Article pubs.acs.org/ac
Characterization and Stability Study of Polysorbate 20 in Therapeutic Monoclonal Antibody Formulation by Multidimensional UltrahighPerformance Liquid Chromatography−Charged Aerosol Detection− Mass Spectrometry Yi Li,† Daniel Hewitt,‡ Yvonne K. Lentz,§ Junyan A. Ji,§ Taylor Y. Zhang,‡ and Kelly Zhang*,† †
Small Molecule Pharmaceutical Sciences, ‡Protein Analytical Chemistry, and §Late Stage Pharmaceutical and Process Development, Genentech, 1 DNA Way, South San Francisco, California 94080, United States ABSTRACT: Polysorbate 20 is a nonionic surfactant commonly used in the formulation of therapeutic monoclonal antibodies (mAb) to prevent protein denaturation and aggregation. It is critical to understand the molecular heterogeneity and stability of polysorbate 20 in mAb formulations as polysorbate can gradually degrade in aqueous solution over time by multiple pathways losing surfactant functions and leading to protein aggregation. The molecular heterogeneity of polysorbate and the interference from proteins and the excipient in the formulation matrix make it a challenge to study polysorbate in protein formulations. In this work, the characterization and stability study of polysorbate 20 in the presence of mAb formulation sample matrix is first reported using two-dimensional liquid chromatography (2DLC) coupled with charged aerosol detection (CAD) and mass spectrometry (MS) detection. A mixed-mode column that has both anion-exchange and reversed-phase properties was used in the first dimension to separate protein and polysorbate in the formulation sample, while polysorbate 20 esters were trapped online and then analyzed using an reversed-phase ultrahigh-performance liquid chromatography (RP-UHPLC) column in the second dimension to further separate the ester species. The MS served as the third dimension to further resolve as well as to identify the polysorbate ester subspecies. Another 2DLC method using a cation-exchange column in the first dimension and the same RPUHPLC method in the second dimension was developed to analyze the degradation products of polysorbate 20. Stability samples of a protein drug product were studied using these two 2DLC−CAD−MS methods to separate, identify, and quantify the multiple ester species in polysorbate 20 and also to monitor the change of their corresponding degradants. We found different polysorbate esters degrade at different rates, and importantly, the degradation rates for some esters are different in the protein formulation compared to a placebo that has no protein. The multidimensional UHPLC−CAD−MS approach provides insights into the heterogeneous stability behaviors of polysorbate 20 subspecies in real-time stability samples of a mAb formulation.
P
40−60% laurate (C12), 14−25% myristate (C14), 7−15% palmitate (C16), and the remaining 20% range from C8 to C18 esters.7 In addition to the esters, polar molecules such as POE sorbitan, POE isosorbide, and POE are also present in polysorbate 20 due to incomplete esterification during the synthesis and as the degradation products of the esters.8 The polysorbate structural heterogeneity, complicated by the presence of proteins at high concentration, makes the characterization of polysorbate highly challenging in protein formulations. There have been several reports about the characterization of polysorbate by NMR, matrix-assisted laser desorption ioniza-
olysorbate 20 (often referred by its trade name Tween 20) is a nonionic surfactant commonly used in the formulation of therapeutic monoclonal antibodies (mAb) due to its biocompatibility, low toxicity, and good stabilizing properties for proteins.1,2 The role of polysorbate 20 in protein formulations is to prevent the formation of aggregates and protect proteins from denaturation.3−5 The amphiphilic nature of polysorbate 20 is provided by two parts: the poly(oxyethylene) (POE) and dehydrated sugar core which comprises the hydrophilic headgroup, and the fatty acids as the hydrophobic tail group. Polysorbate 20 is a complex mixture of esters of different polymeric polar head groups and various fatty acid tails with multiple degrees of esterification. The major esters in polysorbate are POE sorbitan monoesters, POE isosorbide monoesters, POE esters, and diesters of POE sorbitan as a result of polysorbate synthesis shown in Figure 1.6 The fatty acid tails on the esters of polysorbate 20 consist of © 2014 American Chemical Society
Received: March 16, 2014 Accepted: April 21, 2014 Published: April 21, 2014 5150
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
Figure 1. Synthetic scheme and related structures of polysorbate 20.
