Peroxide Formation in Polysorbate 80 and Protein Stability EMILY HA, WEI WANG, Y. JOHN WANG Analytics & Formulation Department, Process Sciences, Bayer Biotechnology, 800 Dwight Way, Berkeley, California 94701
Received 30 January 2002; revised 11 April 2002; accepted 1 May 2002
ABSTRACT: Nonionic surfactants are widely used in the development of protein pharmaceuticals. However, the low level of residual peroxides in surfactants can potentially affect the stability of oxidation-sensitive proteins. In this report, we examined the peroxide formation in polysorbate 80 under a variety of storage conditions and tested the potential of peroxides in polysorbate 80 to oxidize a model protein, IL-2 mutein. For the first time, we demonstrated that peroxides can be easily generated in neat polysorbate 80 in the presence of air during incubation at elevated temperatures. Polysorbate 80 in aqueous solution exhibited a faster rate of peroxide formation and a greater amount of peroxides during incubation, which is further promoted/catalyzed by light. Peroxide formation can be greatly inhibited by preventing any contact with air/ oxygen during storage. IL-2 mutein can be easily oxidized both in liquid and solid states. A lower level of peroxides in polysorbate 80 did not change the rate of IL-2 mutein oxidation in liquid state but significantly accelerated its oxidation in solid state under air. A higher level of peroxides in polysorbate 80 caused a significant increase in IL-2 mutein oxidation both in liquid and solid states, and glutathione can significantly inhibit the peroxide-induced oxidation of IL-2 mutein in a lyophilized formulation. In addition, a higher level of peroxides in polysorbate 80 caused immediate IL-2 mutein oxidation during annealing in lyophilization, suggesting that implementation of an annealing step needs to be carefully evaluated in the development of a lyophilization process for oxidation-sensitive proteins in the presence of polysorbate. ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 91:2252–2264, 2002
Keywords:
polysorbate; peroxide; protein stability; IL-2; oxidation; lyophilization
INTRODUCTION Polysorbates are an important class of nonionic surfactants used in the pharmaceutical industry because of their effectiveness at low concentrations and relative low toxicities. They have long been applied to pharmaceutical preparations to facilitate solubilization of poorly soluble drugs1 and to enhance the stability of emulsions2 or
Emily Ha’s present address is Thomas J. Long School of Pharmacy & Health Sciences, University of the Pacific, 3601 Pacific Avenue, Stockton, CA 95211. Correspondence to: Wei Wang (Telephone: 510-705-4755; Fax: 510-705-5629; E-mail:
[email protected]) Journal of Pharmaceutical Sciences, Vol. 91, 2252–2264 (2002) ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association
2252
microemulsions.3 In the past 2 decades, they were widely used in the biotechnology industry because of their relatively inert nature when mixed with proteins and their strong effect in preventing/ inhibiting protein surface adsorption4 and aggregation under various processing conditions, such as refolding,5 mixing,6 freeze thawing,7 freeze drying,8 and reconstitution.9,10 Even chemical degradation could be inhibited to a certain degree in the presence of polysorbates.11 As a result, several protein pharmaceutical products, both in liquid and solid dosage forms, contain polysorbates as inactive pharmaceutical ingredients, including Actimmune, Activase, Intron A, and Recombinate.12 However, there have been concerns about use of polysorbates in pharmaceutical preparations
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
PEROXIDE FORMATION IN POLYSORBATE 80
because polysorbates contain low levels of residual peroxides,13,14 which may accumulate during storage and cause immediate and/or long-term damage to the active pharmaceutical ingredients. Structurally, polysorbates contain fatty acid esters of polyoxyethylene sorbitan and polysorbate 80 is an ester of a single oleic acid. It has been well documented that surfactants with alkyl polyoxyethylene chains such as polysorbates undergo autoxidation with subsequent chain-shortening degradation.13,15 The autoxidation starts with metal- and/or light-induced decomposition of alkyl polyoxyethylene chain and peroxides (initiation step), propagated with oxygen consumption and formation of hydroperoxidized derivatives, and terminated with collision among radicals. Results of the autoxidation include not only formation of peroxides on the polyoxyethylene chain of the molecule but also changes in physico-chemical properties of the surfactant such as a reduction in the cloud point, pH, and surface tension due to the formation of breakdown products.13 Although peroxide formation in diluted polysorbate 20 or other nonionic surfactants have been carefully examined,13,14,16 peroxidation of polysorbate 80 either in diluted solution or neat form has not been found in the literature. The oxidative damaging effect of peroxides in surfactants on drug molecules has been extensively reported. These include oxidation of benzocaine hydrochloride,17 penicillins,18 and aminophylline19 by peroxides generated from polyoxyethylenic nonionic surfactants. Although the effect of residual peroxides in polysorbates has not been widely reported on protein stability, it is anticipated that even trace amount of peroxides in polysorbates could cause significant damage to proteins because proteins are often formulated at relatively low concentrations. Indeed, a few such cases have been reported. Knepp et al.20 demonstrated that alkyl hydroperoxides in polysorbate 80 induced oxidation, dimerization, and subsequent aggregation of recombinant human ciliary neurotrophic factor (rhCNTF) in solution and the rate of reaction was similar to that induced by hydrogen peroxide at the same concentration. Herman et al.21 were able to correlate the level of peroxides in polysorbate 80 and the degree of oxidation of recombinant human granulocyte colony-stimulating factor (rhG-CSF) during storage. Another paper by Miki et al.22 reported a linkage between the oxidation of hydroperoxidase and peroxides generated in Triton X-100, a surfactant structurally similar to polysorbates. No
2253
reports, however, have been found on the potential effect of peroxides in surfactants on protein stability in solid state. Because surfactants often have to be used in protein pharmaceutical preparations to prevent protein surface adsorption and/or aggregation, proper control of the level of residual peroxides in surfactants is a key in protecting oxidationsensitive proteins. To control the level of residual peroxides in polysorbates, two methods may be applied independently or in combination—control of the source of oxygen, and/or use of antioxidants. Removal of oxygen from the system would prevent peroxidation by blocking the propagation step. Some manufacturers, such as Mazer Chemicals and Pierce, have used this strategy to limit the peroxide formation during storage and shipment by packaging surfactants under nitrogen. Johnson et al.23 showed that removal of oxygen in the headspace of ampoules prevented oxidation of Fenprostalene by peroxides generated in PEG 400 during storage. Several antioxidants, including cysteine, glutathione, and methionine, have been shown to prevent oxidation of recombinant human ciliary neurotrophic factor (rhCNTF) and recombinant human nerve growth factor (rhNGF) caused both by alkyl peroxides in polysorbate 80 and hydrogen peroxide in solution.20 It should be noted that antioxidants may potentially interact with protein molecules, and result in protein degradation or precipitation.20 In this article, we report for the first time how peroxides are generated in polysorbate 80 under a variety of storage conditions, and that peroxidation in polysorbate 80 can be effectively inhibited by reducing its contact with molecular oxygen. A model protein, IL-2 mutein, was used to demonstrate that peroxides generated in polysorbate 80 could accelerate protein oxidation both in liquid and solid states. Finally, we show that peroxideinduced IL-2 mutein oxidation can be effectively prevented/inhibited either by reducing the contact of the protein formulation with molecular oxygen or through addition of an antioxidant.
