Impact of Lot-to-Lot Variability of Cation Exchange Chromatography Resin on Process Performance

ABSTRACT

Cation exchange chromatography is commonly used as a polishing step in the purification of monoclonal antibodies. Cation exchange resins, however, typically have some lot-to-lot variability, which may result in a visually different elution profile, depending on the conditions used in the chromatography step. Product elution volumes also may be significantly different, depending on the product collection criteria. Despite this, there are no significant differences in the clearance of impurities. Differences in product volume, however, could lead to differences in the pH and conductivity of the product that could have an impact on the subsequent unit operation. This paper will present a case study that shows how the lot-to-lot variability of a cation exchange resin impacts its process performance and the performance of the subsequent process step. The study found that differences in product volume and elution profile were more prominent when resin lots containing larger percentages of smaller particles were used.

In a typical antibody purification process (Figure 1), a cation exchange chromatography unit operation is used as a polishing step. Cation exchange is one of the three chromatography steps in the purification process.¹ Protein A chromatography is the initial capture step. This is followed by a low pH hold step to inactivate viruses. Anion exchange chromatography, operated in flow-through mode, precedes the cation exchange (CEX) chromatography step. The purpose of this CEX step is to remove process- and product- related impurities such as aggregates, host cell proteins, and DNA. Downstream from cation exchange is the viral filtration step followed by ultrafiltration and diafiltration to concentrate and formulate the product.

(Millipore Corporation)

This paper presents a case study that shows how the lot-to-lot variability of a cation exchange resin impacts its process performance and the performance of the subsequent process step. In the study, all initial cation exchange development runs performed using resin lot A consistently delivered small eluate volumes. Resin lots B, C, and D, however, produced much larger eluates, one of which was almost double the eluate volume from lot A. In a manufacturing setting, unexpected variability in elution volume could be costly if hold tanks are not big enough to accommodate large product volumes. The product may overflow the tanks and have to be diverted to waste. The installation of larger tanks to contain the extra volumes would be expensive in terms of both time and money.

Figure 1. Flowchart of an antibody purification process

In addition to large elution volumes, significant visual differences in shape of elution profile were observed. These differences included a flat portion followed by a trailing portion on the descending part of the peak, instead of a smooth and relatively rapid fall.

To determine the root cause of the unusually wide elution peaks, bench-scale cation exchange experiments were performed on four columns with unused resins. The first column was packed with lot A, the lot of resin that was used for all the initial cation exchange development work. The remaining three columns were packed with lots B, C, and D, respectively.

MATERIALS AND METHODS

Some details relating to materials and methods could not be included because of confidentiality concerns.

Column Packing and Evaluation

Four 1.1-cm diameter Millipore Vanguard columns were each packed with a different lot of cation exchange resin to a height of 20±2 cm. Equilibration buffer was applied to each column. After the columns were equilibrated, approximately 0.25 mL (1.3% CV) M NaCl was injected. The conductivity of the column effluent was monitored. Column performance was evaluated by calculating the asymmetry and the height of a theoretical plate (HETP) from the NaCl chromatogram. Following this, the columns containing lots B, C, and D were also evaluated with a pulse acetone injection. The resulting signal was measured with UV at 280 nm.

Operation of the Cation Exchange Columns

All the cation exchange runs were performed on AKTA Explorer systems, which were controlled by Unicorn software. At the beginning of each run, equilibration buffer was pumped through each column for five column volumes. Following equilibration, anion exchange product from a single cell culture harvest was loaded onto the columns to a capacity of 19 mg of antibody per mL of resin. The antibody was manufactured and purified through the anion exchange step in-house. When the target loading capacity was reached, the load-line was chased with equilibration buffer for half a column volume to ensure that all the load material in the line reached the column. The columns were then washed with an additional 2.5 column volumes of equilibration buffer. At the end of wash, a step change in salt concentration was applied to the column to elute the antibody.^2,3 Eluate collection was started when UV absorbance reached 0.5 AU on the ascending portion of the peak and stopped when the absorbance on the descending portion of the peak decreased to 0.5 AU. Elution buffer continued to be pumped onto the columns until the absorbance read 0.1 AU. The columns were subsequently sanitized for 30 minutes, flushed with storage buffer for four column volumes, and stored at 2–8 °C when not in use. Linear velocity was maintained at 200 cm/hr for the entire duration of each run. The operation of the cation exchange step is shown schematically in Figure 2.

Figure 2. Operation of the cation exchange step

Use of High Conductivity Elution Buffer

In another set of experiments, elution conditions were changed for two of the columns. In these modified runs, the conductivity of the elution buffer was increased by a factor of two. Bound antibody was eluted with this high conductivity buffer. All the other process parameters remained the same.

