Polysorbates 20 and 80 Used in the Formulation of Protein Biotherapeutics: Structure and Degradation Pathways

BRUCE A. KERWIN
Department of Pharmaceutics, Amgen Inc., One Amgen Center Dr., Thousand Oaks, California 91320

Received 14 June 2007; revised 9 August 2007; accepted 8 September 2007
Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21190

ABSTRACT: Polysorbates 20 and 80 (Tween1 20 and Tween1 80) are used in the formulation of biotherapeutic products for both preventing surface adsorption and as stabilizers against protein aggregation. The polysorbates are amphipathic, nonionic surfactants composed of fatty acid esters of polyoxyethylene sorbitan being polyoxyethy- lene sorbitan monolaurate for polysorbate 20 and polyoxyethylene sorbitan monooleate for polysorbate 80. The polysorbates used in the formulation of biopharmaceuticals are mixtures of different fatty acid esters with the monolaurate fraction of polysorbate 20 making up only 40–60% of the mixture and the monooleate fraction of polysorbate 80 making up >58% of the mixture. The polysorbates undergo autooxidation, cleavage at
the ethylene oxide subunits and hydrolysis of the fatty acid ester bond. Autooxidation
results in hydroperoxide formation, side-chain cleavage and eventually formation of short chain acids such as formic acid all of which could influence the stability of a biopharmaceutical product. Oxidation of the fatty acid moiety while well described in the literature has not been specifically investigated for polysorbate. This review focuses on the chemical structure of the polysorbates, factors influencing micelle formation and factors and excipients influencing stability and degradation of the polyoxyethylene and fatty acid ester linkages. © 2007 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 97:2924–2935, 2008
Keywords: polysorbate; Tween; surfactants; stability; degradation; micelle; proteins; protein formulation; excipients; degradation

INTRODUCTION

The proper formulation of recombinant proteins for pharmaceutical products is vital to their long-term stability and reproducible activity. To maintain biological activity, proteins generally must be

Abbreviations: HSA, human serum albumin; CMC, critical micelle concentration; PEG, polyoxyethylene glycol.
Correspondence to: Bruce A. Kerwin (Telephone: 805-447- 0712; Fax: 805-447-3401; E-mail: [email protected])
Journal of Pharmaceutical Sciences, Vol. 97, 2924–2935 (2008)
© 2007 Wiley-Liss, Inc. and the American Pharmacists Association
maintained in a specific, three-dimensional con- formation. This conformation is only marginally stable, and thus relatively minor perturbing forces can disrupt protein structure causing loss of biological activity or an immunological response. Such perturbations are commonly encountered as proteins are produced, stored, transported, and delivered to patients. For example, it is well known that during common industrial processes such as filtering,1 storage,2 agitation3,4 freeze/ thawing,5–7 lyophilization,8–10 nebulization,11 and spray drying1,12–17 proteins can suffer damage to their native conformation. Further, delivery of

2924 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 8, AUGUST 2008

protein pharmaceuticals to patients may also provoke losses of conformational integrity via unfavorable interactions of proteins with surfaces (e.g., inner walls of catheter tubing or syringes).18 Surface-active agents, or surfactants, are often added to protein solutions to prevent physical damage during purification, filtration, transporta- tion, freeze-drying, spray drying, storage, and delivery. Human serum albumin (HSA) was initi- ally used in a number of protein formulations as a stabilizer to prevent surface adsorption. Recent concerns though about the presence of infectious agents (e.g., viruses, prions) in blood products has prompted many regulatory agencies to no longer allow its use forcing the exploration of other options. Towards that end polysorbates, nonionic surfac- tants, are increasingly being used as replacements for HSA. The polysorbates are both biocompatible and excellent stabilizers against surface adsorption. However, they must be used with caution since under some conditions there degradation can lead to chemical modification of proteins and interac- tions with some proteins may lead to changes in conformational structure with unknown con- sequences.1,19 To prevent problems from occurring it is important to understand the issues associated with there use including their structure, stability,
and mechanisms through which they may interact with proteins. This article reviews the structure and stability of polysorbates 20 and 80, the most commonly used nonionic surfactants in the biotech- nology industry. For information on the interaction of polysorbate 20 and 80 with proteins the reader is referred to the review by Randolph and Jones.20

POLYSORBATE STRUCTURE

The polysorbates are amphipathic, nonionic sur- factants composed of fatty acid esters of polyoxye- thylene sorbitan.21 Polysorbate 20 (polyoxyethylene sorbitan monolaurate) and Polysorbate 80 (poly- oxyethylene sorbitan monooleate) are the most common polysorbates currently used in formulation of protein biopharmaceuticals.22 Their structures are shown in Figure 1. The structures shown here are those of the chemically homogenous poly- sorbates. Solutions of polysorbate sold by manu- facturers and labeled as polysorbate 20 or 80 are chemically diverse mixtures of different fatty acid esters23–25 and will be discussed later. Both types of polysorbates have a common backbone and only differ between the structures of the fatty acid side chains. The hydrocarbon chains provide the

þ þ þ
Figure 1. Chemical structure of polysorbates 20 and 80. w x y z refers to the total number of oxyethylene subunits on each surfactant molecule and may not exceed 20.

