Figure
7. Schematic of mass flow rate control. (1) FproAeluate: protein A product material flow rate,
Facid: acid flow rate, F: total flow into VI SEC column,
Fsp: set point flow rate, Eflow: error
in target flow into SEC column. (2) Elevel: Error in
target vessel level, V: vessel level, Vsp: set point
vessel level. (3) Ratio: target ratio of acid flowrate to product
flowrate.
The following steps are performed to adjust the pump flow rates to
ensure the virus inactivation residence time set point is met. First,
the total flow into the virus inactivation SEC column is calculated as
the sum of the protein A product feed (VI product pump) and the VI acid
pump flow rates. The flow rate set point that corresponds to the virus
inactivation residence time target is then subtracted from the total
inlet flow, providing the flow rate error. The flow rate error is fed
into a PID controller that controls the AEX product pump (the last
product pump in series in this process). Because this pump pulls from
the VI break vessel, any change in this flow rate will immediately
result in a change in VI product vessel level. The error of this vessel
is fed into a PID controller that controls the VI product pump to
maintain a constant VI vessel level. Finally, the VI acid pump output
increases or decreases so that the target ratio between the VI acid and
VI product pumps is always achieved.
The virus inactivation residence time for the third VI sub-run is
displayed in Figure 5c.
This series of independent control loops, connected only by the physical
properties of the VI product vessel allows for cohesive mass flow
control across unit operations. Additional unit operations can be added
between the VI and AEX steps with this strategy. In this case, flow
control cascades up through multiple break vessels.
The perfusion process required a break in mass flow rate at some point
in the process. The perfusion rate must be set constant and for this run
the VI residence time was controlled to a setpoint. Regardless of the
accuracy of pumps, there will inevitably be a difference in mass flow
rates between these two unit operations. The break in mass flow rate
occurred between the protein A chromatography step and the virus
inactivation step. The difference in mass flow rate was absorbed by the
protein A product break vessel. This is the reason the protein A product
vessel operated under min, low, high, max control rather than continuous
PID control and why the virus inactivation step ran as three separate
sub batches during this run. In future runs the VI step will operate
within a range of acceptable residence times which will allow VI flow
rates to adjust to match the changing output of the protein A eluate
vessel. This will enable this step to run continuously for the duration
of the process instead of periodically stopping and restarting.
Viral Clearance Data from continuous Virus
Inactivation
To ensure viral clearance across the flow through virus inactivation SEC
column, live virus spiking studies were performed. These studies aimed
to (1) characterize the breakthrough profile of virus and (2) to ensure
inactivation kinetics occur in the same manner to batch inactivation.
Before testing the column with the virus-spike, the breakthrough profile
of the column was characterized with blue dextran as a virus surrogate.
Blue dextran is a large molecule with a molecular weight of
~2000kDa (~10-20nm diameter). It is
commonly used to determine column void volume and is expected to behave
similarly to enveloped viruses, which are typically larger than 50 nm
(ICH Q5A, Food and Drug Administration, 1999). Blue dextran spiking
studies were performed at various flow rates across the SEC column to
characterize the breakthrough profiles of different column residence
times. Data can be seen in Figure 8.