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 exa-
`mined 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 concentra-
`tions 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: wei.wang.b@bayer.com)
`
`Journal of Pharmaceutical Sciences, Vol. 91, 2252–2264 (2002)
`ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association
`
`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 aggre-
`gation under various processing conditions, such
`as refolding,5 mixing,6 freeze thawing,7 freeze
`drying,8 and reconstitution.9,10 Even chemical de-
`gradation 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 polysor-
`bates as inactive pharmaceutical
`ingredients,
`including Actimmune, Activase, Intron A, and
`Recombinate.12
`However, there have been concerns about use
`of polysorbates in pharmaceutical preparations
`
`2252
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
`
`Ex. 2020-0001
`
`

`
`because polysorbates contain low levels of resi-
`dual 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 polyox-
`yethylene 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 ex-
`tensively 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 demon-
`strated that alkyl hydroperoxides in polysorbate
`80 induced oxidation, dimerization, and subse-
`quent 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 sto-
`rage. Another paper by Miki et al.22 reported a
`linkage between the oxidation of hydroperoxidase
`and peroxides generated in Triton X-100, a sur-
`factant structurally similar to polysorbates. No
`
`PEROXIDE FORMATION IN POLYSORBATE 80
`
`2253
`
`reports, however, have been found on the poten-
`tial 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 oxidation-
`sensitive 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. John-
`son 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 peroxida-
`tion in polysorbate 80 can be effectively inhibited
`by reducing its contact with molecular oxygen. A
`model protein, IL-2 mutein, was used to demon-
`strate that peroxides generated in polysorbate 80
`could accelerate protein oxidation both in liquid
`and solid states. Finally, we show that peroxide-
`induced IL-2 mutein oxidation can be effectively
`prevented/inhibited either by reducing the con-
`tact 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
`
`Ex. 2020-0002
`
`

`
`2254
`
`HA, WANG, AND WANG
`
`solution) were purchased from Sigma Chemicals
`(St. Louis, MO). HPLC-grade water and acetoni-
`trile were obtained from Fisher Scientific (Pitts-
`burgh, PA). High-purity nitrogen (99.99%) was
`obtained from Air Products (Galt, CA). Recombi-
`nant human IL-2 mutein at 3.5 mg/mL in a frozen
`buffered solution was prepared by Bayer Corpo-
`ration. 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 con-
`tent 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 per-
`oxide was used to prepare standard curves and
`therefore, peroxide level in polysorbate 80 was
`obtained as peroxide equivalent to H2O2 stan-
`dards. All neat polysorbate 80 samples were di-
`luted to 20% solution with water before analysis.
`The peroxide level in neat polysorbate 80 samples
`was calculated as milliequivalents (mEq) or micro-
`equivalents (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 (calcu-
`lated 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. Sta-
`bility 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 atmosphe-
`ric pressure or directly sealed with bromobutyl
`rubber stoppers inside the freeze dryer. The stop-
`pered vials were then capped manually. These
`stability samples were incubated at 40, 50, or
`608C with or without light for 8 weeks. The light-
`ing 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% low-
`peroxide polysorbate 80, or 0.1% high-peroxide
`polysorbate 80. The peroxide level in the protein
`solution containing low- and high-peroxide poly-
`sorbate 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 dis-
`pensed 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 stabi-
`lity samples containing high-peroxide polysorbate
`80. Lyophilization included the following steps:
`freezing to 458C at 0.48C/min, holding the tem-
`perature at 458C for 1 h, increasing the tem-
`perature to 208C at 0.48C/min, holding the
`temperature at 208C for 1 h (annealing), de-
`creasing 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
`
`Ex. 2020-0003
`
`

`
`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 oxidiz-
`ed 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 sol-
`vent 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 reten-
`tion 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 re-
`ported in the literature. Therefore, we examined
`peroxide formation in neat polysorbate 80 under
`air or nitrogen. To accelerate formation of pero-
`xides, stability samples were incubated at elevat-
`ed 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 bell-
`shaped curves with time. In contrast, peroxide
`formation in samples under nitrogen was negli-
`gible (Fig. 1). In fact, the peroxide level declin-
`ed 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
`
`PEROXIDE FORMATION IN POLYSORBATE 80
`
`2255
`
`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 for-
`mation in neat polysorbate 80 and by remov-
`ing 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 incuba-
`tion 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 per-
`oxides reached a much higher peak level (approxi-
`mately 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% solu-
`tion (with a starting peroxide concentration of
`0.044 mEq) under nitrogen showed a weak gra-
`dual upward trend in peroxide concentration but
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
`
`Ex. 2020-0004
`
`