þ þ þ
hydrophobic nature of the polysorbates while the hydrophilic nature is provided by the ethylene oxide subunits. The common backbone structure is a sorbitan ring with ethylene oxide polymers attached at three different hydroxyl positions. While the number of repeat ethylene oxide subunits varies at each position their total number (w x y z) equals 20 and is constant for each polysorbate. The fatty acid moieties are attached through an ester linkage to the ethylene oxide oxygen at the z position. The laurate moiety of polysorbate 20 is a straight chain hydrocarbon structure and the oleate moiety of polysorbate 80 contains a double-bond forming a kink in the hydrocarbon chain.
The polysorbates used in the manufacture of drugs, both biotechnology and pharmaceutical, are sold either as polysorbate 20 or 80 or under the trade names Tween1 20 and Tween1 80. In the literature the names polysorbate and Tween are used interchangeably and both should be searched for retrieving articles from Medline1 or other indexing services. As stated earlier poly- sorbate solutions are generally sold as mixtures of the fatty acid esters.23–25 The European Pharma- copoeia26 defines the percentage of the different fatty acid esters of each polyoxyethylene sorbitan that must be present in each solution. The United States Pharmacopoeia does not define the compo- sition of the fatty acid esters present in the solutions. The compositions of the mixtures and chemical structures are shown in Table 1. As can be seen the lauric acid containing component of polysorbate 20 is 40–60% of the total number of
fatty acid species while the oleic acid containing component of polysorbate 80 is 58% of its total. The remaining fatty acids are a mixture of both saturated and unsaturated fatty acids with caproic, caprylic, capric, and lauric acids being present only in polysorbate 20 with all other fatty acid esters present in both polysorbate
≤
20 and 80 solutions. Analysis of polysorbates from different vendors using reverse phase HPLC coupled with a charged aerosol detector has shown only minor differences in the fatty acid contents between mixtures from different suppli- ers (Sungae Park, personal communication). A supply of polysorbate 80 made of 99% pure oleic acid is now available from NOF Corporation albeit this has not made its way into manufactured products. The majority of polysorbates are produc- ed using strictly plant sources and are the requir- ed polysorbates used in biotechnology products. The information regarding the source of the fatty acids used to manufacture the polysorbates can be obtained from the manufacturer.

SOLUTION AND SURFACE ACTIVE PROPERTIES OF POLYSORBATE

Because of their dual hydrophobic/hydrophilic nature, surfactants in solution tend to orient themselves so that the exposure of the hydrophobic portion of the surfactant to the aqueous solution is minimized.20 In systems containing air/water interfaces, surfactants will tend to accumulate at

Table 1. Fatty Acid Contents of Polysorbate 20 and 80

Acid
EU Specifications

PS-20 (%)a PS-80 (%)b

Structure

Caproic ≤1 CH3(CH2)4COOH
Caprylic ≤10 CH3(CH2)6COOH
Capric ≤10 CH3(CH2)8COOH
Lauric 40–60 CH3(CH2)10COOH
Myristic 14–25 ≤5 CH3(CH2)12COOH Palmitic 7–15 ≤16 CH3(CH2)14COOH
Palmitoleic ≤8 CH3(CH2)5CH – CH(CH2)7COOH
Stearic ≤7 ≤6 CH3(CH2)16COOH
Oleic ≤11 ≤58 CH3(CH2)7CH – CH(CH2)7COOH Linoleic ≤3 ≤18 CH3(CH2)5CH – CH(CH2)7COOH
Linolenic ≤4 CH3(CH2)4CH – CHCH2CH – CH(CH2)7COOH
aEuropean Directorate for the Quality of Medicines and Healthcare. European Pharmacopoeia On-line. 5th Edition 2007, 01/ 2005:0426.
bEuropean Directorate for the Quality of Medicines and Healthcare. European Pharmacopoeia On-line. 5th Edition 2007, 04/ 2006:0428.

these interfaces, forming a surface layer of surfactant oriented in such a fashion that only their hydrophilic ends are exposed to water.20,27,28 Such orientation and hydrophobic surface adsorp- tion can also occur at solid/water interfaces such as those found in vials, syringes, and other glass and plastic containers.20 Surfactant adsorption can be in direct competition with protein adsorption at the same interface possibly resulting in reduced binding of the protein at the interface or less damage to the protein adsorbed at the interface.20 The driving force for the surface adsorption is the hydrophobic effect which in this case can be described by the Gibbs free energy equation for the transfer of the hydrocarbon chain from an organic solvent into water.28–30

DGt ¼ DHt — TDSt
At temperatures near 258C the entropic term, DSt, for a hydrocarbon is large and negative becoming more unfavorable with an increase in the size of the hydrocarbon. The contribution from
the entropic term greatly outweighs that from the enthalpic term, DHt, which is zero at or near 258C for many small hydrocarbons.29,31,32 The tem-
¼
perature at which the DHt 0 is known as the Th, the temperature of minimum solubility. Below this temperature DHt is negative while above Th the DHt is positive and increasing with tempera- ture. Therefore, as DSt becomes less negative with increasing temperature the DHt will continue to increase and account for a greater proportion of the free energy change (DGt).29,31,32 Within the
temperature range of 4–378C used for most liquid
protein formulation studies, the entropy term will be dominant.
Transfer of the long chain hydrocarbons of the polysorbates into water is generally assumed to accompany the ordering of water into cage like structures around the hydrocarbon chains result- ing in a loss of solvent entropy. The hydrophobic effect which drives the adsorption process arises from the fact that water would rather hydrogen bond with itself rather than form cage-like waters around the hydrocarbon chains.29,31,32 Adsorption of the polysorbate at the surface through the hydrocarbon chains results in removal of the hydrophobic surface from the water molecules accompanied by a more favorable entropy con- tribution to the solvent system. Additionally, at increasing polysorbate concentrations this allows for van der Waals contact between the hydro- carbon chains at the interface. Due to the
hydrophobic effect surfactant molecules adsorb at interfaces even at low surfactant concentra- tions. Because of thermal motion there will be a balance of adsorption and desorption such that equilibrium interfacial conditions will take some time to establish.28 An example of this for polysorbate 80 above and below the CMC is shown in Figure 4. The time to establish equilibrium at the interface may be an important factor regard- ing the ability of the polysorbate to prevent surface adsorption and aggregation of proteins during acute stresses such as shaking.
~
Surfactant micelles are spherical aggregates of the polysorbates with the ethylene oxide subunits pointing outwards in contact with the surround- ing solution and the hydrocarbon tails in the center away from the water. The interior proper- ties of the micelles are similar to those of liquid hydrocarbon.28 As the surfactant concentration increases and the surfaces become saturated the monomers in solution associate into micelles. This normally occurs within a narrow concentration range and is referred to as the critical micelle concentration or CMC (Fig. 2). The number of polysorbate molecules per micelle is 50–100 but this value can vary depending on the actual polysorbate concentration and solution condi- tions, that is, temperature, solutes, etc. Similar to adsorption, the main driving force behind

