Kinetic Modeling of Methionine Oxidation in Monoclonal Antibodies from Hydrogen Peroxide Spiking Studies

Correlation between Met Oxidation in mAbs and Peroxide Consumption

In the H₂O₂-spiking studies, Met 252 and Met 428 in the Fc region (Fc-Met) of all three mAbs were monitored for oxidation because they were the most surface-exposed Met residues (8) and thus most susceptible to H₂O₂-induced oxidation. Fc-Met oxidation and H₂O₂ residual for mAb1, mAb2, and mAb3 are presented in Tables II, III, and IV, respectively. For mAb1 at 10 mg/mL protein concentration, Fc-Met oxidation was initially detected qualitatively by trypsin peptide mapping for the presence of oxidized peptide in the samples spiked with 1000 and 250 ng/mL H₂O₂ after storage at 5 °C for 3 and 6 months, respectively (Table II). Without spiked-in peroxide, a minimal amount of Met oxidation was observed in the control. mAb1 samples spiked with less than or equal to 100 ng/mL H₂O₂ also had similar oxidation results when compared with the control throughout the stability study. Peroxide content in solution decreased over time for all samples spiked with different levels of H₂O₂, presumably as a result of consumption by the Fc-Mets for oxidation.

For mAb2 at 25 mg/mL protein concentration, Met 252 and Met 428 oxidation was measured separately and quantitatively by Lys-C peptide mapping. No difference in oxidation levels were observed in the mAb2 samples spiked with not more than 50 ng/mL when compared with the unspiked control. At 5 °C, an inverse correlation was observed between the amount of peroxide residual and the level of Met 252 oxidation in the mAb2 sample spiked with 5000 ng/mL peroxide (Figure 1). With 5000 ng/mL H₂O₂ spiked in, a large increase (21.2%) in Met 252 oxidation was observed throughout a period of 3 months at 5 °C that corresponded to a significant decrease of about 85% of the initial in peroxide concentration (Table III). Increased in Met oxidation and decrease in peroxide concentration were also observed at subsequent time points (e.g., 6 and 9 months). However, the degree of oxidation was much less when compared to the initial 3 month time point. Similar results were observed for mAb2 samples containing 250 and 1000 ng/mL H₂O₂ under the same storage conditions. Additionally, Met 428 behaved similarly to Met 252 in terms of oxidation by H₂O₂ in the mAb2 formulation, but the extent of oxidation was approximately one-half of that observed at Met 252. These observations suggested that H₂O₂ in the mAb samples might have been consumed by the Fc-Met residues in the protein molecule during the initial months, which would explain the minimal additional increase in oxidation level thereafter.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Correlation between Met 252 oxidation and loss of H₂O₂ in the mAb2 formulation (25 mg/mL protein concentration) spiked with 5000 ng/mL H₂O₂ and stored at 5 °C.

For mAb3 at 200 mg/mL protein concentration, large increases of Met 252 and Met 428 oxidation induced by H₂O₂ was observed only when a very high peroxide concentration (e.g., 7000 ng/mL) was spiked in. The reaction occurred within the first month at 5 °C, as no further oxidation was observed for both Met 252 and Met 428 after 1 month, and the peroxide level in the sample also decreased significantly (∼99%) at the same time (Table IV). For mAb3 samples spiked with less than or equal to 1000 ng/mL, no difference in both Met 252 and Met 428 oxidation levels were observed when compared with the unspiked control.