profiling the composition of polysorbate in a protein sample matrix. Polysorbate 20 esters are susceptible to degradation in water and lose their surfactant functions as the result of ester hydrolysis.19−21 The degradation of polysorbate is dependent on the solution pH and temperature. Hewitt et al. studied the composition and base hydrolysis of esters with different fatty acid chain length in polysorbate 20 and 80 by RP-LC with tandem MS.22 Since the varying stability of the subspecies within polysorbate could affect the stability of protein drug product (i.e., particulate formation, change in cloud point), it is very important to understand the stability of the different components of polysorbate 20 in protein formulation. However, no literature was found to report on the characterization and stability of individual polysorbate subspecies in the presence of protein sample matrix. This is because most protein drug products contain high concentrations of therapeutic mAb, usually 10−100 mg/mL, and multiple excipients to enhance stability, control pH, and adjust tonicity.23 The high concentrations of antibodies and excipients present significant interference for HPLC analysis of polysorbate, making the characterization of polysorbate by conventional one-dimensional HPLC highly challenging if at all possible. Multidimensional liquid chromatography is a powerful tool to overcome the challenges of analyzing complex samples and enhance resolving power.24−26 By using orthogonal separation mechanisms in each dimension, multidimensional chromatography is multiplicative of peak capacity of each dimension ideally, and thus achieves separation efficiency that cannot be
tion (MALDI) mass spectrometry, and more recently liquid chromatography−mass spectrometry (LC−MS).9−13 However, all these characterizations are based on polysorbate standard solutions, and not the real drug product formulation in the presence of protein sample matrix. For instance, Borisov et al. used LC−MS with a computer-aided peak assignment algorithm to profile the constituents in polysorbates in great detail.14 Erdem et al. used LC and ion mobility coupled with MS to separate isobaric species and species with superimposed isotope patterns so that the esters could be conclusively identified.6 As polysorbates are widely used in protein formulations as a stabilizing agent, studies have been performed to determine the total polysorbate content in protein formulations. Tani et al. described a method for quantitation of polysorbate 80 by size exclusion chromatography.15 Reversed-phase HPLC (RP-LC) with UV detector has also been used for the quantitation of polysorbate 80 in protein solutions by base hydrolysis of all the esters followed by quantifying the oleic acid, a hydrolysis product of polysorbate 80 that has UV absorption at 195 nm.16 Hewitt et al. developed a mixed-mode LC method coupled with an evaporative light scattering detector (ELSD) detector for the measurement of polysorbate 20 in protein solutions without additional sample treatment.17 Recently, He et al. reported a size exclusion chromatography coupled with mixed-mode chromatography for comprehensive profiling of biopharmaceutical drug products composing proteins and various kinds of excipients, including total polysorbate 80.18 However, all these methods were developed to quantify the total polysorbate content, not for 5151
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
with an electrospray ionization (ESI) source. The instrument was operated in a positive mode with capillary voltage at 3.5 kV, capillary temperature at 350 °C, sheath flow rate at 45, and aux flow rate at 10. Full scan spectra were collected over the m/z range of 200−2000. Thermo Xcalibur software was used to acquire MS data. Mixed-Mode Chromatography. Analytes were separated using an Oasis MAX column (20 mm × 2.1 mm, 30 μm, Waters, Milford, MA, U.S.A.). Initial conditions were set at 80% solvent A (0.2% formic acid in water) and 20% solvent B (0.2% formic acid in ACN) and held for 3 min. Solvent B was then increased to 90% at 3.1 min using a step gradient and held for 5 min followed by an equilibration step of 20% B for 5 min. The flow rate was kept at 1.0 mL/min. Column temperature was 25 °C. RP-UHPLC Chromatography. Analytes were separated using an Acquity BEH C18 column (150 mm × 2.1 mm, 1.7 μm, Waters, Milford, MA, U.S.A.). Initial conditions were set at 85% solvent A (0.05% formic acid in water) and 15% solvent B (0.04% formic acid in ACN). Solvent B was increased to 100% in 30 min using linear gradient and held at 100% for 5 min followed by an equilibration step of 15% B for 5 min. The flow rate was kept at 0.5 mL/min. Column temperature was 50 °C. IEC Chromatography. Analytes were separated using a ProPac WCX-10 column (50 mm × 4 mm, Dionex, Sunnyvale, CA, U.S.A.). Initial conditions were set at 100% solvent A (10 mM ammonium acetate, pH 5.0) and held for 3 min. Then 100% (10 mM ammonium acetate, pH 7.0, 250 mM NaCl) was used at 3 min by a step gradient and held for 7 min followed by an equilibration step of 100% A for 5 min. The flow rate was kept at 0.5 mL/min. Column temperature was 25 °C.