MATERIALS AND METHODS Materials NF-grade polysorbate 80 (lot N11662, peroxide level < 100 ppm) and sulfuric acid were purchased from J.T. Baker (Phillipsburg, NJ). Low-peroxide polysorbate 80 (lot 120K7276), Xylenol orange, ferrous chloride, and hydrogen peroxide (30% JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
2254
HA, WANG, AND WANG
solution) were purchased from Sigma Chemicals (St. Louis, MO). HPLC-grade water and acetonitrile were obtained from Fisher Scientific (Pittsburgh, PA). High-purity nitrogen (99.99%) was obtained from Air Products (Galt, CA). Recombinant human IL-2 mutein at 3.5 mg/mL in a frozen buffered solution was prepared by Bayer Corporation. This protein was derived from Chinese hamster ovary cells with a purity of greater than 99% by SDS-PAGE. All materials were used as received. Determination of Peroxide Concentration in Polysorbate 80 The FOX (ferrous oxidation with Xylenol orange) assay was used to determine the peroxide content in polysorbate 80 solutions.14 Briefly, 50 mL of sample was mixed with 5 mL of 10% (w/v) butylated-hydroxy toluene (BHT) in ethanol and 950 mL FOX reagent containing 250 mM ferrous chloride, 100 mM Xylenol orange, and 100 mM sorbitol in 25 mM sulfuric acid. BHT was added to prevent further peroxide generation during the assay process. The mixture was then incubated at room temperature for 20 min prior to reading at 530 nm on a Wallac Victor 2 automated microtiter plate reader (Gaithersburg, MD). Hydrogen peroxide was used to prepare standard curves and therefore, peroxide level in polysorbate 80 was obtained as peroxide equivalent to H2O2 standards. All neat polysorbate 80 samples were diluted to 20% solution with water before analysis. The peroxide level in neat polysorbate 80 samples was calculated as milliequivalents (mEq) or microequivalents (mEq) per kg of neat polysorbate 80 (equivalent to the H2O2 concentration in mM or mM). To make comparison easier, the peroxide level in 20% polysorbate 80 samples was also expressed per kg of neat polysorbate 80 (calculated by multiplying by 5). Stability Studies on Polysorbate 80 Stability studies were conducted on both neat and 20% (w/w) polysorbate 80. The 20% solution was prepared by diluting the neat polysorbate 80 with water for injection at a weight ratio of 1:4. Stability samples were prepared by dispensing 1 mL of neat or 20% polysorbate 80 into 10-mL flint tubing glass vials. These vials were placed in a freeze dryer (Virtis Genesis 35 EL) and cooled down to 0–28C. After the samples were cooled, vacuum was applied gradually to 10 mTorr and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
maintained for 10 min before the vials were either backfilled with air or nitrogen to 1 atmospheric pressure or directly sealed with bromobutyl rubber stoppers inside the freeze dryer. The stoppered vials were then capped manually. These stability samples were incubated at 40, 50, or 608C with or without light for 8 weeks. The lighting condition (460 foot candle) was provided with a fluorescent light box (Model BL1012) purchased from The Back Light Hall Productions (San Luis Obispo, CA). All the samples were stored at 808C until analysis. Preparation of Liquid IL-2 Mutein Stability Samples Recombinant human IL-2 mutein at 3.5 mg/mL was dialyzed extensively overnight into a buffered solution containing 5% mannitol and 20 mM citric acid at pH 5.5. The dialyzed protein was diluted to a final concentration of 1 mg/mL with the dialysis buffer containing no polysorbate 80, 0.1% lowperoxide polysorbate 80, or 0.1% high-peroxide polysorbate 80. The peroxide level in the protein solution containing low- and high-peroxide polysorbate 80 was, respectively, 0.33 mEq and 25 mEq per liter of protein solution. High-peroxide neat polysorbate 80 was obtained by incubating low peroxide polysorbate 80 at 408C under light for 4 days. The diluted IL-2 mutein solution was dispensed at 1 mL into 6-mL flint tubing glass vials. The head space of these sample vials was filled with air or nitrogen in the same way as described for the polysorbate 80 stability samples. Preparation of Lyophilized IL-2 Mutein Stability Samples The initial preparation of lyophilized IL-2 mutein stability samples was similar to that for the liquid stability samples. Reduced glutathione (GSH), an antioxidant, was included at 0.5% in some stability samples containing high-peroxide polysorbate 80. Lyophilization included the following steps: freezing to 458C at 0.48C/min, holding the temperature at 458C for 1 h, increasing the temperature to 208C at 0.48C/min, holding the temperature at 208C for 1 h (annealing), decreasing the temperature to 458C at 0.48C/min, increasing the vacuum to 100 mTorr, increasing the temperature to 258C at 0.48C/min, holding the temperature for 16 h (primary drying), and increasing the temperature to 208C at 0.48C/min and holding the temperature for 24 h (secondary drying). At the end of the drying process, the vials
PEROXIDE FORMATION IN POLYSORBATE 80
2255
were either stoppered immediately under vacuum or backfilled with air or nitrogen. Both liquid and lyophilized IL-2 mutein stability samples were incubated at 408C. Stability samples were stored at 808C until analysis. Reverse-Phase High-Performance Liquid Chromatography (RP-HPLC) RP-HPLC analysis of IL-2 mutein and its oxidized product was done on HP 1100 (Hewlett Packard, Pleasanton, CA). A Vydac C-18 column (250 4.6mm, 5 m, 300 Angstrom) was used (Vydac, Hesperia, CA). The mobile phase contained solvent A (0.1% TFA in water) and B (0.1% TFA in acetonitrile). A gradient elution at a flow rate of 1.0 mL/min was used according to the following program: 45% B to 70% B in 16 min, 75% to 100% B over 10 s and hold for 7 min, 100% to 45% B in 10 s and hold for 7 min. The IL-2 mutein and its oxidized product, respectively, with a retention time of 13.8 and 13.0 min, were monitored at 280 nm.
