Comparing Fed-Batch Cell Culture Performances of Stainless Steel and Disposable Bioreactors

January 1, 2011

A case study to compare the performances of several types of mixing in disposable bags with stainless steel bioreactors.


Single-use bioreactors are commonly used for seeding stainless steel bioreactors or for producing material. The profitability of this equipment has been well demonstrated for more than a decade, but few data on their scalability are published. In May 2010, Merck Serono Biodevelopment began a study to evaluate the performances of disposable bioreactors. Because different technologies were available, this study compared performances of several types of mixing. The evaluation was performed both for seeding application and for clinical material production. This study featured 20–50 L disposable bioreactors with gas flow-rate scaled-down from seeding and production bioreactors. A fed-batch process producing a highly glycosylated molecule was performed twice in four types of disposable bioreactors. The quality of the molecule together with molecule concentration and cell growth were compared among the three single-use technologies. These process performances were also compared to 250 L and 1,250 L bioreactors and to a 3.6 L glass development bioreactor. These comparisons allowed Merck Serono Biodevelopment to conclude on their uses and on the scalability (up and down) of these disposable systems. This study was completed by a characterization of liquid–liquid and gas–liquid transfers inside each disposable bioreactor to estimate their potential in terms of cell culture.

To decrease sterilization, cleaning workload, and reduce costs, the disposables market has been diversified to meet the needs of the biotechnology industry. At Merck Serono Biodevelopment (MSB), bags, tubing, and disposable cell culture containers commonly are used in production processes. Hybrid processes with stainless steel bioreactors, however, remain because a lot of factors have to be considered before implementing such technologies. Besides supplier support and quality, one critical point is the scalability of these bags. Of course it is easier when you design a new plant, but when you have existing stainless steel manufacturing bioreactors and you want to increase your capacity and produce material in disposable bioreactors, comparability and scalability between bioreactors could be a problem. From May to September 2010, MSB performed a study to evaluate and compare performances of different disposable bioreactors. Various equipment models are available on the market, so the options were sorted based on maximum working volume, mixing technology, current good manufacturing practice compliance, availability, data published, supplier experience, and internal requirements. To test different mixing principles, the following technologies were evaluated: ATMI-Pierre Guerin single-use bioreactors; orbital shaking and stirred mixing, both from Sartorius. Processes performed with this equipment began with a growth phase to evaluate bioreactor seeding performances, followed by a fed-batch process producing a high glycosylated antibody-like molecule, to evaluate the impact of culture in disposable bioreactors on molecule quality. Performances were compared to process development bioreactors and manufacturing bioreactors. The evaluation finished with a characterization of transfers by KLa, mixing time measurements, and temperature mapping.

(Merck Serono Biodevelopment)


Cell culture was performed twice in each disposable bioreactor under the same conditions.

Cell Culture

A vial of Chinese hamster ovary (CHO) cells was thawed in commercial medium in a static container. Cells were expanded every two or three days in agitated containers until reaching 8.6.109 cells.

Cell Growth in Disposable Bioreactors

A sterility test was performed at 37 °C in each disposable bag two days before inoculation. The cells were then inoculated in parallel in all disposable bioreactors at the minimum working volume provided by the supplier. Growth was then observed for two days. Working volume was increased to the maximum working volume at the same cell density between bioreactors. Cell growth at maximum working volume lasted three days.

Temperature was regulated at 37 °C. Stirring speed was fixed by the supplier.

Oxygen was regulated at 50% only with oxygen at a flow rate scaled-down from MSB 250 L seeding bioreactor for paddle and stirred mixing bioreactors and at a flow rate fixed by the supplier and MSB according to its experiences with the orbital shaking bioreactor.

Production In Disposable Bioreactors

In the same bag, cell suspension was partially removed and replaced by fresh medium to keep same seeding density between bioreactors. Cells were maintained at 37 °C for a few days, then a temperature switch was performed and production temperature was applied until production day 7.

Oxygen was regulated at 50% with air and oxygen at a flow rate scaled-down from MSB 1,250 L production bioreactor for paddle and stirred mixing bioreactors and at a flow rate fixed by supplier and MSB according to their experiences with the orbital shaking bioreactor. The pH was maintained between 6.8 and 7.2 with sodium hydroxide 1N and CO2 with a flow rate fixed by supplier.

