OR WAIT null SECS
Valerie Pferdeort is senior process development engineer at Amgen, Inc.
Kirsten Hayda is a process development engineer at Amgen, Inc.
Maria Wik is principal process development engineer at Amgen, Inc.
Wet testing of microbial fermenters allows for a greater understanding of the equipment's capabilities.
Improving speed and quality, and reducing the cost of technology transfers is becoming increasingly important in the biopharmaceutical industry. Microbial fermentation processes have challenging equipment requirements, such as high heat and oxygen transfer rates, and ensuring minimal and consistent medium concentration changes resulting from condensate gain during steam-in-place and evaporative losses during media hold and fermentation. This article presents a platform approach to equipment characterization for microbial fermenters. These tests are used to gain understanding of the equipment capabilities before starting process runs. The platform presented here outlines a wet testing approach that has been successfully executed at pilot scale (300-L) for a dual-purpose reactor (used for both mammalian cell culture and microbial fermentation) and in a large-scale (10,000-L) fermenter. Comprehensive equipment characterization using a platform approach streamlines the technology transfer and maximizes success rates during process runs.
Eden Biodesign Ltd.
An optimized and fast-growing microbial culture is a dynamic process, and ensuring that the equipment is suitable for meeting these challenges is necessary for project success. These cultures tend to have a high oxygen demand that the fermenter must be able to meet to sustain cell growth and desired productivity. The culture also generates significant amounts of heat that the fermenter must remove to maintain temperature control. To ensure success during actual process runs, it is important to understand oxygen supply and heat removal capabilities before operating the process in the fermenter. Water-based tests were developed to characterize heat removal and oxygen supply capacity to understand if any equipment modifications would be required to meet process requirements. This testing should be performed early in technology transfer facility fit activities to allow time for equipment modifications, if needed.
Microbial media generally are batched into the fermenter and then steam sterilized-in-place (SIP). The SIP cycle is automated to ensure control within the required temperature range. Temperatures outside the acceptable range could result in insufficiently sterilized medium or overheated medium. Additionally, evaporation or condensation can occur during the SIP cycle. The change in fermenter weight should be characterized or eliminated during SIP operations to ensure the target medium concentration for optimal growth is achieved. After sterilization, the fermentation medium often is held at specified conditions until inoculation. Determining the evaporation rate during the medium hold is essential for achieving the correct medium concentration at the time of inoculation. To compensate for the evaporation rate, additional water can be added during media preparation, and adjustments to the air flow rate can be made to minimize evaporative losses. Water-based testing was developed to understand parameters around media SIP and hold conditions to ensure the target starting media concentration would be met.
Table 1 lists the fermenter dimensions and operational settings for both pilot and large scales.
Table 1. Pilot- and large-scale fermenter characteristics. The agitation setting was scaled between the pilot- and large-scale fermenters by maintaining the ratio of agitator power draw (P) per unit volume (V).
Heat Removal Characterization
To maintain constant temperature control the fermenter must have sufficient heat removal capacity to remove metabolic heat generated by the cells and mechanical heat generated by the agitator. A common heat transfer scale-up challenge stems from the fact that the relative heat transfer area decreases with increasing fermenter scale. The heat transfer rate (HTR) can be calculated using a simple heat transfer equation:
in which m is the mass of water in the fermenter (kg), Cp is the heat capacity of water at 37 °C (4.181 kJ/kg/°C),1 T is the temperature of water in the fermenter (°C), t is time (h), and dT/dt is the rate of temperature change as the slope of water temperature versus time curve.
Strategies for maintaining temperature control involve equipment modifications such as: lowering chilled water or glycol temperature to increase the driving force for heat removal (dT/dt), ensuring jacket and coils have minimum resistance, and adding internal cooling coils to increase heat transfer surface area.
The fermenters were filled to different target weights and heated to approximately 75 °C. The temperature set point was changed to 10 °C, which resulted in a 100% cooling output on the temperature control loop. The rate of temperature change was estimated by the slope of the linear portion of the temperature-versus-time curve from 41 °C to 34 °C. This range encompasses the process temperature set point of 37 °C. The HTR was then calculated using Equation 1.
Oxygen Supply Characterization
The fermenter must have sufficient oxygen transfer to supply enough oxygen to the culture to support cell respiration. Oxygen transfer rates (OTR) are lower as the scale increases because of potentially lower power per volume input, which causes a common scale-up problem. Process modifications can be explored to decrease growth rate and oxygen demand; however, sufficient oxygen supply may be achieved with modifications such as manipulating aeration (air supplemented with oxygen, gas flow rate), agitation (impeller design, number, and location; agitator speed), headspace pressure, and baffles.