Figure 2. Surface tension measurements of poly- sorbate 20 and 80 and model depicting the relationship of the surfactant molecules coating the air/liquid inter- face. Surface tension was measured in both pure water and containing 140 mM sodium chloride using the ring method with a Kruss Processor Tensiometer K100. Model of surfactants coating the air/liquid interface was adopted from Randolph and Jones20 and Porter.27

micelle formation is the tendency of the hydro- carbon chain of the surfactant to minimize water contact, the hydrophobic effect. The ethy- lene oxide chains hydrogen bond with water maintaining the solubility of the surfactant molecules. Polysorbate 80 which is composed of longer hydrocarbon chains than polysorbate 20 (see preceding discussion) has a more negative DSt resulting in a lower CMC value than that for polysorbate 20. As the temperature increases ethylene oxide containing nonionic surfactants exhibit a monotonic lowering in CMC due to dehydration of the ethylene oxide chains.33 Measurements of the CMC for either polysorbate
20 or 80 at temperatures above or below 258C have
not been reported.
Lowering of the CMC with increasing tempera- ture results in phase separation and is known as the cloud point. At temperatures just below the cloud point, nonionic surfactant micelles exhibit a marked increase in size becoming visible even to the naked eye.33 The clouded out phase also contains substantial levels of water, indicating that complete dehydration is not necessary for phase separation.33 The cloud point temperature
has been reported as 768C for a 3% (w/v) solution of polysorbate 20 in 1 M NaCl and 658C for a 3%
(w/v) solution of polysorbate 80 in 1 M NaCl.34 Recent experiments in our lab (Fig. 3) have confirmed the earlier result for polysorbate 80 but showed an increase in the cloud point for

Figure 3. Influence of salt on the cloud point of poly- sorbate 20 and 80. Cloud point of polysorbate (2% w/v solution) in each salt concentration was measured using a Stanford Research Systems MPA100 automated capillary melting point apparatus. The cloud point was recorded at the initiation of clouding.
polysorbate 20 by approximately 108C, possibly due to the increased quality of the polysorbates available in newer preparations or a difference in the overall hydrocarbon makeup of the surfactant. The early experiments by Donbrow et al.34 in 1978 on the cloud point of polysorbate were done in 1 M NaCl and none have been reported since then. Therefore, we tested the effects of NaCl concen- tration and found that the cloud point was also related to the salt concentration in the aqueous solution. Other excipients such as sugars were not examined. As shown in Figure 3 the cloud point appears inversely proportional to the increasing NaCl concentration for both polysorbate 20 and 80, albeit we were unable to observe a cloud point
at temperatures up to 1008C for polysorbate in
water alone. These results are important as the effect of temperature on protein unfolding using techniques such as differential scanning calori- metry or circular dichroism may be influenced by this phenomenon. No mention of salt effects on the cloud point of polysorbates was found in any pub- lications, although it is known that salt increases the hydrophobic effect.30 Further experiments are required to determine if the effect on the cloud point is being exerted on the ethylene oxide or hydrocarbon chains. Ananthapadmanabhan33 noted that an increase in the alkyl chain length of a given nonionic surfactant lowers the cloud point and would explain the difference in cloud points at a given NaCl concentration between polysorbate 20 and 80. This also suggests that the hydrophobic effect on the hydrocarbon moiety driving micelle formation is significant at the temperature at which clouding occurs.
When the concentration of the surfactant is much lower than its CMC, surfactant molecules lay flat at the air/water and hydrophobic solid/ water interfaces (Fig. 2).20,27 As the concentration is increased, more molecules adsorb to these interfaces, such that the surface concentration remains linearly proportional to the bulk concen- tration.30 This crowding forces the surfactant molecules to order themselves such that the hydrophilic groups are oriented towards the bulk water and the hydrocarbon chains are pointed towards the air or hydrophobic solid.27,35 At sufficiently high surfactant concentrations (i.e., at or above the CMC), there is an oriented monolayer of surfactant molecules and maximum surfactant absorption, at the interface. At the air– liquid interface the surface saturation is respon- sible for the sharp slope change (to essentially zero) observed in experimental plots of surface

properties (surface tension, osmotic pressure) versus surfactant concentration.27 Polysorbates at the hydrophobic solid/water interface have not been well characterized and the possibility remains that a well ordered monolayer as well as hemi-micellar structures are also formed. Both types of structures have been demonstrated for other nonionic surfactants.36 Nonetheless, the surface active property of polysorbate at the air/ liquid and liquid/surface interfaces is the reason they are used in protein formulations to prevent protein aggregation37–41 and surface interac- tion.42–45 Studies using protein films at the air/ liquid interface have shown polysorbate 20 can displace bovine serum albumin46 and b-casein/ b-lactoglobulin mixtures.47–50
CMC values can be determined by a number of techniques including surface tension,43,51,52 osmotic pressure,53 fluorescence,54–59 micellar electrokinetic chromatography,60 calorimetry,61–63 light scattering,64–66 electron paramagnetic reso- nance,67 and analytical ultracentrifugation.68 For illustrative purposes surface tension measurements and dye partitioning will be discussed here. The surface tension method uses a surface tensiometer and measures the point at which the solution surface is saturated with surfactant. As the surfactant coats the air–liquid surface the surface tension drops until it reaches a minimum value then levels out. This occurs as a sharp transition and defines the value for the CMC20,69 being approxi- mately 0.007% (w/v) or 55 mM for polysorbate 2052,70 and 0.0017% (w/v) or 13 mM for polysorbate 80.70,71 This assay does not define the interaction of polysorbate free in solution but indicates the concentration of all surface-active species at the liquid surface interface. Patist et al.70 found that polysorbates containing surface-active impurities in technical grades of surfactants gave anomalous CMC results when measured by the surface tension method. Additionally, since the assay is measuring an equilibrium process between polysorbate free in solution and at the varying interfaces the working CMC value of a polysorbate as it relates to a formulation in a specific container may vary de- pending on the container type and surface to volume ratio. For the dye-binding assay an increase in fluorescence is measured when a hydrophobic compound, such as 1,6-diphenyl 1,3,5-hexatriene partitions from the aqueous phase into the oily phase of the micelle interior.56,70 Chattopadhyay and London56 have optimized the use of this system including effects of probe concentration, incubation and light exposure and their work should be
consulted prior to experimentation with this system. In contrast to the surface-tension measurement, dye partitioning provides a picture of the interactions in solution and shows micelle formation occurring over an extended concentration range of surfactant (Fig. 5). This occurs since the formation of micelles is an equilibrium phenomenon between monomers at the air–liquid interface, free in solution and in micelles.20,27,69,71 By the dye-partitioning method initiation of micelle formation is observed at surfactant concentrations of 0.001% (w/v) for poly- sorbate 80 and 0.002% (w/v) for polysorbate 20. Since the polysorbates are nonionic in nature the presence of electrolytes such as sodium chloride or amino acids do not affect the CMC values of the polysorbates72 as demonstrated by the overlap in the data for the solutions containing salts in both polysorbate 20 and 80 (Fig. 5). An additional factor that may also affect CMC values but has not been described in the literature is the affect of surface to volume ratio of a container.