Agitation speed was fixed by the supplier for the paddle bioreactor. Agitation speed for stirred and orbital-shaking bioreactors was fixed by the supplier and increased by MSB to avoid large oxygen variations.

Feed was added twice at the same amount per L, and sampling was performed every day at 1/200 of working volume.

In-Process and Quality Controls

Every day, pH was checked with an external pH-meter and corrected if the difference was above 0.05 pH unit; cell density and viability were measured. During the production process, biochemistry analyses and molecule titration were performed in addition to previous in-process controls. At the end of the process, a capture was performed to check quality attributes (e.g., glycosylation, aggregates).

Bioreactor Characterization

Oxygen transfer coefficient, KLa, was measured at maximum working volume in water at 37 °C by sparging air at the working flow rate applied during production process and at the maximum flow rate allowed by the equipment.

Mixing time was measured at maximum working volume in phosphate saline buffer between pH 6.8 and 7.4 at 37 °C by injecting a small amount of sodium hydroxide 5N to increase the pH by 0.20–0.30 units. The pH was adjusted initial level with 8M acetic acid.

Temperature mapping was performed at maximum working volume in water with six calibrated temperature probes at different places inside each disposable bag. Temperature increase was followed between room temperature up to 37 °C and temperature decrease to 37 and 29 °C.


One of MSB's objectives is to provide grams of high quality recombinant proteins for preclinical and clinical studies. This production can be performed at different scales depending on needs. Thus, process performances in terms of cell growth, but especially protein concentration and quality, should be representative to MSB bioreactors.

Cell Growth During Cell Amplification Phase

Cell growth at minimum and maximum working volume allowed MSB to evaluate seeding performances of disposable bioreactors.

Cell growth was compared between disposable bioreactors and 250 L production seeding bioreactor in terms of doubling time or population doubling level (Figure 1).

Figure 1. Population doubling level (PDL) from inoculation to production phase. For disposable bioreactors, it represents growth at minimum working volume (from working days 0 to 2) and at maximum working volume (from working days 3 to 5). Date for seeding bioreactors were from 1,250 L seeding bioreactors presented later. Data are the means from two bioreactors, with errors bars representing minimum and maximum levels.

Linear regressions of these curves gave cell doubling time (see Table 1).

Table 1. Regression coefficients and slopes of Figure 1 curves

Regression coefficients of all curves were above 0.99; therefore, cells exhibited a regular growth either at minimum working volume or at maximum working volume, whichever bioreactors were used.

These results indicate that flexibility of disposable bioreactors as the minimum working volumes was between half and 1/5 of maximum working volumes. Doubling times were similar between stainless steel and disposable bioreactors. They are all compatible with production bioreactor seeding application.

Cell Growth During Production Phase

The production process includes a temperature switch to compare cell growth between bioreactors; therefore, an integral of viable cells (IVC) was plotted instead of cell density (Figure 2).

Figure 2. Integral of viable cells (IVC) in production phase at maximum working volume for disposable bioreactors, a 1,250 L MSB production bioreactor, and a 3.6 L process development bioreactor. Data are the means from two bioreactors, with errors bars representing minimum and maximum levels.

Cell growth in disposable bioreactors using a paddle and stirred mixing system was similar to cell growth in traditional bioreactors; however, cell growth in the orbital shaking bioreactor was lower.

The variability between runs was quite low, even when runs were carried out with different cell bank vials and cell amplifications. Nevertheless, it is not relevant to conclude on robustness of disposable bioreactors as only two runs were performed.

Viability Profile During Production Phase

Cell viability profile (Figure 3) could be linked to shear stress inside the bioreactor. This parameter could reflect the impact of the mixing type.

Figure 3. Viability in production phase at the maximum working volume for disposable bioreactors, a 1,250 MSB production bioreactor, and a 3.6 L process development bioreactor. Data are the means from two bioreactors, with errors bars representing minimum and maximum levels.

Viability drop started at production day 6 for all bioreactors except the orbital shaking bioreactor. This particular bioreactor has a completely different mixing profile compared with the others because nothing except the bag comes into contact with cells. The agitation is external, while in other bioreactors mixing is generated by an internal element inside the bag. The higher viability observed in orbital shaking could also be linked to the lower IVC presented in Figure 2.