The above equipment settings can influence the OTR in a fermenter. These settings can be optimized to increase the oxygen driving force or the mass transfer coefficient (kLa) of the reactor. The OTR can be calculated using the following equation:
in which kL is the mass transfer coefficient (cm/h), a is the gas/liquid interface area per liquid volume (cm2 /cm3), C* is the saturated dissolved oxygen concentration (mmoles/dm3), and CL is the concentration of dissolved oxygen in the fermenter (mmoles/dm3).
For process transfer and scale-up, kLa is a useful parameter because it helps determine whether a reactor can supply oxygen at a nonlimiting rate. Several methods can be used to estimate kLa. During fermentation, a steady state mass balance calculation can be used to equate the oxygen uptake rate of the culture, as measured by a mass spectrometer, to the OTR. Additionally, chemicals such as sodium sulphite can be used to estimate kLa based on the rate of sodium sulphite oxidation, which is equivalent to the oxygen transfer rate.2
Dynamic pressure and gassing methods also can be used to estimate kLa. A dynamic pressure method can be used by measuring the change in dissolved oxygen (DO) concentration after a small change in system pressure. The dynamic gassing method measures the change in DO concentration versus time for predetermined step changes in the sparge gas oxygen concentration. These dynamic methods can be problematic in microbial systems with high kLa values because the oxygen concentration changes so rapidly that the DO sensor response becomes limiting. However, more accurate kLa can be estimated by accounting for the DO sensor kinetics.2,3
Media Steaming Characterization
The fermenter must have an SIP cycle that controls temperature to ensure efficient sterilization for bioburden control, but also avoids exposing the media components to excessive temperatures that could cause nutrient degradation. Evaporation or condensation can occur during the SIP cycle. It is desirable to characterize (or eliminate) the change in fermenter weight during SIP operations to ensure the target starting medium concentration is achieved. If significant evaporation or condensation occurs, this should be compensated for in the amount of water that is used to batch the medium. Scale effects of the heat-up and cool-down times of the medium as part of the SIP cycle also should be characterized because these durations can significantly change with scale and may have an unexpected effect on media performance.
In this study, the fermenters were filled with water and three SIP cycles were performed in series with full cooling between experiments, to understand variability related to evaporation or condensation.
Evaporative Losses During Media Hold Characterization
To allow for schedule flexibility during production and to ensure a robust process is implemented, the sterilized media must be held for a period of time before inoculation. If the medium is agitated and air is applied to the fermenter, evaporation may occur during this hold period even if a condenser is installed on the gas outlet line of the fermenter. Determining the evaporation rate is essential for achieving the correct medium concentration at the time of inoculation. To compensate for the evaporation rate, additional water can be added during medium preparation; however, this necessitates a fixed hold time for every batch. Alternately, adjustments to the air flow rate can be made to minimize evaporative losses.
At the 300-L pilot scale, water was held under process conditions (temperature, air flow rate, volume, agitation, pressure). Holding the medium at process conditions is advantageous because it eliminates the need for any equilibration time before inoculation. Medium hold data from historical processes in the large scale fermenter (10,000-L) indicated significant weight loss during hold periods. Therefore, for large-scale testing, the airflow rate was decreased from the process set point to a pre-inoculation hold-specific set point. Fermenter weight was continually monitored and used to calculate the evaporation rate at both scales. Weight was plotted against hold time and the slope was used as an estimate for evaporative loss occurring during the media hold.
Heat Removal Characterization
The oxygen uptake rate (OUR) for the culture in question is approximately 300 mmol/L/h. The heat generated by the cells can be estimated using the common method shown in Equation 3:
in which H M is metabolic heat (kJ/kg/h), OUR is the oxygen uptake rate (mmol/kg/h), and 5.2 x 10-4 is a constant.
Per Equation 3, the heat generated by the cells is approximately 150 kJ/kg/h. Typically, the next largest contributor of heat to the fermenter is the agitator. This was measured at large scale using the agitator current draw and estimated to be approximately 18 kJ/kg/h. As such, it is considered a small contributor to the overall heat load, and the HTR required for the fermenters is estimated as 150 kJ/kg/h.
Figure 1 shows the HTR calculated using Equation 1 for each fermenter weight tested at the pilot and large scale, respectively. The calculated HTR at pilot scale varied between 555 and 750 kJ/kg/h with the fermenter weights from 150 to 300 kg, and at large scale the HTR varied between 200 and 235 kJ/kg/h with the fermenter weights between 5,000 and 10,000 kg; therefore both fermenters had sufficient HTR capacity to support the process need of 150 kJ/kg/h.
Figure 1. Heat transfer rate at pilot and large scales.