RESULTS Peroxide Formation and Decomposition in Polysorbate 80 To our knowledge, peroxide formation in neat polysorbate 80 during storage has not been reported in the literature. Therefore, we examined peroxide formation in neat polysorbate 80 under air or nitrogen. To accelerate formation of peroxides, stability samples were incubated at elevated temperatures of 40, 50, or 608C, all in dark. The starting peroxide level in the neat polysorbate 80 was 0.16 mEq. In the presence of air the formation of peroxides was accelerated upon incubation (Fig. 1). The peroxide in the samples under air reached a peak level of 15, 25, and 36 mEq, respectively, at 60, 50, and 408C. Although the peak level of peroxides was reached at week 2 at 60 and 508C, it was reached at week 4 at 408C. Further incubation caused a gradual drop in peroxide level at all temperatures, forming bellshaped curves with time. In contrast, peroxide formation in samples under nitrogen was negligible (Fig. 1). In fact, the peroxide level declined slightly in the first few weeks, especially at 608C and after week 4, the peroxide level started to increase. The slight increase could be due to trace amount of oxygen, which was present in
Figure 1. Effect of temperature and air on the formation of peroxides in neat polysorbate 80 during incubation. Key: }—under nitrogen at 408C; *—under nitrogen at 508C; &—under nitrogen at 608C; ^— under air at 408C; *—under air at 508C; and &—under air at 608C. The error bars represent the standard deviation of values determined from three separate vials and, if not shown, are smaller than the symbols.
nitrogen and/or permeated through the rubber stoppers with time. The nitrogen had a purity of 99.99% and the residual oxygen amount would be approximately 5 nmol, assuming the impurity gas had 20% oxygen. The above results suggest that oxygen in air is the major cause of peroxide formation in neat polysorbate 80 and by removing oxygen, peroxide formation in polysorbate 80 may be inhibited completely during long-term storage. As polysorbate 80 is generally used from a concentrated stock solution, we prepared a 20% aqueous solution and examined the peroxide formation of this solution under similar incubation conditions as for neat polysorbate 80 (Fig. 2). Several differences were observed. First, the initial formation of peroxides in 20% polysorbate 80 solution under air was faster, and the peroxides reached a much higher peak level (approximately 10 times) than that in neat polysorbate 80. Second, there was a clear trend of increasing lag time with decreasing temperature, which resulted in delayed peaking time. Third, the peroxide level did not reach a plateau at 408C even after 8 weeks. Fourth, there was a second increase in peroxide level after week 6 at 508C. Last, the 20% solution (with a starting peroxide concentration of 0.044 mEq) under nitrogen showed a weak gradual upward trend in peroxide concentration but JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
2256
HA, WANG, AND WANG
Figure 2. Effect of temperature and air on the formation of peroxides in 20% polysorbate 80 solution during incubation. Key: }—under nitrogen at 408C; *—under nitrogen at 508C; &—under nitrogen at 608C; ^—under air at 408C; *—under air at 508C; and &—under air at 608C. The error bars represent the standard deviation of values determined from three separate vials and, if not shown, are smaller than the symbols.
remained below 0.4 mEq during the course of the study. The slight increase in peroxide under nitrogen may have to do with the residual oxygen dissolved in the polysorbate 80 solution. Again, these results show that peroxide formation in polysorbate 80 solutions can be effectively inhibited by removing oxygen. The above results indicate that peroxides in polysorbate 80 decompose with time. Because formation and decomposition of peroxides occurred simultaneously under air, the above results could not tell us the true decomposition rate of peroxides. Therefore, we examined the decomposition rate of peroxides in 20% polysorbate 80 solutions under vacuum in dark at 40 or 608C. We chose vacuum condition instead of nitrogen fill because we wanted to reduce the residual oxygen in the sample vials to a minimum to prevent any formation of peroxides. The 20% polysorbate 80 solution had been aged and contained a starting peroxide level of 1 mEq. Upon incubation, the peroxide level dropped at both temperatures (Fig. 3). However, the decline was much faster at 608C than at 408C, and a significant lag time was observed at 408C. In addition, no significant drop was observed at 408C after week 4, suggesting that peroxides below a certain level could be stable at a lower temperaJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
Figure 3. Effect of temperature on the decomposition of peroxides in 20% polysorbate 80 solution under vacuum during incubation. Key: *—408C; and &— 608C. The error bars represent the standard deviation of values determined from three separate vials and, if not shown, are smaller than the symbols.
ture. The higher decomposition rate of peroxides at a higher temperature may explain why a lower level of peroxides accumulated during incubation at a higher incubation temperature both for neat and aqueous solution of polysorbate 80 (Figs. 1 and 2). Factors Affecting Peroxide Formation in Polysorbate 80 Light is a very common factor in controlling stability of chemicals, and has been shown to affect the peroxide formation in polysorbate 20.13 Therefore, we examined the effect of light on the peroxide formation in polysorbate 80. Instead of using neat polysorbate 80, we prepared 20% polysorbate 80 solutions for stability studies because the peroxide formation is faster and any effect can be observed in a shorter period of time. As expected, light dramatically accelerated the formation of peroxides during incubation at 408C (Fig. 4). By the end of a 5-week incubation period, the peroxide level reached 1300 mEq under air, which was eight times higher than that under air but without light exposure. In the absence of air, however, light exposure did not cause any increase in peroxide level (Fig. 4). These results indicate that exposure of polysorbate 80 to light may not cause formation of peroxides in the absence of air.