Extent of Fc-Met Oxidation in mAb Formulation

The results from the H₂O₂ spiking studies for mAb1, mAb2, and mAb3 are in agreement with published reports that Met and H₂O₂ react in a one-to-one stoichiometric ratio (17, 18), and Met 252 is twice as reactive as Met 428 (14). Previous studies demonstrated that Fc-Met oxidation rates of mAbs are similar in solutions with a pH range of 3–8 (19); the extent of Met 252 oxidation should be comparable among mAbs in commonly used formulation buffers that have pH values of 5.5–6.5. Based on these factors, the amount of Met 252 and Met 428 oxidation, as determined by peptide mapping, can be expressed as the following: where 34 represents the molecular weight (MW) of H₂O₂ in Da, and 150 represents the MW (in kDa) of mAb, and 2 represents two Met_{252 or 428} residues; and where C_H2O2 and C_Met are the concentrations of H₂O₂ and Met, respectively. The factors of 2/3 in eq 3 and 1/3 in eq 4 are introduced because Met 252 is twice reactive over Met 428 (e.g., 2/3 of total H₂O₂ will react with Met 252, whereas 1/3 will react with Met 428). Based on these equations, the amount of oxidized Met 252 (in percent) can be calculated for a given concentration of H₂O₂ present in mAb solution at a given protein concentration. A plot of the theoretical percentage of Met 252 oxidation calculated versus H₂O₂ concentrations for various mAb concentrations is shown in Figure 2. For a mAb at 25 mg/mL (with Met 252 or Met 428) at a concentration of 0.333 × 10^–3 M, the expected Met 252 and Met 428 oxidation by H₂O₂ at 1000 ng/mL (i.e., 0.03 × 10^–3 M) should be 6% and 3%, respectively. The actual observed Met 252 and Met 428 oxidation in mAb2 (25 mg/mL) formulation spiked with 1000 ng/mL H₂O₂ was 4.7% and 2.4%, respectively, when the peroxide was totally consumed after 6 months at 5 °C (Table III). The theoretical and observed oxidations of Met residues in the mAb2 formulation are considered comparable.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Theoretical percent Met 252 oxidation based on protein concentration and a ratio of 2:1 for Met 252 to Met 428 oxidation.

Consumption of H₂O₂ by Fc-Met Oxidation in mAb Formulations Can Be Pseudo First-Order Kinetics

When the spiked-in H₂O₂ is consumed in a protein formulation, it is expected that the measured H₂O₂ in the protein solution would slowly disappear until all the H₂O₂ has been expended. The time course of the consumption of H₂O₂ by mAb1 during storage at 5 °C is shown in Figure 3. When taking the log of the H₂O₂ concentrations as a function of time, a linear fit of the results was observed (Figure 4A). Similar results were also observed for the mAb2 and mAb3 formulations when the same calculation was applied to the H₂O₂ concentrations (Figures 4B and 4C). The observed linear regressions indicate that consumption of H₂O₂ in the presence of the mAb follows a pseudo first-order kinetic reaction where the rate constants (k) can be obtained from the slope of the linear regression from the plots. This is assuming that the H₂O₂ is only consumed by Fc-Met residues. The excipients in the formulation have no effect on H₂O₂ consumption, and no other reactive Met residues are present in the mAb. Because the Met residues of the mAbs are most likely being oxidized by the spiked-in H₂O₂, the bi-molecular reaction can be written as With the reaction rate equation: where [Met] is the concentration of Met residues in the protein formulation, and [H₂O₂] is the concentration that is spiked into the samples. When the protein concentration is in excess (i.e., [Met] >>> [H₂O₂]), the Met concentration remains unchanged and the first-order rate equation can be reduced to where Or it can be written as where [H₂O₂]_t0 and [H₂O₂]_t are the concentrations of H₂O₂ at the initial time point and at another specified time point, respectively. The assumption that the protein concentration is always in excess compared with the H₂O₂ concentration is reasonable (in these spiking studies) because the amount of protein is always much greater than the H₂O₂ that is spiked in. The concentration of 10 mg/mL mAb, which has a MW of 150 kDa, is equal to 0.066 mM with two chains each having Met 252 and Met 428 residues, which results in a total concentration of 0.26 mM (0.066 × 4) of Met residues that can be potentially oxidized by H₂O₂—whereas H₂O₂ has a MW of 34 Da, and it can range from 0.00029 mM at 10 ng/mL to 0.21 mM at 7000 ng/mL. Based on these observations, it can be concluded that the H₂O₂ consumption follows the pseudo first-order kinetics (eq 8), where the straight lines were observed in the semi-log plots as shown in Figures 4A, 4B, and 4C. In addition, the reaction rate constant is proportional to the Met concentration, which is proportional to the protein concentration (eq 7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Time course of H₂O₂ consumption in mAb1 formulation (10 mg/mL protein concentration) spiked with 100, 250, and 1000 ng/mL H₂O₂ and stored at 5 °C.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Pseudo first-order kinetic reactions of H₂O₂ consumption by Met oxidation observed for various concentrations of H₂O₂ spiked in (A) mAb1 formulation (10 mg/mL protein concentration); (B) mAb2 formulation (25 mg/mL protein concentration); and (C) mAb3 formulation (200 mg/mL protein concentration).