`
`2256
`
`HA, WANG, AND WANG
`
`Figure 2. Effect of temperature and air on the for-
`mation 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 inhib-
`ited by removing oxygen.
`The above results indicate that peroxides in
`polysorbate 80 decompose with time. Because for-
`mation 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 want-
`ed 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 tempera-
`
`Figure 3. Effect of temperature on the decomposi-
`tion 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 sta-
`bility of chemicals, and has been shown to affect
`the peroxide formation in polysorbate 20.13 There-
`fore, we examined the effect of light on the per-
`oxide formation in polysorbate 80. Instead of using
`neat polysorbate 80, we prepared 20% polysor-
`bate 80 solutions for stability studies because the
`peroxide formation is faster and any effect can be
`observed in a shorter period of time. As expect-
`ed, 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, how-
`ever, 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.
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 10, OCTOBER 2002
`
`Ex. 2020-0005
`
`

`
`PEROXIDE FORMATION IN POLYSORBATE 80
`
`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 presum-
`ably during the fermentation and/or purification
`process. Under nitrogen, oxidized IL-2 mutein
`increased to 5.4% by the end of a 30-day incuba-
`tion 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 appro-
`ximately 5 nmol of oxygen. Because the amount
`of IL-2SA in the sample vials was approxima-
`tely 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 oxida-
`tion 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. How-
`ever, 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 con-
`trast, high-peroxide polysorbate 80 caused a rapid
`increase in IL-2 mutein oxidation during the ini-
`tial 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 formula-
`tion. After 1 week, oxidation of IL-2 mutein under
`air also started to slow down, indicating a dimi-
`nishing 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
`
`Figure 4. Effect of light on the formation of per-
`oxides 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 process-
`ing or formulating proteins. An obvious question
`is whether the peroxide level in polysorbate 80
`can cause any significant damage during proces-
`sing or storage of a formulated protein. Therefore,
`we chose a model protein, IL-2 mutein, and exa-
`mined the potential oxidative effect of peroxides
`in polysorbate 80 on formulated IL-2 mutein dur-
`ing storage. This protein has been reported to have
`potent antitumor activity and better tolerabi-
`lity 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 oxida-
`tion,27 and any oxidative effect of peroxide-con-
`taining 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 MWþ 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%),
`
`Ex. 2020-0006
`
`

`
`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
`
`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 lyophi-
`lized IL-2 mutein with the same composition as
`the liquid formulation. Although not as signifi-
`cant as in liquid state, air-induced oxidation of IL-
`2 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 low-
`peroxide 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 prepara-
`tion (lyophilization) as well as increased oxidation
`during incubation. After lyophilization, the oxi-
`dized IL-2 mutein jumped from 3 to 5%. Addition
`of glutathione did not change the amount of
`oxidized IL-2 mutein formed during lyophili-
`zation, although this antioxidant significantly re-
`duced the rate of IL-2 mutein oxidation during
`incubation. Under vacuum, high-peroxide poly-
`sorbate 80 also accelerated IL-2 mutein oxidation
`as expected. However, the amount of oxidized IL-2
`mutein reached a maximum at week 4, suggesting
`
`Figure 6. Formation of oxidized IL-2 mutein in
`solution during incubation at 408C. The protein solu-
`tion 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; }—low-
`peroxide polysorbate 80 under nitrogen; ^—low-per-
`oxide 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
`
`Ex. 2020-0007
`
`

`
`PEROXIDE FORMATION IN POLYSORBATE 80
`
`2259
`
`because the peroxide assay recommended in the
`European Pharmacopoeia (EP) requires a larger
`amount of sample (5 g) and offers less sensiti-
`vity. 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 Chemi-
`cal, ICI Specialties, Spectrum, Croda, and Emery.
`The difference in peroxide level in polysorbate 80
`may reflect possible differences in the manu-
`facturing 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% poly-
`sorbate 20 solutions13 and 3% Cetomacrogol solu-
`tions.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 ter-
`mination. The increased formation of peroxides in
`polysorbate 80 under light suggests photocata-
`lyzed initiation of peroxide formation process,
`probably through the conversion of triplet to
`singlet oxygen27,28 and/or light-induced decom-
`position 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 forma-
`tion and decomposition of peroxides. The initial
`increase was due to a faster rate of peroxide for-
`mation 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 con-
`ditions and the level of residual peroxides in poly-
`sorbate 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
`
`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 de-
`viation 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 con-
`taining 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 mu-
`tein during lyophilization occurred in both freez-
`ing and annealing steps and mostly during
`annealing.
`
`DISCUSSION
`
`Peroxides in Polysorbate 80
`
`In this study, we examined the peroxide forma-
`tion in neat and 20% polysorbate 80. We chose the
`FOX assay in the determination of peroxides
`
`Ex. 2020-0008
`
`

`
`2260
`
`HA, WANG, AND WANG
`
`indicate that removal of oxygen is simple and ef-
`fective in preventing peroxide formation in poly-
`sorbate 80 during long-term storage.
`Peroxide formation in polysorbate 80 experi-
`enced 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 relation-
`ship