1. Introduction
Following the proposal to use continuous processes for monoclonal
antibody (mAb) production (Konstantinov and Cooney, 2015), many studies
had demonstrated benefits such as reduced footprint or cost reduction.
However, even now, there are few examples of its adoption into actual
production. The reasons raised for not adopting this technology include
that downstream processes are not yet developed enough to support
continuous processes compared to the upstream process or that it is very
complicated and difficult to operate each step of the production process
while conducting inline quality control and feed adjustment
(Somasundaram et al., 2018; Gerstweiler et al., 2021; Kumar et al.,
2020). It has been reported that most pharmaceutical manufacturers
adopting continuous processes incorporate batch processes into the
overall process instead of conducting end-to-end continuous processes
(Coffman et al., 2021), and there are also proposals to utilize existing
fed-batch process as part of continuous GMP processes (Ferreira et al.,
2022). There are also reports of concrete proposals for making
continuous processes robust, controllable and practical (David et al.,
2020; Rathore et al., 2021; Coffiman et al., 2021).
One of the greatest expected benefits of the continuous process is its
cost advantage over batch process, and it was shown that combining
continuous processes and single-use technology has a clear cost
advantage, especially for processes with relatively small annual
production volumes (Pollard et al., 2016; Arnold et al., 2018; Hummel et
al., 2019; Farid et al., 2020; Mahal et al., 2021). Furthermore, the
benefits of continuous process contribute not only to cost reduction but
also to PMI (process mass intensity) reduction (Catalda et al., 2020).
While the benefits and challenges of continuous processes have been
extensively examined, there are few reports discussing virus filtration
processes to prevent virus contamination, which is a critical step in
the production process. Virus filtration is not only a critical step in
product safety, but also one of the most costly steps along with media
consumption and Protein A chromatography step (Pollard et al., 2016).
Compared to batch processes with flux of around 50 to 60 LMH (liters per
effective surface area of filter in square meters per hour), David et
al. (2019) reported that flux for a virus filter incorporated into a
continuous process is extremely small at 0.3 LMH but virus logarithmic
reduction value (LRV) of > 4 was still achieved even in
these ultra-low flux conditions and the processing volume per unit
membrane area became larger with lower flux. To verify the robustness of
virus filtration in a long-term continuous process, Lute et al. (2019)
performed continuous filtration over 4 days and showed that virus LRV
> 4 was achieved even at a throughput of 6900
L/m2. Bohonak et al. (2021) investigated a continuous
mAb process with a connected column process and virus filter operated in
constant pressure mode, and they reported that the intensified process
in which the mAb solution was concentrated by single-pass tangential
flow filtration (TFF) before the column process showed improved
purification efficiency of mAb and reduced virus filter membrane area
usage. Coolbaugh et al. (2021) achieved continuous operation for 25 days
for a virus filtration process operating in TFF mode incorporated into
an end-to-end integrated and continuous process.
Validation of the virus filter incorporated into a continuous process
cannot be evaluated individually, unlike in a batch process; rather, it
is necessary to consider an appropriate virus validation method for
continuous process. Proposals for various inline virus spiking methods
adapted for virus validation in a continuous process and confirmation of
robust virus removal by these methods has been reported by Lute et al.
(2011), Bohonak et al. (2021), Shirataki et al. (2021a) and Malakian et
al. (2022).
One of the major issues in adopting continuous processes for commercial
production is the definition of the product batch, and ICH Q13 Guideline
(2021) describe that the product batch should be clearly defined as
quantity of output material, quantity of input material and run time at
a defined mass flow rate. Use of the greatest common divisor (GCD) of
the time required for each step of the process has been reported to be
an appropriate definition of batch that meets this requirement (Lali et
al., 2021).
In this study, we present an end-to-end continuous process incorporating
a constant flux filtration process for the virus filter with feed
adjustment between each step and a clear batch concept of the product.
Here, the batch is set as the unit of product pooled in the low pH
inactivation step. In this process, the solution harvested from the
perfusion cell culture is continuously supplied to the affinity process
with Protein A column, and column eluate is pooled for the virus
inactivation step of low pH treatment for 1 h in an automated manner.
After low pH inactivation, the pH-adjusted mAb solution is supplied to a
total flow-through two column polishing process including direct
connection to a virus filter. The combination of the AEX and CEX columns
used here with the virus filter has already been reported to show very
high impurity removal, recovery rate and filterability of the virus
filter in a total flow-through process (Shirataki et al., 2021a). Though
activated carbon is reported to be effective for impurity and virus
clearance by flow-through (Ichikawa et al., 2019; Kikuchi et al., 2022),
we adopted flow-through two column chromatography in this study.
The total flow-through integrated polishing process has been reported to
achieve high mAb purification and robust virus reduction (Shirataki et
al., 2021a). In this report, we verify the effectiveness of this
integrated process when incorporated into a practical scaled-down
end-to-end continuous process. In our previous report (Shirataki et al.
2021a), robust reduction of minute virus of mice (MVM) and xenotropic
murine leukemia virus (X-MuLV) using inline spiking was confirmed for a
mAb process with throughput to about 100 L/m2 and low
flux levels expected in the continuous process (5 LMH lowest). In this
report, we examined MVM removal in a mAb solution on two different
hollow fiber virus filters (Planov BioEX and S20N, Asahi Kasei Medical)
to a throughput of about 1000 L/m2 and flux of 5 LMH
down to 1.5 LMH with and without process pause to especially test the
effect of process pause in low flux filtration. To test a process with
reduced filterability, we conducted spiking with X-MuLV at increased
titer and compared the performance of two types of hollow fiber virus
filters.