obtained by one-dimensional HPLC. The coupling of twodimensional liquid chromatography (2DLC) with MS has been reported as an effective way to reduce signal suppression caused by matrix interference.27,28 Due to the molecular complexity of both polysorbate and its degradation products, not only do we need to remove proteins and other excipients from the sample matrix in the first dimension, but also a high-resolution HPLC method in the second dimension is required to separate the different polysorbate subspecies and their degradation products. This work represents the first attempt for polysorbate molecular heterogeneity characterization and stability study in protein formulation sample matrix using an online 2DLC setup. As polysorbate 20 has no UV absorbing chromophores, the 2DLC was coupled with a charged aerosol detector (CAD), a universal detector for nonvolatile compounds, and an MS detector to characterize the components in polysorbate 20. The MS also served as the third dimension to further resolve and identify the polysorbate subspecies. A high-resolution ultrahighperformance liquid chromatography (UHPLC) method was developed and used in the second dimension to separate different polysorbate species. Online heart-cutting 2DLC mode29 was used to avoid sample loss from the first dimension and allow longer run time on the second dimension to achieve high resolution. Two 2DLC methods were developed, one using a mixed-mode chromatography in the first dimension and RP-UHPLC chromatography in the second dimension to isolate and characterize the ester subspecies in polysorbate 20. Another 2DLC method using ion-exchange chromatography (IEC) in the first dimension and RP-UHPLC chromatography in the second dimension was used to monitor the change of degradation products to obtain the full picture of polysorbate degradation in mAb formulations. Both 2DLC methods used two very different separation mechanisms in each dimension to take advantage of the orthogonality provided by the 2DLC system.
■
RESULTS AND DISCUSSION Characterization of Polysorbate 20 Esters in mAb Formulation by Mixed-Mode RP 2DLC. We studied a therapeutic mAb formulation that contains 30 mg/mL mAb, 0.02% polysorbate 20, and other excipients at pH 6.0. To study the esters in polysorbate 20 in protein formulation samples, we first need to remove the protein from the formulation matrix and then analyze the ester subspecies. 2DLC provides an excellent platform for this purpose as we can utilize the first dimension to separate polysorbate esters from proteins, selectively trap the esters in a sample loop, and further analyze them in the second dimension. A mixed-mode HPLC method (first dimension) coupled with a RP-UHPLC method (second dimension) was developed to study the composition of polysorbate 20 esters in protein formulation. Figure 2a shows the separation of esters in polysorbate 20 from proteins by the mixed-mode method in first dimension. The mixed-mode column has both anion-exchange and reversed-phase properties. A two-step elution was used for this first-dimension method. The first step used a high aqueous mobile phase acidified with formic acid to remove proteins as they are positively charged at low pH and are washed out by electrostatic repulsion with the stationary phase. The neutral hydrophobic esters, on the other hand, are retained on the column through hydrophobic interactions. The second step used high organic mobile phase to elute all the esters in one peak. The esters that eluted at 4.1 min were trapped in a 500 μL stainless steel loop and then sent to second dimension for further separation of the esters subspecies. Previous studies have used Zorbax SB C8 and Vydac 214MS C4 columns with 5 μm particles to separate the components in
■
EXPERIMENTAL SECTION Reagents and Materials. Acetonitrile (ACN, HPLC grade) was purchased from J. T. Baker (Phillipsburg, NJ, U.S.A.). Polysorbate 20, ammonium formate, formic acid, and sodium chloride were purchased from Sigma-Aldrich (St. Louis, MO, U.S.A.). Deionized water was from an in-house Milli-Q (Millpore, Billerica, MA, U.S.A.). The therapeutic protein formulation stability samples were provided by Genentech (South San Francisco, CA, U.S.A.). The polysorbate 20 standard solution was prepared at 0.2 mg/mL in water. Multidimensional UHPLC−CAD−MS System. The details of the 2DLC system setup used in this study have been reported in our previous work.29 The 2DLC system was built based on the UltiMate 3000 series from Dionex (Sunnyvale, CA, U.S.A.) that has two quaternary pumps, two thermostatic column compartments, and two diode array detectors (DAD). The interface between the two dimensions was a pair of MXT 715-105 valves (six-position, seven-port) purchased from Rheodyne (Oak Harbor, WA, U.S.A.)29 equipped with stainless steel loops for sample trapping. The 2DLC system is coupled with both CAD and MS detectors. The Corona Ultra CAD detector was purchased from ESA Inc. (Chelmsford, MA, U.S.A.) and operated at a nitrogen pressure of 35 psi. Chromeleon 6.70 Chromatography Management software (Dionex, Sunnyvale, CA, U.S.A.) was used for system control and data processing. LCQ Fleet MS was purchased from Thermo Scientific (Waltham, MA, U.S.A.) 5152
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
Figure 2. (a) Separation of esters in polysorbate 20 from proteins by mixed-mode HPLC in the first dimension. (b) Separation of esters subspecies in polysorbate 20 by RP-UHPLC in the second dimension. (c−e) Representative mass spectra of peak 1 (POE sorbitan monolaurate), peak 2 (POE isosorbide monolaurate), and peak 3 (POE laurate), respectively.