PEROXIDE FORMATION IN POLYSORBATE 80
Figure 4. Effect of light on the formation of peroxides in 20% polysorbate 80 solution during incubation at 408C. Key: }—light under vacuum; *—light under air at 408C; and *—dark under air. The error bars represent the standard deviation of values determined from three separate vials and, if not shown, are smaller than the symbols.
Effect of Peroxides in Polysorbate 80 on IL-2 Mutein Stability in Liquid State The above data demonstrate that peroxides can be easily generated in neat or diluted polysorbate 80 during storage under air. As mentioned earlier, polysorbate 80 has been frequently used in processing or formulating proteins. An obvious question is whether the peroxide level in polysorbate 80 can cause any significant damage during processing or storage of a formulated protein. Therefore, we chose a model protein, IL-2 mutein, and examined the potential oxidative effect of peroxides in polysorbate 80 on formulated IL-2 mutein during storage. This protein has been reported to have potent antitumor activity and better tolerability in vivo than Proleukin, a commercial IL-2 product.24 Additional reasons for choosing this protein include (1) oxidation at Met104 could easily occur in recombinant IL-225 and desAla1Ser125 IL-226 during storage at room temperature; (2) peroxides specifically catalyze methionine oxidation,27 and any oxidative effect of peroxide-containing polysorbate 80 can be easily assessed; and (3) the oxidized IL-2 can be easily separated and quantitated by RP-HPLC (Fig. 5). Oxidation of the IL-2 mutein was confirmed by MS analysis of the RP-HPLC eluate and both MW þ 18 (major) and MW þ 32 were detected, although the site of oxidation was not verified. Because the level of polysorbate 80 in protein formulations is usually low (typically below 1%),
2257
the residual level of peroxides in polysorbate 80 may or may not have a significant effect on protein oxidation. To answer this question, we prepared liquid IL-2 mutein formulation containing both fresh (low-peroxide) and stressed (high-peroxide) polysorbate 80 at 0.1%. The peroxide level in the protein formulations was determined to be, respectively, 0.33 and 25 mEq. Incubation was conducted at 408C under air or nitrogen. The starting material of IL-2 mutein contained 2.8% of oxidized IL-2 mutein, which was formed presumably during the fermentation and/or purification process. Under nitrogen, oxidized IL-2 mutein increased to 5.4% by the end of a 30-day incubation period (Fig. 6). In contrast, air caused an increase in the amount of oxidized IL-2 mutein to 10.9% in the same period, suggesting that oxygen in air catalyzed IL-2 mutein oxidation. The slight increase in oxidized IL-2 mutein under nitrogen may reflect a combined effect of dissolved oxygen and the residual oxygen present in nitrogen. As calculated before, the nitrogen may contain approximately 5 nmol of oxygen. Because the amount of IL-2SA in the sample vials was approximately 66 nmol, up to 7.6% of IL-2 mutein could be potentially oxidized by the residual oxygen if their reaction molar ratio was 1:1. In the presence of low-peroxide polysorbate 80, the base-line oxidation of IL-2 mutein did not change under air or under nitrogen to a significant degree, suggesting that the amount of peroxides in polysorbate 80 did not induce any significant oxidative effect. However, the amount of oxidized IL-2 mutein in the presence of low-peroxide polysorbate 80 at the end of the 30-day incubation period was slightly higher than the control, suggesting possible effect of peroxides generated during storage. In contrast, high-peroxide polysorbate 80 caused a rapid increase in IL-2 mutein oxidation during the initial incubation period. The rate of oxidation was the same either under air or nitrogen within 4 days, suggesting the peroxides in polysorbate 80 were responsible for the IL-2 mutein oxidation. After day 4, oxidation of IL-2 mutein in these samples became slower under nitrogen, which may reflect a gradual depletion of peroxides in the formulation. After 1 week, oxidation of IL-2 mutein under air also started to slow down, indicating a diminishing contribution of peroxide-induced IL-2 mutein oxidation relative to that due to molecular oxygen. These results indicate that the oxidative effect of peroxides in polysorbate 80 may or may not be an issue, depending on its relative quantity in a protein formulation. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
2258
HA, WANG, AND WANG
Figure 5. Separation and quantitation of IL-2 mutein and its oxidized product by reverse-phase HPLC (see details under Materials and Methods). The retention time for IL-2 mutein and its oxidized product was, respectively, 13.8 and 13.0 min.