In a same way as IVC, viability profiles with paddle and stirred mixing bioreactors were similar to traditional bioreactors.

Disposable bioreactors with internal agitation seemed to reproduce cell growth and viability profiles better than traditional bioreactors.

Glucose Metabolism During Production Phase

Glucose is a key nutrient of cell metabolism and was followed (Figure 4) with a semi-automatic biochemistry analyzer.

Figure 4. Glucose concentration in production phase at the maximum working volume for disposable bioreactors, a 1,250 L MSB production bioreactor, and a 3.6 L process development bioreactor. Data are the means from two bioreactors, with errors bars representing minimum and maximum levels.

Despite different cell densities especially in the orbital shaking bioreactor, glucose concentrations were similar between all bioreactors throughout runs.

Target Protein Production

Concentration of the protein was measured by semi-automatic equipment with 10% of accuracy. Figure 5 shows protein concentration (molecule titer), and Figure 6 shows protein concentration measured during runs but taking into account the accuracy of the method.

Figure 5. Molecule titer in production phase at the maximum working volume for disposable bioreactors, a 1,250 L MSB production bioreactor, and a 3.6 L process development bioreactor. Data are the means from two bioreactors, with errors bars representing minimum and maximum levels.

Despite different mixing technologies, lower IVC, and higher viability in orbital shaking bioreactor, all disposable bioreactors demonstrated consistent protein concentration.

Figure 6. Recombinant protein titer in production phase at maximum working volume for disposable bioreactors, a 1,250 L MSB production bioreactor, and a 3.6 L process development bioreactor. Data are the means from two bioreactors, with errors bars representing minimum and maximum levels with 10% accuracy of the quality control method.

Their production performances seemed to be closer to process development bioreactors than production bioreactors.

When the accuracy of the quality control (QC) method was taken into account, protein concentration differences between bioreactors were less obvious.

Process variability taking into account QC accuracy seemed to be higher in disposable bioreactors than traditional bioreactors. This statement is not surprising as this process was better handled in traditional bioreactors than in new equipment.

Target Protein Quality Attributes

A recombinant protein was chosen for this project because its glycosylation pattern depended on cell culture conditions.

Quality attributes (Figure 7) are critical for protein function, pharmacokinetics variables such as half-life and bioavailability, immune response. They also may depend on the upstream part of the production process.

Figure 7. Recombinant protein quality for disposable bioreactors and 1250L MSB production bioreactor. HMW is for high molecular weight molecule (aggregates). High molecular weight molecules could not be compared to 1250L production bioreactors due to differences in sample treatment. Data are means from 2 bioreactors with errors bars representing minimum and maximum levels.

To generate a high quality molecule, a minimum specification for glycosylation was fixed (red line in Figure 7). All bioreactors performed beyond this specification: disposable bioreactors, whatever mixing type, did not have a critical impact on this quality attribute.

Figure 8. The pH measured after injecting sodium hydroxide in disposable bioreactors

This observation also is valid for the aggregates; no major differences were observed between disposable bioreactors.

For that process, cell growth and viability profiles of disposable bioreactors were close to traditional bioreactors. The productivity of disposable bioreactors was similar to MSB bioreactors. No impact of the mixing type was observed on molecule quality attributes. The equipment can be used either as a seeding bioreactor or as a production bioreactor.

Bioreactor Characterization

Bioreactor characterization allows evaluating mixing capacity and homogeneity inside the bioreactor. It simulates parameter evolutions while regulations are ongoing during a process.

Oxygen transfers. The oxygen transfer coefficient (KLa) was measured with air at the previous working flow rate and at the maximum flow rate allowed by the equipment. The air-flow rates applied during the process were representatives to 1,250 L production bioreactor flow rates.

Stirred and paddle mixing bioreactors have a sparger to introduce air and oxygen during dissolved oxygen regulation, while an orbital shaking bioreactor regulates oxygen concentration by injecting air and oxygen only in the headspace.

The agitation speed was the same as the one applied during the production process (see Table 2).