CHEMICAL STABILITY OF POLYSORBATES

Chemical stability of the polysorbates is another important consideration for their use. The poly- sorbates are notorious for undergoing auto- oxidation34,73–79 and cleavage at the ethylene oxide subunits34,74 as well as hydrolysis of the fatty acid ester bond.34,80 A generalized reaction scheme depicting these processes is shown in Figure 6. Autooxidation of the ethylene oxide

Figure 4. Surface tension measurements of poly- sorbate 80 over time at a fixed polysorbate concentra- tions. Polysorbate 80 in water was mixed for 1 min then the surface tension measured over time using the ring method with a Kruss Processor Tensiometer K100.

Figure 5. Dye partitioning of 1,6-diphenyl 1,3,5- hexatriene in solutions with increasing concentrations of polysorbate. Method was adopted from that of Chattopadhyay and London.56 Open squares represent polysorbate 80 in water, filled circles represent poly- sorbate 80 in a salt solution, open triangles represent polysorbate 20 in water and filled diamonds represent polysorbate 20 in a salt solution.
results in hydroperoxide formation, side-chain cleavage and eventually formation of short chain acids such as formic acid.73 Hydroperoxide for- mation by polysorbate is known to cause oxidation of proteins as was shown for the two therapeutic proteins recombinant human ciliary neurotrophic factor (rhCNTF) and recombinant human nerve growth factor (rhNGF),81 The oxidation was pre- vented by addition of antioxidants such as cys- teine, thioglycerol, and methionine.81 Hydrolysis of the fatty acid ester bond results in formation of the long chain fatty acids linoeic for polysorbate 20 and oleic for polysorbate 80.34,80,82 The reactions are dependent on the solution pH, presence of oxygen, peroxides, heat, UV light and metal ions
such as copper. Storage of the polysorbates at room temperature (258C) primarily results in hydrolysis of the fatty acid ester while storage at
higher temperatures also favors autooxidation of the ethylene oxide subunits. Storing the poly- sorbate solutions away from light, in the presence of nitrogen and at lower temperatures helps prevent degradation from occurring. Degradation

Figure 6. Reaction scheme depicting the hydrolysis of the fatty acid ester of poly- sorbate 20 and the autooxidation process of the ethylene oxide subunits. Reaction schemes were adapted from those described by Donbrow.34,73,83

can also be prevented by addition of low concen- trations of the antioxidant butylated hydroxyto- luene (BHT).83
The kinetics of hydrolysis for the fatty acid ester bond is affected by the pH and temperature of the solution. Hydrolysis of the ester bond is both acid and base catalyzed80 and the mechanism of ester hydrolysis has been described previously for both acid catalyzed84 and base catalyzed85 reac- tions. Based on the similarity in structure of both polysorbates 20 and 80 around the ester bond the kinetics of the reaction were expected to be similar for the compounds. Bates et al.80 found nearly identical rates of hydrolysis for poly- sorbates 40, 60, and 80. Donbrow et al.34 used the activation energies and rates determined by Bates for polysorbate 80 to predict the hydrolysis rates for a solution of polysorbate 20 at 25 and
408C and found good agreement between the
theoretical and experimental results in a pH 3.95 solution. Studies of polysorbate 80 at 808C showed that the reaction was catalyzed markedly below
pH 3 and above pH 7.6. The lowest rates were obtained between pH 3–7.6 being essentially constant within that range. In addition to the pH dependence the rates of hydrolysis were also concentration dependent. When tested between values of 0.02 to 0.1% (w/v) at pH values of 1.1 and
10.3 and 808C the initial rates of cleavage
decreased with increasing polysorbate concentra- tion. It was suggested that this was due to an increase in the fraction of monomers existing in micelles and or a marked alteration in the micellar structure making it more difficult for the protons or hydroxide anions to reach the ester bonds. Interestingly, different polysorbates with a five- fold change in their CMC values demonstrated similar rates of hydrolysis when examined under acidic conditions.
Autooxidation of the polysorbates occurs along the ethylene oxide moieties of both polysorbate
20 and polysorbate 80. Since the polyethylene oxide structures of the polysorbates are similar then the oxidative degradation processes are expected to be the same. The overall oxidation reactions (Fig. 6) which occur in distinct phases were initially described by Donbrow et al.34,73,83 Phase I is initiation of radical formation by light or a catalyst forming a carbon based radical on either the a or b carbon. Phase II is reaction of the carbon radical with oxygen forming an organic peroxide followed by abstraction of a hydrogen to form an acid and a new carbon radical allowing propaga- tion of the reaction. Finally the reaction is
terminated by self-quenching of the radicals through bimolecular interactions.34,73 The auto- oxidation process involves scission of both C–O and C–C bonds producing short chain acids such as formic acid and acetic acid, suggesting that the peroxide formation mainly occurs at the ethylene oxide moieties near the ends of the chains result- ing in shortening of the polyoxyethylene chains. Low concentrations of formaldehyde have also been identified as contaminants in polysorbate solutions86,87 consistent with formaldehyde iden- tified by Donbrow et al.73 resulting from the C–C fission of the hydroperoxides or free radicals. The chain shortening will change the hydrophilic to lipophilic ratio of the surfactants resulting in modification of their physical characteristics such as CMC value and cloud point formation.34,51 Jaeger et al.75 used catalase to demonstrate that 75% of the peroxide formed in polysorbate 80 is in the form of hydrogen peroxide with the remainder being other types of peroxides. The peroxide formation was prevented by storing the solutions under an atmosphere of nitrogen.21
~
Factors affecting the autooxidation process in- clude temperature, light and the presence of transition state metals such as copper. Recently, Harmon et al.88 have taken advantage of this property of polysorbate 80 together with iron III to generate peroxy radicals and developed an oxidant stressing system for studying degradation of small molecules. As stated above the autooxidation process results in formation of hydroperoxides and acids that can be assayed for using various analytical techniques.21,34,75,76,78 An interesting feature of peroxide formation in polysorbate solutions is that it undergoes an initial lag period followed by a rapid rise in peroxide formation and eventually a decline in the peroxide concentra- tion.21,34 If peroxide content was the only measure used for estimating polysorbate quality then the degree of degradation in a sample could be greatly underestimated at later stages of the reaction. As the peroxides increase and eventually decline, the acid content of the polysorbate solutions increase leading to a decrease in the solution pH. Depending on the storage temperature the solution may fall to a pH value of 2–4 for a 3% (w/v) polysorbate solution, while nondegraded polysorbate solutions have a pH of 6–7.34 Lower temperatures of 25–
408C favor hydrolysis of the ester bond leading to
~
~
long chain fatty acids with a pKa of 4.9 with higher temperatures also leading to autooxidation and short chain acids with pKa values of 3.6. Light also increases the rate of polysorbate