polysorbate 20 by RP-HPLC.13,22 As many esters are only partially separated, computer-assisted data analysis was needed to deconvolute the mass spectra for peak identification. UHPLC is a continuing trend in modern HPLC as it provides better resolution and shorter analysis time by using sub-2 μm particle column with ultrahigh pressures. An RP-UHPLC method was developed for the second-dimension separation using an Acquity BEH C18 column with 1.7 μm particles to separate and identify the ester components in polysorbate 20. This UHPLC method provides superior resolution of the esters than the previously reported methods using 5 μm particle columns. Figure 2b shows the separation of the trapped polysorbate 20 esters by this RP-UHPLC method in the second
dimension. Peak identifications were performed by MS directly without deconvolution. In general, the retention time of the esters on the second-dimension UHPLC column is laurate (C12) < myristate (C14) < palmitate (C16), and for esters of a particular fatty acid, POE sorbitan ester < POE isosorbide ester < POE ester. The diesters are more hydrophobic and they elute later after 27 min. Example mass spectra of the laurate esters of three different head groups, POE sorbitan, POE isosorbide, and POE, are shown in Figure 2c−e. The most abundant ester is the POE sorbitan monolaurate that elutes at 19 min (peak 1 in Figure 2b). It was identified by the [M + Na]+ of 1293.78, 1337.80, and 1381.83 series of ions, correlating to the POE sorbitan 5153
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
Table 1. Identification and Characterization of Major Esters in Polysorbate 20 peak no.
RT (min)
1 2 3 4 5 6 7 8 9 10
19.0 20.2 20.5 21.5 23.0 23.2 23.9 28.3 29.9 31.2
peak ID POE POE POE POE POE POE POE POE POE POE
sorbitan monolaurate (C12) isosorbide monolaurate (C12) monolaurate (C12) sorbitan monomyristate (C14) isosorbide monomyristate (C14) monomyristate (C14) sorbitan monopalmitate (C16) sorbitan dilaurate (C12) sorbitan laurate (C12)/myristate (C14) sorbitan laurate (C12)/palmitate (C16)
obsd [M + Na]+
theor [M + Na]+
1337.80 879.54 707.47 1277.79 907.58 691.48 1305.82 1608.04 1636.06 1664.10
1337.82 879.53 707.46 1277.78 907.56 691.46 1305.81 1608.02 1636.05 1664.08
(n (n (n (n (n (n (n (n (n (n
= = = = = = = = = =
22) 12) 11) 20) 12) 10) 20) 24) 25) 24)
(n (n (n (n (n (n (n (n (n (n
= = = = = = = = = =
22) 12) 11) 20) 12) 10) 20) 24) 24) 24)
area % 28 10 6 8 3 1 2 22 12 7
Figure 3. (a) Chromatogram of the IEC method on the first dimension that separates the polyols from protein and (b) RP-UHPLC method on second dimension that separates the different polyols in polysorbate 20.