H₂O₂ Consumption Rates are Proportional to mAb Concentration

Based on the pseudo first-order kinetics observed in the three mAb formulations (Figures 4A, 4B, and 4C), the consumption rate of H₂O₂ can be based on protein concentration. In the H₂O₂ spiking studies, the higher the protein concentration in the formulation, the faster the H₂O₂ was consumed. As a result, the amount of time required for the mAb to consume the spiked-in H₂O₂ can be calculated theoretically based on its concentration given the assumptions presented. Based on this theoretical calculation, the required duration for the H₂O₂ spiking stability study may be estimated. The spiking study can be carried out until the spiked-in H₂O₂ can no longer be detected. If residual amounts of H₂O₂ are still detectable in the protein formulation, the potential for the mAb to be oxidized still exists. Based on Figure 3, it took approximately 24 months at 5 °C for the H₂O₂ in the mAb1 (10 mg/mL) formulation spiked with 1000 ng/mL H₂O₂ to reach an undetectable level. Because the mAb2 formulation is 2.5 times the protein concentration (25 mg/mL) of the mAb1 formulation, the consumption rate of H₂O₂ in the mAb2 formulation is expected to be 2.5 times faster than that which was observed for mAb1, as both mAbs have the same MW of 150 kDa and have the same IgG1 backbone and Met residues are similarly surface-exposed. The spiked-in 1000 ng/mL H₂O₂ in the mAb2 (25 mg/mL) formulation should be consumed in 9.6 months of storage at 5 °C, theoretically. For the mAb3 formulation with a protein concentration of 200 mg/mL, the H₂O₂ is expected to reach an undetectable level within 1.2 months under the same storage condition because the protein concentration is 20 times that of mAb1. The actual H₂O₂ consumption rates observed for the mAb2 and mAb3 formulations in this study basically followed this assumption with some exceptions. The consumption rates observed were faster than the predicted rates, as shown in Table V. Additionally, the predicted consumption rates of mAb2 and mAb3 were not within the calculated 95% confidence intervals of the observed consumption rates from the spiking studies. The fact that all three mAb formulations were formulated using different excipients may have also contributed to the increase in consumption rates of the H₂O₂.

Formulation Excipients' Impact on H₂O₂ Degradation

The observation of faster-than-expected H₂O₂ consumption rates in the spiking studies for mAb2 and mAb3 led us to investigate if excipients in the mAb formulations may affect the degradation of H₂O₂. To determine the formulation excipient effect, H₂O₂ at 200 ng/mL was spiked into the mAb1 formulation buffer (citrate buffer at pH 6.5 with sodium chloride and polysorbate 80) and the mAb2 formulation buffer (phosphate buffer at pH 6.2 with trehalose and polysorbate 20). In the absence of protein, it is expected that the H₂O₂ should remain fairly constant in both formulation buffers because there is nothing present that can consume the spiked-in H₂O₂. After 28 days at 30 °C, the amount of H₂O₂ (200 ng/mL) in the mAb1 formulation buffer remained constant. However, the 200 ng/mL H₂O₂ spiked into the mAb2 formulation buffer was depleted after 14 days at 30 °C as shown in Figure 5. Similar results were also observed when a higher level of H₂O₂ (2000 ng/mL) was spiked into the same buffers (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Degradation of 200 ng/mL H₂O₂ in mAb1 and mAb2 formulation buffers (without protein) over time at 30 °C.