degradation in solution. Ha et al.21 demonstrated that after 5 weeks at 408C solutions of polysorbate 80 contained eightfold as much peroxide as solut-
ions stored in the dark. Interestingly, removal of the air prevented peroxide formation even in the presence of light suggesting that the oxygen in the air was the likely culprit involved with the autooxidation process. The presence of transition state metals such as copper may also catalyze autooxidation of the polysorbate. Donbrow et al.34 showed that addition of copper sulfate to solutions significantly increased the rates of polysorbate degradation.
Oxidation of unsaturated and polyunsaturated fatty acids is also a well-known process but has not been discussed in the literature in relation to polysorbate oxidation. As described in the preceding paragraphs oxidation of polysorbate is mainly associated with oxidation of the ethylene oxide moiety and not the unsaturated fatty acid moieties such as oleate and linoleate present in solutions of polysorbate 20 and 80 to varying degrees (Tab. 1). The oxidation of unsaturated fatty acids occurs in a manner similar to that for polyethylene oxide (PEG). The reaction mechanism that will be described here was derived from studies of unsaturated fatty acids in organic solvents such as benzene but will provide a basis for understanding of what may occur in the polysorbates as well. Similar to PEG autooxidation of the unsaturated fatty acids is a free radical chain process consisting of initiation, propagation and termination.89–91 The key event in initiation is formation of a lipid radical, R.. This can occur by thermal or photochemical hemolytic cleavage of an RH bond or by hydrogen atom abstraction from R–H by an initiator free radical. The propagation step normally begins with addition of molecular dioxygen to R. and the second step of propagation, the rate limiting step, is abstraction of a hydrogen atom from RH by a peroxyl radical ROO. to generate ROOH and another R. eventually result- ing in reactions leading to non reactive species.89–91 Because these reactions are propagated by reaction with molecular oxygen the removal of dioxygen would be expected to eliminate this process as well. As stated though this process has not been described in the literature as related to polysor- bates and as such will be a useful area of research.

CONCLUDING REMARKS

Polysorbates are present in a large number of biopharmaceutical drugs listed in the 2006
Physicians Desk Reference. The concentrations used in the formulations range from 0.0003% (w/v) to 0.3% (w/v). Over the past decade the quality of polysorbate solutions has increased dramatically such that many manufacturers now offer highly purified, low peroxide, and low acid content solutions. While the solutions are initially low in reactive oxygen species care must still be used when storing drug products containing the poly- sorbates since, as described above, degradation can occur under a variety of conditions leading to formation of acids and peroxides that can readily degrade proteins. A good idea is to use the lowest amount of polysorbate possible in a formulation to minimize the risk of damage to a protein product. Additionally, it should also be kept in mind that polysorbate solutions are generally not a homo- genous mixture of a single fatty acid ester but are usually a mixture of the fatty acids mentioned in Table 1. Pure oleic acid esters for polysorbate
80 can be obtained, albeit how this versus a mixture might affect protein stability has not been reported. Finally, data are now emerging that suggest polysorbates not only can bind to proteins (discussed extensively in Randolph and Jones20) but may also affect the structure of the protein in solution, an area of investigation that could provide a fruitful area for future research.

ACKNOWLEDGMENTS

The author would like to thank Dr. Sungae Park for the information from the European Pharma- copoeia, Arnold McAuley, and Hyo Jin Le for the surface tension measurements and Dr. Darren Reid, Dr. David Brems, and Dr. Richard Remmele for many useful discussions and critical reading of the manuscript.

REFERENCES

⦁ Maa YF, Hsu CC. 1998. Investigation on fouling mechanisms for recombinant human growth hor- mone sterile filtration. J Pharm Sci 87:808–812.
⦁ McLeod AG, Walker IR, Zheng S, Hayward CP. 2000. Loss of factor VIII activity during storage in PVC containers due to adsorption. Haemophilia 62:89–92.
⦁ Maa Y-F, Hsu CC. 1997. Protein denaturation by combined effect of shear and air-liquid interface. Biotechnol Bioeng 54:503–512.
⦁ Thurow H, Geisen K. 1984. Stabilisation of dis- solved proteins against denaturation at hydropho- bic interfaces. Diabetologia 27:212–218.