Effect of Peroxides in Polysorbate 80 on IL-2 Mutein Stability during Lyophilization and in Solid State
Figure 6. Formation of oxidized IL-2 mutein in solution during incubation at 408C. The protein solution contained 1.0 mg/mL IL-2 mutein, 5% mannitol, and 20 mM citrate at pH 5.5 with and without 0.1% polysorbate 80. Key: *—no polysorbate 80 under nitrogen; *—no polysorbate 80 under air; }—lowperoxide polysorbate 80 under nitrogen; ^—low-peroxide polysorbate 80 under air; &—high-peroxide polysorbate 80 under nitrogen; and &—high-peroxide polysorbate 80 under air. The error bars represent the standard deviation of values determined from three separate vials and, if not shown, are smaller than the symbols. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
Lyophilization has been used frequently to extend the shelf-life of a protein formulation. Therefore, we also examined the potential oxidative damage caused by peroxides in polysorbate 80 on lyophilized IL-2 mutein with the same composition as the liquid formulation. Although not as significant as in liquid state, air-induced oxidation of IL2 mutein also occurred in the solid state and 5.5% of IL-2 mutein was oxidized by the end of 9-week incubation period (Fig. 7). In contrast to the liquid state, IL-2 mutein oxidation in solid state was significantly increased in the presence of lowperoxide polysorbate 80 and the oxidation rate gradually decreased to that of the air-induced oxidation by week 7. A surprising finding was that high-peroxide polysorbate 80 caused a significant formation of oxidized IL-2 mutein during preparation (lyophilization) as well as increased oxidation during incubation. After lyophilization, the oxidized IL-2 mutein jumped from 3 to 5%. Addition of glutathione did not change the amount of oxidized IL-2 mutein formed during lyophilization, although this antioxidant significantly reduced the rate of IL-2 mutein oxidation during incubation. Under vacuum, high-peroxide polysorbate 80 also accelerated IL-2 mutein oxidation as expected. However, the amount of oxidized IL-2 mutein reached a maximum at week 4, suggesting
PEROXIDE FORMATION IN POLYSORBATE 80
Figure 7. Formation of oxidized IL-2 mutein in a lyophilized formulation during incubation at 408C. The formulation before lyophilization contained 1.0 mg/mL IL-2 mutein, 5% mannitol, and 20 mM citrate at pH 5.5 with and without 0.1% polysorbate 80. Key: *—no polysorbate 80 under air; ^—low-peroxide polysorbate 80 under air; &—high-peroxide polysorbate 80 under air; &—high-peroxide polysorbate 80 under vacuum; and ~—high-peroxide polysorbate 80 under air plus 0.5% GSH. The error bars represent the standard deviation of values determined from three separate vials and, if not shown, are smaller than the symbols.
that the peroxide-induced IL-2 mutein oxidation, if any, reached a nondetectable level after week 4. The rapid formation of oxidized IL-2 mutein during lyophilization prompted us to investigate each lyophilization step where IL-2 mutein can possibly be oxidized. To magnify the oxidizing effect, we prepared IL-2 mutein formulation containing hydrogen peroxide at 50 mM instead of high-peroxide polysorbate 80. We sampled each step during lyophilization and found that the total amount of oxidized IL-2 mutein after freezing, annealing, and drying was, respectively, 4.8, 17.4, and 17.4%. It is clear that oxidation of IL-2 mutein during lyophilization occurred in both freezing and annealing steps and mostly during annealing.
DISCUSSION Peroxides in Polysorbate 80 In this study, we examined the peroxide formation in neat and 20% polysorbate 80. We chose the FOX assay in the determination of peroxides
2259
because the peroxide assay recommended in the European Pharmacopoeia (EP) requires a larger amount of sample (5 g) and offers less sensitivity. Fresh polysorbates usually contain a variable level of residual peroxides depending on the source. We found the peroxide level in polysorbate 80 from J.T. Baker (lot N11662) and Sigma Chemicals (lot 120K7276) was, respectively, 0.16 and 0.33 mEq. Knepp et al.20 reported peroxide levels between 0.76 to 27.8 mEq in polysorbate 80 from several suppliers, including Mazer Chemical, ICI Specialties, Spectrum, Croda, and Emery. The difference in peroxide level in polysorbate 80 may reflect possible differences in the manufacturing and purification (bleaching) processes, packaging, and storage conditions. During incubation at elevated temperatures, both neat and 20% polysorbate 80 experienced a rise and fall in peroxide content with time. Similar observations were also made with other types of nonionic surfactants including 3% polysorbate 20 solutions13 and 3% Cetomacrogol solutions.16 The rise and fall in peroxide content was considered typical of radical chain autoxidation followed by degradation.13 Autoxidation is a term for the uncatalyzed oxidation of a substrate by molecular oxygen,27 and our data clearly indicate the participation of air (oxygen) in the peroxide formation. A chain autoxidation process consists of three phases: initiation, propagation, and termination. The increased formation of peroxides in polysorbate 80 under light suggests photocatalyzed initiation of peroxide formation process, probably through the conversion of triplet to singlet oxygen27,28 and/or light-induced decomposition of trace amounts of peroxides, triggering the chain oxidation. The changing time course of peroxide formation in polysorbate 80 reflects a relative balance between simultaneous formation and decomposition of peroxides. The initial increase was due to a faster rate of peroxide formation than that of decomposition. At the peak level, the rate of formation and decomposition reached a steady state. The decline in peroxide level indicates a faster decomposition of peroxides than peroxide formation. These results suggest that the level of peroxides in polysorbate 80 can change dramatically under different storage conditions and the level of residual peroxides in polysorbate 80 is not a reliable indicator of the age or past storage condition of polysorbate 80. Although peroxide formation in surfactants can be inhibited to a great degree by including an antioxidant such as butylated hydroxytoluene,14,29 our data JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
2260
HA, WANG, AND WANG