Table 2. Comparison of KLa measured at working flow rate and maximum flow rate between disposable bioreactors, a 1,250 L MSB production bioreactor, and a 3.6 L process development bioreactor.

The cells studied did not require a lot of oxygen. A low KLa (0.44h–1) was sufficient to run the process.

The air-flow rates applied in the orbital shaking bioreactor during the process were defined according to Merck Serono Biodevelopment and supplier experiences. These flow rates should have been decreased as KLa was above traditional bioreactors.

The stirred mixing disposable bioreactor obtained the maximum oxygen transfer. Despite air injection only through the headspace, the orbital shaking bioreactor had a KLa close to the stirred mixing bioreactor.

The paddle mixing bioreactor had a maximum KLa close to the traditional bioreactor.

Mixing time. Mixing time was also measured in phosphate saline buffer at the maximum working volume and the agitation speed applied during production process.

Mixing time was measured by injecting concentrated sodium hydroxide from the top of the bag (mixing down) and from the bottom of the bag (mixing up) (Table 3).

Table 3 : Comparison of mixing time measured in seconds between disposable bioreactors, 1,250 L MSB production bioreactor, and 3.6L process development bioreactor.

The paddle mixing system has classical probes with a transmitter, while orbital shaking and stirred mixing bioreactors have optical probes. Optical probes were more stable because they took a measure every five seconds. Bioreactors with disposable probes have a mixing time higher than traditional bioreactors because disposable probes have a higher response time compared to traditional ones. Classical probes were less stable and showed variations of around 0.03 pH unit even a few minutes after injection.

Mixing times on all disposable bioreactors were above traditional bioreactors. For the stirred mixing bioreactor, the mixing time was close to one minute, which is recommended for mammalian cell culture.1 Orbital shaking and paddle mixing were below two minutes.

Nevertheless, after sodium hydroxide injection, pH measured stayed or remained below or close to the final pH value. Even if the mixing time were longer than usual, cells would not be subjected to high feed or sodium hydroxide concentrations during a run.

Temperature mapping. Temperature mapping was performed because bags could be heated either by a jacket or by a blanket. Thus the temperature mapping allowed for assessing the efficiency and homogeneity of each heating system. It was measured in water at the maximum working volume and the agitation speed applied during production process. Temperature set points applied were from room temperature to 37 °C and from 37 °C to 29 °C (see Table 4).

Table 4. Comparison of temperature mapping measured between disposable bioreactors

Bags are in plastic, which does not conduct heat efficiently. Jackets provided higher temperature performances than blankets but also generated temperature gradients inside bags during heating and cooling activities.

The water inside the paddle mixing bioreactor jacket was from a water bath. When the system heats, the water bath is at 60 °C, so the temperature inside the bag can be 3°C above the set point. With this equipment, a 0.8 °C difference has been noticed between equipment temperature probe and calibrated Merck Serono temperature probes.


Merck Serono Biodevelopment performed a four-month study to evaluate disposable bioreactors. Different mixing principles were assessed in a disposable bag for seeding and fed-batch production processes. Process performances were similar to 250 L seeding bioreactors, 1,250 L production bioreactors, and 3.6 L process development bioreactors. Either cell growth, viability, molecule concentration, or quality were quite similar between bioreactors. Bioreactor characterization showed points to be improved in terms of probes and heating system.

During this project, MSB managed to set up a bioreactor in less than 30 minutes. Within three months, 11 runs were performed without any internal equipment validation or qualification. No contamination was observed despite filling and removing liquid several times in the bag. Disposable bioreactors were easy to manipulate and changed cell-culture process down time.

Aurore Poles Lahille is an assistant scientist, specialist new technologies, Richard C is a process specialist, Fisch S is USP production technician, Pedelaborde D is an equipment specialist, Gerby S is a trainee ENSTBB, Perrier V is head of the process development unit, Balbuena D is a process development and QC manager, Trieau R is head of the process improvement unit, Valognes L is the production manager, Peyret D is director of Merck Serono Biodevelopment, all at Merck Serono Biodevelopment, Martillac, France, +33 (0) 557 960 960.


1. Riba JP. Réacteurs enzymatiques et fermenteurs. Traité Génie des procédés, Technique de l'ingénieur. 1998.