⦁ Eckhardt BM, Oeswein JQ, Bewley TA. 1991. Effect of freezing on aggregation of human growth hor- mone. Pharm Res 8:1360–1364.
⦁ Izutsu K-I, Yoshioka S, Terao T. 1994. Stabilizing effect of amphiphilic excipients on the freeze-thaw- ing and freeze-drying of lactate dehydrogenase. Biotechnol Bioeng 43:1102–1107.
⦁ Nema S, Avis KE. 1993. Freeze-thaw studies of a model protein, lactate dehydrogenase, in the pre- sence of cryoprotectants. J Parent Sci Technol 47: 76–83.
⦁ Carpenter JF, Chang BS, Garzon-Rodriguez W, Randolph TW. 2002. Rational design of stable lyo- philized protein formulations: Theory and practice. Pharm Biotechnol 13:109–133.
⦁ Carpenter JF, Pikal MJ, Chang BS, Randolph TW. 1997. Rational design of stable lyophilized protein formulations: Some practical advice. Pharm Res 14:969–975.
⦁ Carpenter JF, Prestrelski SJ, Dong A. 1998. Appli- cation of infrared spectroscopy to development of stable lyophilized protein formulations. Eur J Pharm Biopharm 45:231–238.
⦁ Ip AY, Arakawa T, Silvers H, Ransone CM, Niven RW. 1995. Stability of recombinant consensus interferon to air-jet and ultrasonic nebulization. J Pharm Sci 84:1210–1214.
⦁ Adler M, Lee G. 1999. Stability and surface activity of lactate dehydrogenase in spray-dried trehalose. J Pharm Sci 88:199–208.
⦁ Broadhead J, Rouan SK, Hau I, Rhodes CT. 1994. The effect of process and formulation variables on the properties of spray-dried beta-galactosidase. J Pharm Pharmacol 46.
⦁ Maa YF, Nguyen PA, Andya JD, Dasovich N, Swee- ney TD, Shire SJ, Hsu CC. 1998. Effect of spray drying and subsequent processing conditions on residual moisture content and physical/biochemical stability of protein inhalation powders. Pharm Res 15:768–775.
⦁ Millqvist-Fureby A, Malmsten M, Bergenstahl B. 1999. Spray-drying of trypsin— Surface character- isation and activity preservation. Int J Pharm 188: 243–253.
⦁ Mumenthaler M, Hsu CC, Pearlman R. 1994. Fea- sibility study on spray-drying protein pharmaceu- ticals: Recombinant human growth hormone and tissue-type plasminogen activator. Pharm Res 11:12–20.
⦁ Tzannis ST, Prestrelski SJ. 1999. Activity-stability considerations of trypsinogen during spray dry- ing: Effects of sucrose. J Pharm Sci 88:351–359.
⦁ Tzannis ST, Hrushesky WJ, Wood PA, Przybycien TM. 1996. Irreversible inactivation of interleukin 2 in a pump-based delivery environment. Proc Natl Acad Sci USA 93:5460–5465.
⦁ Katakam M, Bell LN, Banga AK. 1995. Effect of surfactants on the physical stability of recombinant
human growth hormone. J Pharm Sci 84:713– 716.
⦁ Randolph TW, Jones LS. 2002. Surfactant-protein interactions. Pharm Biotechnol 13:159–175.
⦁ Ha E, Wang W, Wang YJ. 2002. Peroxide formation in polysorbate 80 and protein stability. J Pharm Sci 91:2252–2264.
⦁ Nema S, Washkuhn RJ, Brendel RJ. 1997. Excipi- ents and their use in injectable products. J Pharm Sci Technol 51:166–171.
⦁ Ayorinde FO, Gelain SV, Johnson JH, Jr., Wan LW. 2000. Analysis of some commercial poly- sorbate formulations using matrix-assisted laser desorption/ionization time-of-flight mass spectro- metry. Rapid Commun Mass Spectrom 14:2116– 2124.
⦁ Brandner JD. 1998. The composition of NF-defined emulsifiers: Sorbitan monolaurate, monopalmitate, monostearate, monooleate, polysorbate 20, poly- sorbate 40, polysorbate 60, and polysorbate 80. Drug Dev Ind Pharm 24:1049–1054.
⦁ Frison-Norrie S, Sporns P. 2001. Investigating the molecular heterogeneity of polysorbate emulsifiers by MALDI-TOF MS. J Agric Food Chem 49:3335– 3340.
⦁ European Pharmacopoeia. 2007. On-line 5th edition: European Directorate for the Quality of Medicines and Healthcare.
⦁ Porter M. 1994. Handbook of Surfactants. 2nd ed. Glasgow: Blackie Academic & Professional.
⦁ Schramm LL, Marangoni G. 2000. Surfactants and their solutions: Basic principles. In: Schramm LL, editor. Surfactants: Fundamentals and applica- tions in the petroleum industry. 1st ed. Cambridge: University Press. pp 1–49.
⦁ Privalov PL, Gill SJ. 1989. The hydrophobic effect: A reappraisal. Pure Appl Chem 61:1097–1104.
⦁ Tanford C. 1973. The hydrophobic effect: Formation of micelles and biological membranes. 2nd ed. New York: John Wiley & Sons.
⦁ Silverstein KAT, Haymet ADJ, Dill KA. 1998. A simple model of water and the hydrophobic effect. J Am Chem Soc 120:3166–3175.
⦁ Southall NT, Dill KA, Haymet ADJ. 2002. A view of the hydrophobic effect. J Phys Chem B 106:521– 533.
⦁ Ananthapadmanabhan KP. 1993. Surfactant solu- tions: Adsorption and aggregation properties. In: Goddard ED, Ananthapadmanabhan KP, edi- tors. Interactions of surfactants with polymers and proteins. 1st ed. Boca Raton: CRC Press. pp 40–51.
⦁ Donbrow M, Azaz E, Pillersdorf A. 1978. Autoxida- tion of polysorbates. J Pharm Sci 67:1676–1681.
⦁ Fainerman VB, Miller R, Aksenenko EV, Makievski AV, Kragel J, Loglio G, Liggieri L. 2000. Effect of surfactant interfacial orientation/aggregation on adsorption dynamics. Adv Colloid Interface Sci 86:83–101.