indicate that removal of oxygen is simple and effective in preventing peroxide formation in polysorbate 80 during long-term storage. Peroxide formation in polysorbate 80 experienced a temperature-dependent lag time during incubation. This lag time was also termed as the induction period and considered to be the time period when the peroxide level is below 5 mEq.13 Assuming this level demarcates the end of a lag time, we found a linear Arrhenius-type relationship between the lag time and temperature in the peroxide formation in 20% polysorbate 80 solution (Fig. 8). Based on this relationship, the extrapolated lag time at 258C is 340 days. Such reverse relationships between lag time and temperature were also observed with other types of non-ionic surfactants, such as polysorbate 20.13 In another study with 3% aqueous Cetomacrogol solution, Hamburger et al.16 demonstrated a linear relationship between the lag time and incubation temperature between 40 to 908C and the extrapolated lag time at 258C was approximately 120 days. The difference in the extrapolated lag time between polysorbate 80 and Cetomacrogol is likely due to the differences in structure and concentration used in these studies. The shorter lag time at higher temperatures was believed to be due to a faster formation of free radicals.16 Another finding is that the peroxide formation in 20% solution was much faster than that in neat polysorbate 80. This observation agrees with the
Figure 8. Effect of temperature on the lag time of peroxide formation in 20% polysorbate solution during storage. The lag time was the duration of time in days when the peroxide level was below 5 mEq. It was estimated assuming a linear relationship between each data point. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
trend of other type of polyethylene surfactants at different aqueous solutions. The peroxide formation rate in Cetomacrogol was inversely proportional to the surfactant concentration in the range of 3 to 20%, although no explanations were offered.16 To explain the rate difference in peroxide formation between neat polysorbate 80 and its aqueous solution, we determined the content of dissolved oxygen in polysorbate 80 solutions at different concentrations up to 25% (high viscosity restricted analysis at higher concentrations). The concentration of dissolved oxygen decreased linearly with increasing polysorbate 80 concentrations in the range tested (data not shown). Therefore, the higher rate of peroxide formation in 20% polysorbate 80 solutions is at least partially due to the higher amount of dissolved oxygen, which is a reactant for peroxidation. In addition, peroxidation could be diffusion controlled, as the reactions between radicals and oxygen were found to be slower in more viscous solvents.30 Polysorbate 80 is a viscous liquid and thus, diffusion-controlled reactions are expected to be slow in neat polysorbate 80. These results suggest that polysorbates may be preferably stored as neat or concentrated solution to minimize the rate of peroxide formation. Peroxides in Polysorbate 80 and Protein Stability in Liquid State An IL-2 mutein was used to test the effect of residual peroxides in polysorbate 80 on protein stability. In the presence of 0.1% low-peroxide polysorbate 80, oxidation of IL-2 mutein in liquid state was not affected to a significant degree either under air or nitrogen at 408C for at least 14 days. This suggests that peroxides were not generated to a significant level to cause additional oxidation of IL-2 mutein. In contrast, high-peroxide polysorbate 80 at the same concentration caused a rapid increase in IL-2 mutein oxidation under the same conditions. The starting peroxide concentration in the protein stability samples containing low- and high-peroxide polysorbate 80 was, respectively, 0.33 and 25 mEq and that of IL-2 mutein was around 66 mM. Assuming the reaction stoichiometry between peroxides and IL-2 mutein was 1:1, the maximum percentage of oxidized IL-2 mutein generated through peroxides would be 0.5% in the presence of low-peroxide polysorbate 80, but 38% in the presence of high-peroxide polysorbate 80. Given the variation of the RP-HPLC, detection of a 0.5% change in oxidized IL-2 mutein
PEROXIDE FORMATION IN POLYSORBATE 80
is difficult. Because oxidation of IL-2 mutein was also observed in samples containing no polysorbate 80 under nitrogen or air, the total amount of oxidized IL-2 mutein in the presence of highperoxide polysorbate 80 is a sum of both oxygeninduced and peroxide-induced oxidation processes. After subtracting the contribution from oxygeninduced oxidation under nitrogen or air, the net percentage of oxidized IL-2 mutein induced by peroxides in the presence of high-peroxide polysorbate 80 can be obtained (Fig. 9). The maximum percentage of oxidized IL-2 mutein induced by peroxides was extrapolated, through nonlinear regression analysis, to be 31 and 36%, respectively, under nitrogen and air. This suggests that the reaction stoichiometry between peroxides in polysorbate 80 and IL-2 mutein is likely to be 1:1. The lower percentage under nitrogen could be due to decomposition of a small portion of peroxides. Peroxides in Polysorbate 80 and Protein Stability in Solid State As mentioned, many unstable proteins need to be lyophilized for long-term storage. Yet, protein degradation does occur in solid state including oxidation (see reviews by Lai31 and Wang32).
Figure 9. Formation of oxidized IL-2 mutein in solution during incubation at 408C. The protein solution contained 1.0 mg/mL IL-2 mutein, 5% mannitol, and 20 mM citrate (pH 5.5) and 0.1% high-peroxide polysorbate 80. The amount of peroxide-induced IL-2 mutein oxidation was the difference in oxidized protein in the presence and absence of high-peroxide polysorbate 80. Key: &—under air; &—under vacuum.
2261