Analysis of Samples

All cation exchange load and eluate samples were assayed for yield, product purity, and product quality. Yield was calculated by assaying for the concentration of IgG in load and eluate samples using absorbance obtained at 280 nm. Host cell proteins and leached Protein A concentrations in the load and eluate samples were measured using an in-house ELISA method, whereas DNA was quantified using a Q-PCR method. Charged isoforms of the antibody was measured with an in-house method using an analytical CEX column.

RESULTS AND DISCUSSION

Elution Volumes and Profiles

More than 50 initial cation exchange development runs performed using resin lot A consistently delivered eluates of less than 1.5 column volumes. In this study, lot A had an eluate volume of 1.4 CV. Cation exchange resin lot D had an unusually wide eluate peak of 2.6 CV. Lots B and C also obtained elution volumes larger than 1.5 CV. The eluate from lot B was 1.6 CV, and the eluate produced by lot C was 2.1 CV. These volumes are shown graphically in Figure 3.

Figure 3. Lot-to-lot variation in elution volume

There were also significant visual differences in shape of elution profile. The UV traces of the elution profiles for lots C and D appear to have a flat portion followed by a trailing portion. This feature is not apparent for lots A and B, as shown in Figure 4.

Figure 4. Lot-to-lot variability in shape of elution peak

The discussion into the probable root cause of the wide product volumes and unusual elution profiles continues below.

Column Evaluation with NaCl

Table 1 shows that the column asymmetries, evaluated with 1 M NaCl, varied from 1.0 to 1.7.⁴ The table also points to a correlation between column asymmetry, as evaluated with NaCl and elution volume. Generally, as asymmetry increases, column volume also increases. This suggests that the columns may have been poorly packed, resulting in varying flow patterns. To determine if this was the case, the probable correlation was investigated using two methods.

Table 1. Column asymmetries

Column evaluation with acetone. In the first method, ionic interactions between the analyte and the resin were eliminated by using a 3% v/v acetone injection. Acetone should not interact ionically with the cation exchange resin. Column asymmetries when evaluated with acetone were almost identical (Figure 5). Acetone evaluation showed that in the absence of ionic interactions, all four columns showed equivalent flow characteristics.

Figure 5. Peaks from acetone injections

Elution with high conductivity buffer. The second method of investigation used a high conductivity buffer to elute the bound antibody off the columns. Columns containing cation exchange resin lots A and D were used because they represented the extremes of asymmetry and column volume. The purpose of this experiment was to see if the different shapes of the elution peaks were caused by poor column packing or if they were because of different selectivity of the resin for the charged isoforms of the antibody. The use of a much stronger elution buffer would significantly reduce the selectivity for all the isoforms so that if column packing was predominantly responsible for the different peak shapes, these differences would still be observed.⁵ The chromatograms are shown in Figure 6.

Figure 6. Elution with high conductivity buffer

The solid lines show the widely varying elution profiles and volumes that were produced when the normal elution buffer was used. The dashed lines show that with the high conductivity buffer, the elution peaks were almost identical. The fact that no difference in the elution profile was observed under high conductivity conditions indicates that the variation in the elution under normal conditions is largely because of the different lots of resin having a difference in selectivity for the charged isoforms of the antibody. Thus, this variation is not likely to be related to column packing. To further understand the reason for this different selectivity, we examined the ionic capacity and the particle size distribution of the four resin lots.

Ionic Capacity

Certificates of analysis from the manufacturer contained the ionic capacity of each lot of resin. Table 2 lists the ionic capacities of each of the four lots. They are very similar and no trends are apparent. Ionic capacity of the resin lots did not cause the lot-to-lot variability in peak shape and elution volume that we had observed.

Table 2. No variation in ionic capacity

Resin Particle Size Distribution

The manufacturer guarantees that at least 80% of the particles that make up any lot of the cation exchange resin it releases have diameters that fall within the range of 40–90 μm. The certificates of analysis, which accompany each lot of resin, report the percentage of particles falling within this range. Table 3 summarizes this percentage for the four lots of resin we tested.

Table 3. Percent of particles falling within 40–90 μm

From the table, we can see that the percentage of particles within 40–90 μm diameter is variable from lot-to-lot, although all the lots would pass the vendor's release specification. In addition, the table shows that elution volume increases as the percentage of smaller particles, in this case those within the 40–90 μm range, decreases.^6–8 On request, the manufacturer provided more detailed particle size distribution information of lots A, B, C, and D. In this information, almost all of the particles falling outside of the 40–90 μm range are smaller than 40 μm. The percentage of these small particles can be seen in Table 4.

Table 4. Percent of particles smaller than 40 μm

Figure 7 illustrates the fact that the resin lots containing a larger percentage of particles less than 40 μm in diameter produce wider elution peaks.

Figure 7. Effect of resin particle size on elution volume

It is likely that smaller CEX resin particles allow for higher accessibility of the antibody to the charged sites by providing a shorter diffusional length inside the pores. This likely increases the interaction of the antibody with the charged sites, resulting in a wider elution peak and greater resolution (or higher selectivity) between the charged isoforms of the antibody.