⦁ Paria S, Khilar KC, Paria S, Khilar KC. 2004. A review on experimental studies of surfactant adsorption at the hydrophilic solid-water interface. Adv Colloid Interface Sci 110:75–95.
⦁ Mahler HC, Muller R, Friess W, Delille A, Matheus
S. 2005. Induction and analysis of aggregates in a liquid IgG1-antibody formulation. Eur J Pharm Biopharm 59:407–417.
⦁ Kerwin BA, Heller MC, Levin SH, Randolph TW. 1998. Effects of Tween 80 and sucrose on acute short-term stability and long-term storage at —20 degrees C of a recombinant hemoglobin. J Pharm Sci 87:1062–1068.
⦁ Chou DK, Krishnamurthy R, Randolph TW, Car- penter JF, Manning MC. 2005. Effects of Tween 20 and Tween 80 on the stability of Albutropin during agitation. J Pharm Sci 94:1368–1381.
⦁ 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.
⦁ Kreilgaard L, Jones LS, Randolph TW, Frokjaer S, Flink JM, Manning MC, Carpenter JF. 1998. Effect of Tween 20 on freeze-thawing- and agitation- induced aggregation of recombinant human factor
XIII. J Pharm Sci 87:1597–1603.
⦁ Patro SY, Freund E, Chang BS. 2002. Protein for- mulation and fill-finish operations. Biotechnol Annu Rev 8:55–84.
⦁ Levine HL, Ransohoff TC, Kawahata RT, Mc Gregor WC. 1991. The use of surface tension mea- surements in the design of antibody-based product formulations. J Parent Sci Technol 45:160– 165.
⦁ Feng M, Morales AB, Poot A, Beugeling T, Bantjes
⦁ 1995. Effects of Tween 20 on the desorption of proteins from polymer surfaces. J Biomater Sci Polym Ed 7:415–424.
⦁ Norman PS, Marsh DG. 1978. Human serum albumin and Tween 80 as stabilizers of allergen solutions. J Allergy Clin Immun 62:314–319.
⦁ Nino RR, Patino JMR. 1998. Surface tension of bovine serum albumin and Tween 20 at the air- aqueous interface. J Am Oil Chem Soc 75:1241– 1248.
⦁ Gunning AP, Mackie AR, Kirby AR, Morris VJ. 2001. Scanning near field optical microscopy of phase separated regions in a mixed interfacial pro- tein (BSA)-surfactant (Tween 20) film. Langmuir 17:2013–2018.
⦁ Gunning PA, Mackie AR, Gunning AP, Wilde PJ, Woodward NC, Morris VJ. 2004. The effect of sur- factant type on protein displacement from the air- water interface. Food Hydrocolloids 18:509–515.
⦁ Gunning PA, Mackie AR, Gunning AP, Woodward NC, Wilde PJ, Morris VJ. 2004. Effect of surfactant type on surfactant-protein interactions at the air-water interface. Biomacromolecules 5:984–991.
⦁ Mackie AR, Gunning AP, Ridout MJ, Wilde PJ, Morris VJ. 2001. Orogenic displacement in mixed beta-lactoglobulin/beta-casein films at the air/ water interface. Langmuir 17:6593–6598.
⦁ Donbrow M, Hamburger R, Azaz E. 1975. Surface tension and cloud point changes of polyoxyethylenic non-ionic surfactants during autoxidation. J Pharm Pharmacol 27:160–166.
⦁ Mittal KL. 1972. Determination of CMC of poly- sorbate 20 in aqueous solution by surface tension method. J Pharm Sci 61:1334–1335.
⦁ Rusanov AI. 1996. Determination of micellar char- acteristics by measuring quantities related to the surfactant chemical potential. Mendeleev Commun 6:217–219.
⦁ Ohyashiki T, Mohri T. 1983. Fluorometric analysis of the micelle formation process of surfactants in aqueous solution 1. Utility of pyrene in determina- tion of the critical micelle concentration. Chem Pharm Bull 31:1296–1300.
⦁ Brito RMM, Vaz WLC. 1986. Determination of the critical micelle concentration of surfactants using the fluorescent probe N phenyl-1-naphthylamine. Anal Biochem 152:250–255.
⦁ Chattopadhyay A, London E. 1984. Fluorimetric determination of critical micelle concentration avoiding interference from detergent charge. Anal Biochem 139:408–412.
⦁ Maiti NC, Krishna MMG, Britto PJ, Periasamy N. 1997. Fluorescence dynamics of dye probes in micelles. J Phys Chem B 101:11051–11060.
⦁ Tummino PJ, Gafni A. 1993. Determination of the aggregation number of detergent micelles using steady-state fluorescence quenching. Biophys J 64: 1580–1587.
⦁ Zhang X, Jackson JK, Burt HM. 1996. Determina- tion of surfactant critical micelle concentration by a novel fluorescence depolarization technique. J Biochem Biophys Methods 31:145–150.
⦁ Fuguet E, Rafols C, Roses M, Bosch E. 2005. Critical micelle concentration of surfactants in aqueous buf- fered and unbuffered systems. Anal Chim Acta 548:95–100.
⦁ Heerklotz H, Seelig J, Heerklotz H, Seelig J. 2000. Titration calorimetry of surfactant-membrane par- titioning and membrane solubilization. Biochim Biophys Acta 1508:69–85.
⦁ Lasch J, Hildebrand A, Lasch J, Hildebrand A. 2002. Isothermic titration calorimetry to study CMCs of neutral surfactants and of the liposome- forming bolaamphiphile dequalinium. J Lipos Res 12:51–56.
⦁ Li X, Wettig SD, Verrall RE, Li X, Wettig SD, Verrall RE. 2005. Isothermal titration calorimetry and dynamic light scattering studies of interactions between gemini surfactants of different structure and Pluronic block copolymers. J Colloid Interface Sci 282:466–477.