Examples include oxidation of methionine in lyophilized human growth hormone in a vial containing only 0.4% oxygen during storage at 258C,33 in lyophilized recombinant human insulin-like growth factor I (hIGF-I ),34 and in lyophilized recombinant IL-2.26 In fact, the oxidation rate of lyophilized hIGF-I under air was rather similar to that in a solution at either 25 or 308C.34 Our study clearly show that IL-2 mutein was also oxidized in the lyophilized state under air during incubation at 408C. However, the oxidation rate was much slower than that in the liquid state. Although inclusion of low-peroxide polysorbate 80 did not change the baseline oxidation of IL-2 mutein in liquid state, it caused a significant increase in oxidation in solid state. We attribute this effect to the concentration of peroxides during the lyophilization process. It is well known that freezing can concentrate all solutes in a solution as water freezes to form ice crystals. The dissolved oxygen can be concentrated more than 1000 times during freezing.35 Therefore, it is conceivable that the concentrated peroxides can readily react with IL-2 mutein at elevated temperatures. Freezinginduced concentration of peroxides and protein also occurred in high-peroxide polysorbate 80 samples, so much so that some of the IL-2 mutein was oxidized during the lyophilization process. Inclusion of an antioxidant GSH could not inhibit the oxidation process during lyophilization, although it did exert a significant inhibition effect during incubation (Fig. 7). Chang and Bock29 showed that oxidizing species in nonionic surfactants, including polysorbate 80, could easily oxidize sulfhydryl groups. Therefore, peroxides in polysorbate 80 could oxidize sulfhydryl groups in GSH during incubation. Consequently, IL-2 mutein was spared. Even though GSH offered significant antioxidation effect during incubation, the amount of oxidized IL-2 mutein in GSHcontaining samples was still higher than those under vacuum after incubation for 2 weeks. This clearly indicates that GSH could not protect IL-2 mutein completely during incubation. Because the amount of GSH was in excess (16 mM) in the formulation relative to IL-2 mutein, the incomplete protection during incubation and the nonprotection during lyophilization suggest that oxidation of IL-2 mutein might be easier than oxidation of GSH under these conditions. Under vacuum, the oxidation of IL-2 mutein in the presence of high-peroxide polysorbate 80 was much slower than under air, suggesting that either the peroxides in the solid state might not be as JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
2262
HA, WANG, AND WANG
reactive as in the liquid state in the absence of air, and/or oxygen facilitated the peroxide-induced oxidation process in the solid state. The latter interpretation seems more plausible, as the oxidized IL-2SA in the presence of low-peroxide polysorbate 80 increased by more than 7% during incubation, which is much higher than what was predicted based on 1:1 reaction ratio. Currently, there is no specification on the limit of residual peroxides in polysorbate 80 both in the USP and the JP. Although the limit is specified in EP, the maximum allowable amount is 10 in terms of peroxide value. In our study, the low-peroxide polysorbate 80 has a peroxide level of 0.33 mEq, which is 0.66 peroxide value as defined in EP. Because such a low level of peroxides in polysorbate 80 accelerated IL-2 mutein oxidation in solid state, any use of peroxide-containing polysorbate 80, that meets the EP requirement, can cause potential protein instability at least in the solid state. Our data demonstrated that the peroxideinduced oxidation of IL-2 mutein during the lyophilization process occurred mainly in the annealing step, which is a temperature holding step at 208C for 1 h post deep freezing at 458C. The IL-2 mutein formulation contains 5% mannitol and 20 mM citric acid (pH 5.5) as a buffering agent. The glass transition temperature (Tg0 ) of mannitol was reported to be around 308C.36,37 Annealing the formulation at 208C is likely above the glass transition temperature, which could promote crystallization of mannitol and cause further concentration of solutes, including peroxides and oxidation of IL-2 mutein. Another possibility is that annealing caused a change in IL-2 mutein conformation and increased peroxides’ accessibility to the oxidation site. Indeed, several proteins such as b-galactosidase and LDH were destabilized during the annealing process in mannitol-containing formulations.38,39 These results suggest that mannitol may not be the best formulation excipient for IL-2 mutein, especially in a peroxide-containing formulation during lyophilization. Other protein stabilizers, particularly the ones that do not easily crystallize or phase-separate during lyophilization, could potentially minimize the peroxideinduced oxidation of IL-2 mutein.
CONCLUSIONS For the first time, we demonstrated that peroxides can be easily generated in neat polysorbate JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
80 in the presence of air during incubation at elevated temperatures. Dilution of polysorbate 80 resulted in a faster rate of peroxide formation and a greater amount of peroxides during incubation, which is further promoted/catalyzed by light. Peroxide formation can be blocked or greatly inhibited by preventing any contact with air/oxygen during storage. IL-2 mutein, a model protein in this study, can be easily oxidized both in liquid and solid states under air. A lower level of peroxides in polysorbate 80 did not change the rate of IL-2 mutein oxidation in liquid state but significantly accelerated its oxidation in solid state. A higher level of peroxides in polysorbate 80 caused a significant increase in IL-2 mutein oxidation both in liquid and solid states, and GSH can significantly inhibit the peroxide-induced oxidation of IL-2 mutein during incubation. In addition, a higher level of peroxides in polysorbate 80 caused immediate IL-2 mutein oxidation during annealing in lyophilization, suggesting that implementation of an annealing step needs to be carefully evaluated in the development of a lyophilization process for oxidation-sensitive proteins.
ACKNOWLEDGMENTS The authors sincerely thank Dr. Shian-Jiun Shih for close collaboration, Bruce Gardner for technical support in lyophilization, Mike Shearer for help with sample filling and capping, and other formulation group members for assistance in various aspects of the study. We are also grateful to Drs. Sheryl Martin-Moe, Bob Kuhn, and Michael Coan for their critical review of the manuscript and their support of this project.
REFERENCES 1. Saha P, Kou JH. 2000. Effect of solubilizing excipients on permeation of poorly water-soluble compounds across caco-2 cell monolayers. Eur J Pharm Biopharm 50:403–411. 2. Trotta M, Gallarate M, Patterino F, Morel S. 2001. Emulsions containing partially water-miscible solvents for preparation of drug nanosuspensions. J Control Release 76:119–128. 3. Kim CK, Cho YJ, Gao ZG. 2001. Preparation and evaluation of biphenyl dimethyl dicarboxylate microemulsions for oral delivery. J Control Release 70:149–155. 4. Gombotz WR, Pankey SC, Bouchard LS, Phan DH, MacKenzie AP. 1996. Stability, characterization,
PEROXIDE FORMATION IN POLYSORBATE 80
5.
6.
7.
8.
9.
10.
11.
12. 13. 14.
15.
16.
17.
18.
19.
20.