As a result of the H₂O₂ degradation observed in the mAb2 formulation buffer, each component of the buffer (i.e., phosphate, trehalose, and polysorbate 20) was investigated to determine if any of the excipient(s) may potentially degrade or consume the peroxide. H₂O₂ was spiked into the following five buffers at a final concentration of 200 ng/mL and placed at 30 °C for up to 7 days: (1) mAb2 buffer as positive control, (2) mAb2 buffer without trehalose and polysorbate 20, (3) mAb2 buffer without trehalose, (4) mAb2 buffer without polysorbate 20, and (5) mAb2 buffer plus 0.05% EDTA. EDTA was added to the mAb2 formulation buffer to evaluate whether a chelating agent might help prevent the degradation of H₂O₂. When comparing the residual peroxide concentration across all five buffer solutions within 7 days at 30 °C, significant degradation of H₂O₂ was only observed in the buffer solutions that contained trehalose (Figure 6). In fact, the peroxide was completely depleted after 3–4 days at 30 °C for the trehalose-containing buffer solutions. For the buffer solutions containing no trehalose, minimal loss of the peroxide was detected for the same incubation period. The presence of EDTA as chelating agent reduced the degradation rate of the H₂O₂ in the mAb2 buffer, suggesting that trace impurities of metal ions may be present in the buffer (possibly from trehalose) that might have catalyzed the degradation of H₂O₂ and resulted in a faster degradation rate.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Effect of formulation excipients on degradation of 200 ng/mL H₂O₂ over time at 30 °C in the following buffers: (#1) mAb2 formulation buffer (phosphate buffer at pH 6.2 with trehalose and polysorbate 20); (#2) mAb2 formulation buffer without trehalose and polysorbate 20; (#3) mAb2 formulation buffer without trehalose; (#4) mAb2 formulation buffer without polysorbate 20; and (#5) mAb2 formulation buffer with 0.05% EDTA.

Because many protein formulations contain either trehalose or sucrose as a cryo-protectant for drug substance storage, the effect of sucrose on H₂O₂ degradation was also evaluated. With the presence of 120 mM or 240 mM sucrose and 200 ng/mL H₂O₂ in phosphate buffer at pH 6.2, the degradation rates of H₂O₂ were similar to the trehalose-containing buffer solutions at 30 °C (Figure 7). However, the rate of peroxide depletion was not dependent on sugar concentration, and it was reduced by the addition of EDTA. Keeping this observation in mind, the kinetic model presented earlier in this report may be considered a worst-case assumption in regards to the amount of H₂O₂ and time required for the spiking studies. Therefore, the kinetic model is considered a valuable tool for planning such spiking studies to support the acceptable level of residual peroxide for manufacture of protein drug products using VPHP for sterilization in isolators.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Effect of 120 mM sucrose and 240 mM sucrose with and without 0.05% EDTA on degradation of 200 ng/mL H₂O₂ in phosphate buffer at pH 6.2 (PB) over 7 days at 30 °C.

Consumption of H₂O₂ by L-Methionine (Free) in Formulation Buffers and Activation Energy

Because the duration of a H₂O₂ spiking study can be calculated based on the correlation between Met oxidation in the mAb and peroxide consumption, similar calculations may also be useful if it can be demonstrated that free L-methionine behaves similarly in H₂O₂ spiking studies.

Because Fc-Met residues in a mAb can consume the spiked-in H₂O₂, free L-methionine should act similarly as Fc-Met residues in consuming the H₂O₂ based on the assumption that Fc-Met residues are strongly surface-exposed and react in a similar way as L-methionine (20). A study for determining the reaction rate between L-methionine and H₂O₂ was conducted by spiking 1000 ng/mL H₂O₂ into a histidine buffer consisting of sucrose and polysorbate 20 at pH 5.5 with different concentrations of L-methionine that ranged from 0 to 10 mM. The samples were placed at 5 °C, 25 °C, and 40 °C for up to 72 h. Residual H₂O₂ was monitored throughout the study to determine whether the L-methionine in the formulation buffer also behaves similarly to the mAb formulations in terms of H₂O₂ consumption. Similar to H₂O₂ spiking studies with mAb formulations, pseudo first-order kinetics was observed for all L-methionine concentrations (data not shown) under accelerated and stress conditions. When plotting the H₂O₂ consumption rates of L-methionine in an Arrhenius plot (Figure 8), the slopes of these linear fits are nearly identical for the buffer containing various amounts of L-methionine, demonstrating that the activation energy of the reaction between L-methionine and H₂O₂ are comparable. When comparing the linear fits of mAb2 (phosphate buffer at pH 6.2 with 0.3 mM mAb2) and L-methionine (Figure 8), the slope of mAb2 is slightly steeper than the slopes of L-methionine at all concentrations. Using the slopes that were generated from Figure 8, the activation energies were calculated. The average activation energy between L-methionine and H₂O₂ was approximately 38 kJ/mol, and it was 59 kJ/mol for mAb2 (Table VI). L-methionine at all concentrations was formulated in a buffer containing a cryo-protectant that might contain trace impurities of metal ions as catalysts for the H₂O₂-induced oxidation to occur. Thus, less activation energy might be required for the reaction when compared with mAb2 containing no sugar. This observation is consistent with the results obtained from the formulation excipient investigation, where trehalose-containing buffer had a faster rate of H₂O₂ degradation as compared with the buffer containing no sugar.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