⦁ Reis S, Moutinho CG, Matos C, de Castro B, Gameiro P, Lima JLFC. 2004. Noninvasive meth- ods to determine the critical micelle concentration of some bile acid salts. Anal Biochem 334:117–126.
⦁ Saito Y, Kondo Y, Abe M, Sato T. 1994. Effects of temperature on micellar formation and micellar structure of pluronic L-64. Yakugaku Zasshi 114: 325–332.
⦁ Villalobos AP, Gunturi SR, Heavner GA. 2005. Interaction of polysorbate 80 with erythropoietin: A Case Study in Protein-Surfactant Interactions. Pharm Res 22:1186–1194.
⦁ Bam NB, Randolph TW, Cleland JL. 1995. Stability of protein formulations: Investigation of surfactant effects by a novel EPR spectroscopic technique. Pharm Res 12:2–11.
⦁ Zhang R, Somasundaran P. 2004. Abnormal micel- lar growth in sugar-based and ethoxylated nonionic surfactants and their mixtures in dilute regimes using analytical ultracentrifugation. Langmuir 20: 8552–8558.
⦁ Ananthapadmanabhan KP. 1993. Surfactant solu- tions: Adsorption and aggregation properties. In: Goddard ED, Ananthapadmanabhan KP, editors. Interactions of surfactants with polymers and proteins. 1st ed. Boca Raton: CRC Press. pp 11–15.
⦁ Patist A, Bhagwat SS, Penfield KW, Aikens P, Shah DO. 2000. On the measurement of critical micelle concentrations of pure and technical-grade nonio- nic surfactants. J Surfactants Deterg 3:53–58.
⦁ Helenius A, McCaslin DR, Fries E, Tanford C. 1979. Properties of detergents. Methods Enzymol 56:734– 749.
⦁ Acharya KR, Bhattacharya SC, Moulik SP. 1997. Salt effects on surfactant aggregation and dye- micelle complexation. Indian J Chem Sect A 36: 137–143.
⦁ Donbrow M, Hamburger R, Azaz E, Pillersdorf
⦁ 1978. Development of acidity in non-ionic surfac- tants: Formic and acetic acid. Analyst 103:400–402.
⦁ Hamburger R, Azaz E, Donbrow M. 1975. Auto- xidation of polyoxyethylenic non-ionic surfactants and of polyethylene glycols. Pharm Acta Helv 50: 10–117.
⦁ Jaeger J, Sorensen K, Wolff SP. 1994. Peroxide accumulation in detergents. J Biochem Biophys Methods 29:77–81.
⦁ Khossravi M, Kao YH, Mrsny RJ, Sweeney TD. 2002. Analysis methods of polysorbate 20: A new method to assess the stability of polysorbate
20 and established methods that may overlook degraded polysorbate 20. Pharm Res 19:634–639.
⦁ Lever M. 1977. Peroxides in detergents as interfer- ing factors in biochemical analysis. Anal Biochem 83:274–284.
⦁ Magill A, Becker AR. 1984. Spectrophotometric method for quantitation of peroxides in sorbitan
monooleate and monostearate. J Pharm Sci 73: 1663–1664.
⦁ Segal R, Azaz E, Donbrow M. 1979. Peroxide removal from non-ionic surfactants. J Pharm Phar- macol 31:39–40.
⦁ Bates TR, Nightingale CH, Dixon E. 1973. Kinetics of hydrolysis of polyoxyethylene (20) sorbitan fatty acid ester surfactants. J Pharm Pharmacol 25:470– 477.
⦁ Knepp VM, Whatley JL, Muchnik A, Calderwood TS, Knepp VM, Whatley JL, Muchnik A, Calder- wood TS. 1996. Identification of antioxidants for prevention of peroxide-mediated oxidation of recombinant human ciliary neurotrophic factor and recombinant human nerve growth factor. J Pharm Sci Technol 50:163–171.
⦁ Hu M, Niculescu M, Zhang XM, Hui A. 2003. High- performance liquid chromatographic determination of polysorbate 80 in pharmaceutical suspensions. J Chromatogr A 984:233–236.
⦁ Donbrow M. 1987. Stability of the polyoxyethylene chain. In: Schick MJ, editor. Nonionic surfactants: Physical chemistry. 1st ed. New York: Marcel Dekker Inc. pp 1060–1065.
⦁ Roberts I, Urey HC. 1939. The mechanism of acid catalyzed ester hydrolysis, esterification and oxy- gen exchange of carboxylic acids. J Am Chem Soc 61:2584–2587.
⦁ Stefanidis D, Jencks WP. 1993. General base cata- lysis of ester hydrolysis. J Am Chem Soc 115: 6045– 6050.
⦁ Chafetz L, Hong WH, Tsilifonis DC, Taylor AK, Philip J, Chafetz L, Hong WH, Tsilifonis DC, Taylor AK, Philip J. 1984. Decrease in the rate of capsule dissolution due to formaldehyde from polysorbate 80 autoxidation. J Pharm Sci 73:1186–1187.
⦁ Nassar MN, Nesarikar VN, Lozano R, Parker WL, Huang Y, Palaniswamy V, Xu W, Khaselev N, Nassar MN, Nesarikar VN, Lozano R, Parker WL, Huang Y, Palaniswamy V, Xu W, Khaselev
N. 2004. Influence of formaldehyde impurity in polysorbate 80 and PEG-300 on the stability of a parenteral formulation of BMS-204352: Identifica- tion and control of the degradation product. Pharm Dev Technol 9:189–195.
⦁ Harmon PA, Kosuda K, Nelson E, Mowery M, Reed RA. 2006. A novel peroxy radical based oxidative stressing system for ranking the oxidizability of drug substances. J Pharm Sci 95:2014–2028.
⦁ Porter NA, Calwell SE, Mills KA. 1995. Mecha- nisms of free radical oxidation of unsaturated lipids. Lipids 30:277–290.
⦁ Porter NA, Wagner CR. 1986. Phospholipid oxida- tion. Adv Free Rad Bio 2:283–323.
⦁ Yin H, Porter NA. 2005. New insights regarding the autooxidation of polyunsaturated fatty acids. Anti- oxid Redox Sign 7:170–184.