formulation and delivery system development for transforming growth factor-beta1. In: Pearlmean R, Wang YJ, editors. Formulation, characterization, and stability of protein drugs. New York: Plenum Press, pp 219–245. Bam L, Cleland JL, Randolph TW. 1996. Molten globule intermediate of recombinant human growth hormone: Stabilization with surfactants. Biotechnol Prog 12:801–809. Katakam M, ell LN, Banga AK. 1995. Effect of surfactants on the physical stability of recombinant human growth hormone. J Pharm Sci 84:713–716. Chang BS, Kendrick BS, Carpenter JF. 1996. Surface-induced denaturation of proteins during freezing and its inhibition by surfactants. J Pharm Sci 85:1325–1330. Sarciaux JM, Mansour S, Hageman MJ, Nail SL. 1999. Effects of buffer composition and processing conditions on aggregation of bovine IgG during freeze-drying. J Pharm Sci 88:1345–1361. Zhang MZ, Wen J, Arakawa T, Prestrelski SJ. 1995. A new strategy for enhancing the stability of lyophilized protein: The effect of the reconstitution medium on keratinocyte growth factor. Pharm Res 12:1447–1452. Zhang MZ, Pikal K, Nguyen T, Arakawa T, Prestrelski SJ. 1996. The effect of the reconstitution medium on aggregation of lyophilized recombinant interleukinl-2 and ribonuclease a. Pharm Res 13:643–646. Son K, Kwon C. 1995. Stabilization of human epidermal growth factor (hEGF) in aqueous formulation. Pharm Res 12:451–454. Physician’s Desk Reference, 53rd ed. 1999. Montvale, NJ: Medical Economics Company, Inc. Donbrow M, Azaz E, Pillersdorf A. 1978. Autoxidation of polysorbates. J Pharm Sci 67:1676–1681. Jaeger J, Sorensen K, Wolff SP. 1994. Peroxide accumulation in detergents. J Biochem Biophys Methods 29:77–81. Bates TR, Nightingale CH, Dixon E. 1973. Kinetics of hydrolysis of polyoxyethylene (20) sorbitan fatty acid ester surfactants. J Pharm Pharmacol 25:470– 477. Hamburger R, Azaz E, Donbrow M. 1975. Autoxidation of polyethylenic non-ionic surfactants and of polyethylene glycols. Pharm Acta Helv 50:10–17. Azaz E, Donbrow M, Hamburger R. 1975. Incompatibility of non-ionic surfactants with oxidisable drugs. Pharm J 211:15. Coates LV, Pahley MM, Tattersall K. 1961. The stability of antibacterials in polyethylene glycol mixtures. J Pharm Pharmacol 13:620–624. James KC, Leach RH. 1970. The stability of chloramphenicol in topical formulations. J Pharm Pharmacol 22:607–611. Knepp VM, Whatley JL, Muchnik A, Calderwood TS. 1996. Identification of antioxidants for prevention
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
2263
of peroxide-mediated oxidation of recombinant human ciliary neurotrophic factor and recombinant human nerve growth factor. J Pharm Sci Technol 50:163–171. Herman AC, Boone TC, Lu HS. 1996. Characterization, formulation, and stability of neupogen (filgrastim), a recombinant human granuloctycolony stimulating factor. In: Pearlman R, Wang YJ, editors. Formulation, characterization, and stability of protein drugs. New York: Plenum Press, pp 303–328. Miki T, Orii Y. 1985. The reaction of horseradish peroxidase with hydroperoxides derived from triton x-100. Anal Biochem 146:28–34. Johnson DM, Taylor WF. 1984. Degradation of fenprostalene in polyethylene glycol 400 solution. J Pharm Sci 73:1414–1417. Shanafelt AB, Lin Y, Shanafelt MC, Forte CP, Dubois-Stringfellow N, Carter C, Gibbons JA, Cheng SL, Delaria KA, Fleischer R, Greve JM, Gundel R, Harris K, Kelly R, Koh B, Li Y, Lantz L, Mak P, Neyer L, Plym MJ, Roczniak S, Serban D, Thrift J, Tsuchiyama L, Wetzel M, Wong M, Zolotorev A. 2000. A t-cell-selective interleukin 2 mutein exhibits potent antitumor activity and is well tolerated in vivo. Nat Biotechnol 18:1197– 1202. Sasaoki K, Hiroshima T, Kusumoto S, Nishi K. 1992. Deamidation at asparagine-88 in recombinant human interleukin 2. Chem Pharm Bull 40:976–980. Hora MS, Rana RK, Wilcox CL, Katre NV, Hirtzer P, Wolfe SN, Thomson JW. 1992. Development of a lyophilized formulation of interleukin-2. Dev Biol Stand 74:295–303. Hovorka S, Schoneich C. 2001. Oxidative degradation of pharmaceuticals: Theory, mechanisms and inhibition. J Pharm Sci 90:253–269. Ralph Rawls H, van Santen PJ. 1970. A possible role of singlet oxygen in the initiation of fatty acid autoxidation. J Am Oil Chemists Soc 47:121— 125. Chang HW, Bock E. 1980. Pitfalls in the use of commercial nonionic detergents for the solubilization of integral membrane proteins: Sulfhydryl oxidizing contaminants and their elimination. Anal Biochem 104:112–117. Maillard B, Ingold KU, Scaiano JC. 1983. Rate constants for the reactions of free radicals with oxygen in solution. J Am Chem Soc 105:5095–5099. Lai MC, Topp EM. 1999. Solid-state chemical stability of proteins and peptides. J Pharm Sci 88:489–500. Wang W. 2000. Lyophilization and development of solid protein pharmaceuticals. Int J Pharm 203:1– 60. Pikal M, Dellerman K, Roy M. 1992. Formulation and stability of freeze-dried proteins: Effects of
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
2264
HA, WANG, AND WANG
moisture and oxygen on the stability of freeze-dried formulations of human growth hormone. Dev Biol Stand 74:21–37. 34. Fransson J, Florin-Robertson E, Axelsson K, Nyhlen C. 1996. Oxidation of human insulin-like growth factor i in formulation studies, ii. Effects of oxygen, visible light, and phosphate on methionine oxidation in aqueous solution and evaluation of possible mechanisms. Pharm Res 13:1252–1257. 35. Wisniewski R. 1998. Developing large-scale cryopreservation systems for biopharmaceutical products. Bio Pharm 11:50–60. 36. Meredith P, Donald AM, Payne RS. 1996. Freezedrying: In situ observations using the cryoenviron-
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
mental scanning electron microscopy and differential scanning calorimetry. J Pharm Sci 85:631–637. 37. Lueckel B, Helk B, Bodmer D, Leuenberger H. 1998. Effects of formulation and process variables on the aggregation of freeze-dried interleukin-6 (il-6) after lyophilization and on storage. Pharm Dev Technol 3:337–346. 38. Izutsu K, Yoshioka S, Terao T. 1993. Decreased protein-stabilization effects of cryoprotectantss due to crystallization. Pharm Res 10:1232–1237. 39. Izutsu K, Yoshioka S, Terao T. 1994. Effect of mannitol crystallinity on the stabilization of enzymes during freeze-drying. Chem Pharm Bull (Tokyo) 42:5–8.