headgroup that contains n = 21 (theoretical [M + Na]+ = 1293.80), n = 22 (theoretical [M + Na]+ = 1337.82), and n = 23 (theoretical [M + Na]+ =1381.85) ethylene oxide subunits. The [M + Na]+ masses of the POE sorbitan laurate are 182 Da, the molecular weight of lauric acid minus water, higher than the corresponding POE sorbitans with the same ethylene oxide units, confirming it is laurate ester. These [M + Na]+ masses are 44 Da apart, the molecular weight of one ethylene oxide subunit. The MS coupled to the second-dimension UHPLC provides a third dimension to the 2DLC system to resolve the polymeric esters of various ethylene oxide units by their molecular masses. The POE isosorbide monolaurate and POE laurate in polysorbate 20 were identified in the same way by matching their [M + Na]+ masses of different ethylene oxide unit to their theoretical values, as shown in Figure 2, parts d
and e. The esters of different fatty acid chain lengths, such as myristate (C14) and palmitate (C16), have similar MS pattern as the laurate (C12) ester. The [M + Na]+ masses of myristate shifted 28 Da higher compare to laurate as it contains one more ethylene group, and the [M + Na]+ masses of palmitate shifted 56 Da higher as it has two more ethylene groups. Table 1 summarizes the identification and characterization of the major polysorbate 20 esters in the protein formulation. Peak numbers in Table 1 correlate to those labeled in Figure 2b. The observed [M + Na]+ reported in the table are the most abundant mass with number n ethylene oxide subunits. The area% is based on response of the CAD detector. The performance of this online-coupled mixed-mode RPUHPLC method was evaluated by five replicate injections of 0.2 mg/mL polysorbate 20 standard solution to assess the 5154
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
Figure 4. Stability of polysorbate 20 in protein drug product by 2DLC. (a) Second-dimension chromatogram of the esters in the stability samples studied by mixed-mode RP 2DLC method and (b) second-dimension chromatogram of polyols in the stability samples studied by the IEC−RP 2DLC method.
precision of the 2DLC method. The percent relative standard deviations (%RSDs) of the retention time and peak area of the most abundant species, POE sorbitan monolaurate, are 0.2% and 1.0%, respectively, demonstrating good precision of the 2DLC method. The recovery of POE sorbitan monolaurate is 97%, which was determined by calculating the ratio of the peak area on the second dimension to that obtained from the direct analysis by the same RP-UHPLC method with the same sample, indicating the esters were completely trapped from the first dimension and sent to second dimension with negligible sample loss. Analysis of Degradation Products of Polysorbate 20 in mAb Formulation by IEC−RP 2DLC. POE sorbitan, POE isosorbide, and POE are hydrophilic polyol molecules present in polysorbate 20 without surfactant activity due to the absence of hydrophobic fatty tail group. They are the byproducts in the manufacture of polysorbate 20 and also the degradation products of polysorbate 20. As these polyols are polar molecules, they are poorly retained on the mixed-mode column and elute together with protein in the first-dimension method described in last section. Therefore, in order to fully understand the degradation pathway of polysorbate 20 in a protein formulation, another method that can analyze the polyols in polysorbate 20 in the protein formulation would be highly desirable. An IEC method was developed in the first dimension to separate the polyols from the proteins as polyols are neutral molecules, while proteins can be charged depending on solution pH. The same RP-UHPLC method was used in the second dimension to monitor the change of various polyols in the protein formulation samples. A cation-exchange ProPac WCX-10 column was used in first dimension with step gradients to separate the polyols from protein. An initial step of 100% 10 mM ammonium acetate, pH 5.0 mobile phase was used to elute the negatively charged and polar neutral excipients, such as the polyols in polysorbate 20, and other
excipients used in the protein formulation. These excipients came off from the column around 0.8 min while the proteins were retained on the column as they are positively charged at pH 5.0. The nonretained species that eluted at 0.8 min were trapped in a 200 μL loop and analyzed by the same RPUHPLC method used in last section. Protein was eluted after 3 min using 10 mM ammonium acetate and 250 mM NaCl, pH 7.0 mobile phase as they become mostly neutral at pH 7 and are easily washed off from the cation-exchange column by highconcentration salt eluent. Figure 3a shows the separation of polyols from protein by the IEC method in first dimension, and Figure 3b shows the separation and identification of various polyols using the RPUHPLC method in the second dimension. The large peak before 2 min in the second dimension is excipients used in protein formulation. They are very polar molecules that do not retain on the RP column and therefore do not interfere with the polyols peaks. The polyols elute between 5 and 14 min. This region contains three groups of molecules, POE, POE isosorbide, and POE sorbitan, all very well separated by the UHPLC method in the second dimension. Each group consists of a series of peaks differing by one ethylene oxide as identified by MS. The example MS of POE sorbitan peaks that have n = 23, n = 24, and n = 25 ethylene oxide units are also shown in Figure 3b. The mass difference between the neighboring peaks is 44 Da, which is the mass of one ethylene oxide unit. The method precision was demonstrated by calculating the RSD of five replicate injections of a polysorbate 20 standard solution using the IEC−RP 2DLC method. The RSDs of the retention time of POE sorbitan, POE isosorbide, and POE are 0.23%, 0.22%, and 0.20%, respectively. The %RSDs of the peak area of POE sorbitan, POE isosorbide, and POE are 0.9%, 1.2%, and 1.5%, respectively. The recoveries of POE sorbitan, POE isosorbide, and POE are within 95−105%. 5155
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
Figure 5. Change of major polysorbate 20 esters in protein drug product and placebo at different stability time points.
