Early Platform Modifications
The original process was developed at Pfizer for a 500-L scale bioreactor. To accommodate an expanding pipeline, the scale
of clinical manufacture was increased to 1,200 L, enabling the production of sufficient supplies in a shorter campaign. This
increase in scale made it necessary to introduce changes into the platform process. One change reflected the efficiency of
the initial depth filtration operation to remove cells and debris. During scale-up, we noted that effective filtration required
a much larger membrane area than was feasible, so we introduced a continuous flow centrifugation process, followed by depth
filtration to minimize membrane area, reduce costs, and increase the efficiency of cell removal.5
To increase process efficiency, we also changed the order of the polishing chromatography steps, allowing the intermediate
diafiltration step to be removed and thus reducing the total number of unit operations. Using this process, the pH of the
Protein A affinity chromatography eluate was adjusted upward after pH inactivation and loaded directly onto the anion exchange
resin. The anion exchange flow-through pool was then adjusted down to an optimal pH for the subsequent cation exchange step.
All other unit operations remained the same. The resulting process is shown in Figure 1 as process 2.
In the next iteration of the process, further modifications were made to process 2 to reduce the clinical manufacturing time.
The cation exchange polishing step was removed and the anion exchange chromatography step was changed from a resin-based operation
to one involving membrane adsorbers. This increased the loading capacity of this step by a factor of up to 100. All other
unit operations remained the same. The resulting process is shown in Figure 1 as process 3.
Then, the timeline for processes 2 and 3 was reduced by 30% (Figure 2). This was accomplished by performing previously defined
experiments for each unit operation with specified ranges for buffer systems, load capacities, and pH/conductivity values.
These experiments allowed ~80% of the process to be determined, with the remainder defined through additional bench-scale
experiments specific to the antibody product. Subsequently, the experimental conditions promoting maximum yield and product
purity were used to establish the parameters for each unit operation in the platform process. After the purification scheme
was established, a larger demonstration run was carried out to evaluate scale-up effects before transferring the process to
the pilot plant for manufacturing clinical supplies.
Current Platform Modifications
Although process 3 worked well for many Pfizer antibodies, the performance was not satisfactory for all antibodies, particularly
those produced at high titers with correspondingly high levels of impurities. In these cases, there were occasional issues
surrounding virus clearance, raising safety concerns especially for antibodies required in large doses. To address these issues
and increase the robustness of the platform, a cation exchange chromatography step was added to process 3 to define the current
process (Figure 1). Based on knowledge gained from the implementation of the original process, and modified processes 2 and
3, the load capacities, buffer solutions, and other operational parameters were standardized to maximize purification efficiency
for the majority of the antibodies in Pfizer's pipeline.
These changes reduced the timelines for purification development by 50% compared to the original process. The defined, optimized
platform process has been tested at the bench scale to verify that it is suitable for the production of clinical material.
At this point, it must meet yield and purity specifications before it is scaled up for the demonstration run, or transferred
to the pilot plant. This process is designed to meet the purification needs for 80% of the antibody products in our pipeline,
which means the majority of the purification processes can be verified without further development. The remaining antibody
projects may need slight modifications to meet individual specifications. These modifications will be evaluated on a case-by-case
Table 1 shows the impurity levels for various Pfizer antibodies using the processes outlined in Figure 1. The impurity profiles
are collected after the final chromatography step. The percent monomer is determined by size exclusion high performance liquid
chromatography, using two G3000 SWXL columns (Tosoh, Montgomeryville, PA) in series with detection of the antibody at 214
nm. HCP is detected using a high-sensitivity sandwich Enzyme-Linked ImmunoSorbent Assay (ELISA) kit specific for either CHO
or NS0 proteins following the manufacturer's instructions (Cygnus Technologies, Southport, NC). We measured residual recombinant
protein A levels using a sandwich ELISA kit from Repligen (Waltham, MA). Finally, residual total DNA was evaluated by extraction
using the DNA extractor kit from Wako (Richmond, VA), followed by detection of the single-stranded DNA using a threshold kit
(Molecular Devices, Sunnyvale, CA).
The results indicate that for the representative antibodies shown for the different process permutations, the current purification
scheme as outlined is sufficient to produce an acceptable product. A few of the antibodies required minor changes to the platform
process to produce the desired product purity, as explained in Table 1. Figure 3 shows the overall virus reduction achieved
for the early-phase representative antibodies in the different platform processes. Again, the results indicate that all processes
achieved satisfactory viral clearance. However, the current process is advantageous as we increase column capacities to accommodate
increasing titers, as insurance for acceptable virus clearance results.
Table 1. Impurity profiles and classifications for various therapeutic monoclonal antibodies using the platform processes
outlined in this article