Operational Performance of the Single-Use System
Figure 1. Operation comparison of a stainless steel ultrafiltration–diafiltration (SS UF–DF) system and a single-use (SU)
system
|
A typical UF–DF operation includes flow path and membrane installation, pre-use CIP, integrity test, UF, DF, product recovery,
and post-use CIP. Figure 1 compares a SS UF–DF system and a single-use UF–DF system. The implementation of the Millipore SU
Mobius FlexReady Solution for TFF offers some key advantages over the existing SS UF–DF system in terms of preparation, membrane
integrity testing, UF–DF general operation, critical parameters (feed flow rate, efficiency of mixing in retentate tank),
and product recovery.
Equipment preparation and set up: The disposable flow path is gamma irradiated and ready for use when it arrives. With the SS UF–DF system, cleaning occurs
in two steps: the UF–DF system and membrane. The gamma irradiated retentate bag and flow path have eliminated the system pre-use
and post-use CIP, rinse samples and protein swabs, and storage steps, however cassette cleaning must still be performed. The
disposable flow path is relatively easy to install, and typically can be done in 60 minutes or less. The flow path is discarded
after each use, minimizing the potential for product cross-contamination.
Membrane integrity testing: The feed pump on the SS UF–DF system is a rotary lobe pump. An auxiliary peristaltic pump is required to perform membrane
integrity testing for the SS UF/DF system. The SU UF–DF system eliminates the need for an auxiliary peristaltic pump for membrane
integrity testing.
UF–DF general operation: Four fully automated operations (initial fill, fed-batch, DF, and batch concentration) make the system user friendly. Process
parameters are easy to input, and the resulting data are captured in our data acquisition system, eliminating the need to
transcribe data into a paper-based GMP batch record. This facilitates trending or comparisons from batch to batch. Electronic
data also simplify the technology transfer from pilot plant to commercial manufacturing.
Feed flow rate: One parameter with a significant impact on the UF–DF operation is the feed flow rate. To maximize the permeate flux, UF–DF
operations typically require flow rates in the range of 240–360 LHM to maintain specific crossflow characteristics. With the
SU Mobius FlexReady Solution, the peristaltic feed pump can achieve 20 L/min with 30 psi back pressure. The feed pump is either
controlled by ΔP or fixed pump speed, which is correlated to the flow rate. The flow rate meets our current process feed flow
rate specification, and the automated TMP control using the retentate PCV is relatively stable.
Figure 2. Constant-volume diafiltration in the Millipore single-use Mobius FlexReady solution for TFF. The solid line is experimental
and the dotted line is theoretical: Retentate residue (%) = 100 * e (R–1)*N, where R is the Donnan effect and N is diavolume.
|
Mixing Efficiency: Mixing is critical during the diafiltration step to ensure a homogeneous solution for efficient buffer exchange. With the
previous SS UF–DF system, retentate return flow distribution was the only method for mixing in the retentate tank. The SU
Mobius TFF system design incorporates flow distribution by a retentate diverter plate and a magnetically coupled agitator
for enhanced mixing in the retentate tank. Eliminating dead legs from the flow path is critical to achieving efficient buffer
exchange. To this end, the design of the low dead-volume t-connectors for the pressure indicators in the disposable flow path
assembly is important. In Figure 2, diafiltration from 1.1 M NaCl to 0.15 M NaCl was performed at a constant-volume diafiltration
(50 L) and 9% agitation speed with a 5 m2 30 kD Millipore Biomax membrane. The maximum diafiltration volume with low agitation speed was considered the worst-case
scenario. The experimental curve, is comparable to the theoretical curve indicating good mixing.
Product recovery: To increase the percent recovery, a buffer flush step was performed to recover product retained in the system and membrane.
With the SU Mobius FlexReady TFF system, buffer can be transferred to the retentate tank and weighed directly. This saves
time and eliminates the step of weighing the flush buffer, which was required with the previous SS UF–DF system.
System Process Control
To meet the requirement of process control, a retentate PCV was applied for constant TMP control. The retentate PCV position
can be selected between 0 and 100% open for feed flow rate control. The 100% open position was chosen for PCV at the start
of diafiltration. Table 1 summarizes the performance of constant TMP control at diafiltration phase for six GMP runs with
a feed flow rate of 180 LHM. The diafiltration started with the feeding flow rate controlled by the pump speed. Once the feed
flow rate was in the recommended operating range (ROP), constant TMP control was applied. The TMP setting was randomly selected
in the TMP ROP range (<20 psi). It was found that TMP could be controlled at a narrow fluctuation range with a coefficient
of variation (CV) less than 10% in these runs.
The system uses an ultrasonic permeate flow meter to calculate cumulative diafiltration volume (DV), which is used to target
the diafiltration process endpoint. During the factory acceptance testing, the accuracy of the permeate flow meter was tested
using four solutions with different densities (2 M guanidine HCl, 2 M urea, 100 mM NaCl, and pure water). The results suggest
that the solution densities have little impact on the flow meter reading (data not shown).
Figure 3. Comparison of measuring diafiltration volume by permeate flow meter (light columns) versus by weigh scale (dark columns)
|
In GMP production, the accuracy of the flow meter was evaluated by comparing DV values using a flow meter versus those obtained
form a weight scale. The comparison of results is summarized in Figure 3, where the dark columns represent DV results obtained
from the weigh scale and the light columns represent DV values obtained from the permeate flow meter. The average difference
between these two measurements is 3.3%.
|