Stability of Polysorbate 20 in Protein Drug Product Formulation. As polysorbate 20 is susceptible to degradation in water and loses its surfactant activity, we studied the stability of polysorbate 20 in a protein drug product using these two 2DLC methods. The protein drug product samples stored at 5 °C were pulled at 2 years and 3 years for stability testing. These stability samples were analyzed by the mixed-mode RP-UHPLC 2DLC method with CAD detection to study the composition of ester subspecies in polysorbate 20, and then the same samples were run on the IEC−RP-UHPLC 2DLC method to study the change of polyols. Figure 4a shows that almost all the esters species slowly decreased over time. The degradation was further supported by the increase of polyols analyzed by the IEC−RP-UHPLC method as shown in Figure 4b. The POE sorbitan monoesters of different fatty acid chain lengths, such as laurate (C12), myristate (C14), and palmitate (C16), all degrade to the same POE sorbitan product and the corresponding fatty acids, causing the increase of POE sorbitan peaks. Likewise, the increases of POE isosorbide and POE peaks result from the degradation of POE isosorbide esters and POE esters of different fatty acid chain length, as illustrated by the arrows in Figure 4. To further understand the degradation rates of different ester species in polysorbate 20, we studied the peak area change of major esters in the protein drug product compared to the placebo (see Figure 5) at different time points. The placebo sample contains the exact same components as the protein drug product except the therapeutic antibody, and it was stored at the same condition of 5 °C. The results show that polysorbate 20 monoesters in the protein drug product degrade at similar rates as those in the placebo sample. However, the diesters degrade slower in the protein drug product than in the placebo. We do not fully understand why the diesters degrade at different rate with and without the protein at this point, and the mechanism of polysorbate degradation is out of the scope of this paper. However, this finding demonstrates the importance of characterizing the molecular heterogeneity of polysorbate 20 in protein formulations as the information gained from the study of polysorbate standard solution may not be fully applicable to the behavior of polysorbate in protein drug products. Another interesting observation is the increase of POE sorbitan monolaurate while all other major esters
decreased over time. This is likely due to the degradation of POE sorbitan mixed diesters or higher order esters containing laurate into the POE sorbitan monolaurate. The results also suggest that POE isosorbide esters and POE esters degrade faster than POE sorbitan esters. This observation is consistent with the change of the three different types of polyols as shown in Figure 6 where POE isosorbide and POE peaks increase
Figure 6. Change of polyols in polysorbate 20 in protein drug product at different stability time points.
more than POE sorbitan. As both POE isosorbide esters and POE esters are considered byproducts in polysorbate 20 manufacture, it is important to control these byproducts in the synthetic process to ensure product quality from long-term stability perspective.