For both fermenters, only the bottom impeller was submerged at the lowest test weight. The maximum heat transfer rate was observed at 200 and 7,000 kg, the pilot-scale and large-scale, respectively. The upper impeller is partially or just barely submerged at these weights. The lowest HTR was observed at the maximum working volume of 300 kg for the pilot scale and 10,000 kg for the large-scale fermenter. The lower HTR at the maximum working volume is expected and likely caused by the decreased surface area to volume ratio at the higher volume. Another factor that may have contributed to the heat removal rates was the air sparge rate, which was kept constant for all fermenter weights; therefore, the volume of sparge air per volume of water (VVM) decreased as the fermenter weight increased (shown in Figure 1). Introducing gas into the liquid likely reduces the apparent density of the fluid and because of gas hold-up, the liquid comes in contact with additional cooling surface.
Oxygen Supply Characterization
To estimate the kLa of the pilot scale reactor at processing conditions, the dynamic gassing method was explored. A procedure was drafted to measure the kLa while accounting for the DO sensor lag. This method development is on-going.
Media Steaming Characterization
Table 2 summarizes the gain or loss observed during SIP cycles at pilot and large scale. For the large scale, no experiments were performed because historical SIP data showed minimal and consistent gain and losses. Testing performed at pilot scale demonstrated that the weight change during SIP also was consistent and minimal. Therefore, it was concluded that the water amount used to batch medium did not need to be adjusted to account for any considerable gain or loss during the SIP operation. From the evaluation of data from SIP cycles done on actual media at the pilot scale, it can be concluded that media and water behave similarly and that water is a good surrogate for testing purposes before starting actual production runs.
Table 2. Sterilize-in-place (SIP) gain or loss results. Pilot scale test results are presented. For large scale, historical SIP data are presented.
Evaporative Losses During Media Hold Characterization
Figure 2 shows the evaporation rate as observed at the pilot scale. The test was performed at process conditions. Tests 1 and 2 show very consistent results with an average evaporative loss of approximately 0.12% of starting fermenter weight/h. Media hold conditions or the amount of water used to prepare the medium should be further evaluated to ensure the target starting media concentration is met.
Figure 2. Pilot-scale evaporative losses. Tests 1 and 2 were performed with the condenser on the fermenter enabled, and test 3 was performed with the condenser disabled. The condenser reduces the evaporative loss by almost a factor of two.
Large-scale data demonstrated that the need for any water adjustments to achieve the target media concentration is eliminated when the airflow rate is decreased during the media hold. With the airflow rate at 10% of the processing set point, the evaporative loss was negligible. Over approximately 40 h, the reactor weight dropped by approximately 0.02% of starting weight (Figure 3), which is within the calibration tolerance of the fermenter load cells. At the end of the hold, the air flow rate was increased to the processing set point. The data indicated that no additional equilibration time was needed to achieve DO saturation.
Figure 3. Large-scale evaporative losses.
The wet testing presented in this article represents a platform approach to characterizing microbial fermenters. These tests can be used to gain an understanding of the equipment capabilities before starting actual process runs and further ensure project timelines and success criteria are met. Measuring the heat removal and oxygen transfer capability of a fermenter helps ensure a reactor meets process needs. Water-based testing can be performed to help develop appropriate media SIP and hold conditions, thus ensuring the target starting media concentration is met.
These equipment characterization studies are not required to be performed for every new product that is introduced into a facility as long as major equipment changes have not occurred (e.g., SIP cycles). However, it is useful to confirm that there have been no shifts in equipment capability by periodically repeating these tests.
The author would like to thank Greg Naugle and Sushil Abraham for reviewing the manuscript, and Arun Tholudur for troubleshooting support during testing and useful technical insights, Justin Bingham for providing data from large scale testing, and Colter Davidson, Jaime Foster, and Tom Folger for testing at the pilot scale.
KIRSTEN HAYDA is a process development engineer, MARIA WIK is principal process development engineer, and VALERIE PFERDEORT is senior process development engineer, all at Amgen, Inc., Longmont, CO, 303.401.5156, email@example.com
1. Geankoplis, Christy. Transport Processes and Unit Operations. 3rd ed. Englewood Cliffs, New Jersey: Prentice Hall P T R; 1993.
2. Standbury, Whitaker, Hall. Principles of Fermentation Technology. 2nd ed. Burlington, MA: Butterworth–Heinemann; 1995.
3. Badino Jr, Facciotti, and Schmidell. Improving kLa determination in fungal fermentation, taking into account electrode response time. J Chem Technol Biotechnol. 2000; 75:469-474.
4. Bailey and Ollis. Biochemical Engineering Fundamentals. 2nd ed, Boston, MA: McGraw-Hill; 1986.