Arrhenius plot of H₂O₂ consumption by 0.5, 1, 5, and 10 mM L-methionine in histidine buffer consisting of sucrose and polysorbate 20 at pH 5.5 and 0.3 mM mAb2 in phosphate buffer at pH 6.2. The H₂O₂ consumption rate constants (in reciprocal day) versus reciprocals of storage temperatures are shown. The activation energies for the consumption of H₂O₂ by free L-methionine and mAb2 were calculated from the slope of the lines.

Types of H₂O₂ and Their Effect on Fc-Met Oxidation

To determine if there is any difference between the stabilized and unstabilized forms of H₂O₂ in terms of degradation, both types of peroxide were spiked into the mAb1 formulation buffer (citrate buffer at pH 6.5 with sodium chloride and polysorbate 80) and the mAb2 formulation buffer (phosphate buffer at pH 6.2 with trehalose and polysorbate 20) at a final concentration of 200 ng/mL, and the degradation of the peroxides was monitored at 30 °C. No differences were observed between the stabilized and unstabilized H₂O₂ in terms of their degradation behavior over time in these two buffers (Figure 9). Loss of H₂O₂ for both types was not detected in the mAb1 formulation buffer containing no trehalose after 3 months at 30 °C. In contrast, both types of H₂O₂ in the mAb2 formulation buffer that contained trehalose were completely depleted within 7 days at 30 °C. Because the H₂O₂ concentration was quite low (diluted from 35% or 350 mg/mL to 200 ng/mL) in the formulation buffers, the amount of stabilizers originally present in the stock solution may have been too dilute to have an impact on peroxide degradation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9

Degradation of 200 ng/mL H₂O₂ (stabilized versus unstabilized) in mAb1 formulation buffer (no trehalose) and mAb2 formulation buffer (with trehalose).

Because there were no differences in degradation between the stabilized and unstabilized H₂O₂ in the formulation buffers, it is expected that there should also be no difference in the kinetics of H₂O₂-induced Met oxidation in mAbs. However, there was a concern that the stabilizers present in the stabilized H₂O₂ may have an effect on Met oxidation. For this reason, both types of H₂O₂ (final concentration of 500 ng/mL each) were spiked into the mAb2 formulated at 10 mg/mL in phosphate buffer at pH 6.2 to assess the difference between stabilized and unstabilized H₂O₂ in terms of H₂O₂ consumption and Met oxidation at 30 °C. Trehalose and polysorbate were excluded in the buffer in order to eliminate the effect of impurities such as metal ions on H₂O₂ degradation. After 21 days at 30 °C, the amount of Met 428 and Met 252 oxidation and the total H₂O₂ consumed in the stabilized and unstabilized peroxide-spiked samples were comparable (Table VII). A slight increase in both Met 428 and Met 252 oxidation of the mAb2 control was also observed when compared with the initial time point, which is expected at an elevated incubation temperature of 30 °C. Overall, the amount of Met 252 oxidation observed was approximately twice the amount of Met 428 oxidation for both stabilized and unstabilized peroxide-spiked samples, which had a ratio of about 2:1. This result is consistent with other studies that have also shown that when mAbs are exposed to an oxidizing agent (e.g., H₂O₂), the amount of Met 252 oxidation observed is about twice the amount of Met 428 oxidation (6, 21). With the results from our studies (with and without protein), it can be concluded that both stabilized and unstabilized H₂O₂ was comparable.