Effect on the Product

It is important to understand the impact of this variable elution profile on the composition of the different charged isoforms in the product. Additionally, it is important to assess if the lot-to-lot variability results in any differences in the performance attributes of this cation exchange step such as yield and clearance of process and product related impurities. Analytical results show that yield clearance of product and process related impurities were not affected by the different cation exchange chromatography profiles obtained with the four different lots. The results are summarized in Table 5.

Table 5. Product yield and removal of impurities

The distribution of the different charged isoforms (N- and C- terminal charged isoforms) of the product was not affected either. These results, which are almost identical for each lot tested, are shown in Table 6.

Table 6. N- and C-terminal charged isoform distribution

Impact of Different Elution Profiles on Viral Filtration Step

As mentioned above, the CEX step is followed by a viral filtration using a normal flow parvovirus (NFP) filter is subsequently used in the downstream step. During the development of this process step, it was determined that controlling the concentration of NFP load to approximately 8.5 mg/mL and diluting the concentrated cation exchange product with a lower conductivity cation exchange equilibration buffer to reduce the conductivity resulted in the optimal performance of this viral filtration step. However, a consequence of having a wider than usual elution peak is a lower product concentration and a higher elution volume, allowing little or no dilution with the equilibration buffer because of volumetric constraints in a manufacturing process. This could result in a higher conductivity load for the NFP step. A higher conductivity load has the potential to impact the flux across the filtration process, as shown in Figure 8.

Figure 8. Effect of conductivity on volumetric flux across an NFP filter

Process Development and Manufacturing Considerations

It may be possible to overcome the problem of lot-to-lot variability in resin particle size so that the cation exchange step and subsequent steps are not affected. A suitable resin defining process could potentially remove the smaller particles, making the resin lots much more uniform. Although such a process would likely produce a more normal elution profile, it could be expensive in terms of the volume of resin discarded.

CONCLUSIONS

In this study, significant visual differences in elution profile and product volume were observed for the resin lots that had > 10% of particles smaller than 40 μm. Concomitantly, there is an increase in product volume on elution with an increase in the percentage of smaller particles. This increase in product volume generally parallels the increase in the asymmetry factor measured with 1 M NaCl as the analyte. Studies with higher strength elution buffer and acetone rule out differences in column packing as possible causes of the variable elution profiles. Instead, the unusual elution profile is likely caused by higher accessibility of the antibody to the charged sites inside the pores. This higher accessibility is probably caused by the shorter diffusional length inside the pores of smaller particles. An effect of this is probably a stronger interaction and more resolution of the charged isoforms of the antibody, resulting in a broader elution profile. Despite these differences, performance of the cation exchange step was not affected. No differences in yield, clearance of the impurities, or the distribution of charged isoforms were observed for these lots. However, an increase in the cation exchange pool conductivity was observed and this could impact the volumetric flux across the subsequent viral filtration step, especially at higher volumetric loadings.

Jane Wahome is a former development engineer from PDL BioPharma, Brooklyn Park, MN, jane.wahome@gmail.comWeichang Zhou, PhD, is the senior director, bioprocess engineering, technology development, at Genzyme, Framingham, MA, and Amitava Kundu, PhD, is the associate director, process development at GenMab, Brooklyn Park, MN.

REFERENCES

1. Ishihara T, Kadoya T. Accelerated purification process development of monoclonal antibodies for shortening time to clinic. Design and case study of chromatography processes. J Chromatogr A. 2007 Dec 28;1176(1–2):149–156.

2. Yamamoto S, Nomura M, Sano Y. Stepwise elution chromatography as a method for both purification and concentration of proteins. Chem Eng Sci. 1992 Jan;47(1):185–188.

3. DiLeo M, Haimes E, Ley A, Chen J. Interaction of Resin-Buffer-Antibody Product Influences the development of an optimal cation exchange purification process for specific monoclonal antibody products. Dyax Corporation. 2007.

4. Cabanne C, Raedts M, Zavadzky E, Santarelli X. Evaluation of radial chromatography versus axial chromatography, practical approach. J Chromatogr B. 2007 Jan 15;845(2):191–199.

5. Janson JC, Ryden L. Protein purification-principles, high resolution methods, and applications. New York: Wiley-VHC; 1998.

6. Carta G, Ubiera A. Particle size distribution effects in batch adsorption. AICHE J. 2004;49(12):3066–3073.

7. Karau A, Benknen C, Thommes J, Kula MR. The influence of particle size distribution and operating conditions on the adsorption performance in fluidized beds. Biotechnol Bioeng. 1997 Jul 5:55(1):54–64.

8. Billen J, Davy D, Rudaz S, Veuthey JL, Ritchie H, Grady B, Desmet G. Relation between the particle size distribution and the kinetic performance of packed columns: Application to a commercial sub-2 μm particle material. J Chromatogr A. 2007 Aug 17;1161(1–2):224–233.