 and Fig. 6 are explanatory drawings rather than real equipment or measurements, and that the hero image and Fig. 5 are AI-generated images | Our note (commentary) | Commentary |
Last updated 23 September 2026. Sources are limited to primary material (peer-reviewed papers, ICH guidelines, and published material from the US FDA, the MHLW and PMDA). Biopharmaceuticals approved with continuous manufacturing linking perfusion culture directly to purification, and the current state of large-scale cell retention challenges, are not stated here because no published primary source could be confirmed. Some figures in Sections 4, 5 and 8 are this article's calculations based on stated assumptions. All figures are explanatory concept graphics. Figs. 1 to 4 and Fig. 6 are vector drawings, and the hero image and Fig. 5 are AI-generated images; none of them shows real equipment, a product or a measurement.