■
CONCLUSIONS We characterized polysorbate 20 molecular heterogeneity and stability in the presence of a mAb formulation sample matrix using multidimensional UHPLC−CAD−MS. A mixed-mode column was used in the first dimension to separate the polysorbate esters from the protein in the formulation sample, and the esters were further separated by a RP-UHPLC method in the second dimension to quantify the multiple ester subspecies. Another 2DLC method using a cation-exchange column in the first dimension and the same RP-UHPLC in the second dimension was used for the analysis of degradation products of polysorbate 20. The decrease of esters in 5156
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
Analytical Chemistry
Article
(20) Khossravi, M.; Kao, Y. H.; Mrsny, R. J.; Sweeney, T. D. Pharm. Res. 2002, 19, 634−639. (21) Bates, T. R.; Nightingale, C. H.; Dixon, E. J. Pharm. Pharmacol. 1973, 25, 470−477. (22) Hewitt, D.; Alvarez, M.; Robinson, K.; Ji, J.; Wang, Y. J.; Kao, Y.H.; Zhang, T. J. Chromatogr., A 2011, 1218, 2138−2145. (23) Powell, M. F.; Nguyen, T.; Baloian, L. PDA J. Pharm. Sci. Technol. 1998, 52, 238−311. (24) Opiteck, G. J.; Jorgenson, J. W.; Anderegg, R. J. Anal. Chem. 1997, 69, 2283−2291. (25) Stoll, D. R.; Wang, X. L.; Carr, P. W. Anal. Chem. 2008, 80, 268−278. (26) Zhang, K.; Wang, J.; Tsang, M.; Wigman, L.; Chetwyn, N. P. Am. Pharm. Rev. 2013, 16, 39−44. (27) Pascoe, R.; Foley, J. P.; Gusev, A. I. Anal. Chem. 2001, 73, 6014−6023. (28) Matejicek, D. J. Chromatogr., A 2012, 1231, 52−58. (29) Zhang, K.; Li, Y.; Tsang, M.; Chetwyn, N. P. J. Sep. Sci. 2013, 36, 2986−2992.
polysorbate 20 and the increase of polyols confirmed that polysorbate 20 did undergo degradation in both the protein drug product and the placebo samples over 3 years of storage at 5 °C. However, some polysorbate 20 esters exhibit different degradation rates in solutions with and without antibodies. This demonstrates the importance of studying the molecular heterogeneity of polysorbate 20 in protein formulations for drug formulation development. We have also found that the POE isosorbide esters and POE esters in polysorbate 20 degrade faster than POE sorbitan esters. The multidimensional UHPLC−CAD−MS approach allows us to get insight about the profiling of polysorbate 20 in protein formulation and help understand the potential impact on formulation stability.
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +1 650 467 8470. Fax: +1 650 225 6238. E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Jackie Tyler for providing some of the study materials. We thank Dr. Colin Medley, Dr. Michael Dong, Dr. Pete Yehl, and Dr. Larry Wigman of Genentech for manuscript review and helpful suggestions.
■
REFERENCES
(1) Garidel, P.; Hoffmann, C.; Blume, A. Biophys. Chem. 2009, 143, 70−78. (2) Patapoff, T. W.; Esue, O. Pharm. Dev. Technol. 2009, 14, 659− 664. (3) Bam, N. B.; Cleland, J. L.; Randolph, T. W. Biotechnol. Prog. 1996, 12, 801−809. (4) Bam, N. B.; Cleland, J. L.; Yang, J.; Manning, M. C.; Carpenter, J. F.; Kelley, R. F.; Randolph, J. W. J. Pharm. Sci. 1998, 87, 1554−1559. (5) Randolph, T. W.; Jones, L. S. Pharm. Biotechnol. 2002, 13, 159− 175. (6) Erdem, N. S.; Alawani, N.; Wesdemiotis, C. Anal. Chim. Acta 2014, 808, 83−93. (7) European Pharmacopoeia Online, 8th ed.; European Directorate for the Quality of Medicines & HealthCare; Council of Europe: Strasbourg, France, 2014. (8) Zhang, R.; Wang, Y.; Ji, Y.; Shi, B.-j.; Zhang, Z.-p.; Zhang, H.-y.; Yang, M.; Wang, Y.-m. J. Chromatogr., A 2013, 1272, 73−80. (9) Ayorinde, F. O.; Gelain, S. V.; Johnson, J. H., Jr.; Wan, L. W. Rapid Commun. Mass Spectrom. 2000, 14, 2116−2124. (10) Frison-Norrie, S.; Sporns, P. J. Agric. Food Chem. 2001, 49, 3335−3340. (11) Raith, K.; Schmelzer, C. E. H.; Neubert, R. H. H. Int. J. Pharm. 2006, 319, 1−12. (12) Vu Dang, H.; Gray, A. I.; Watson, D.; Bates, C. D.; Scholes, P.; Eccleston, G. M. J. Pharm. Biomed. Anal. 2006, 40, 1155−1165. (13) Hvattum, E.; Yip, W. L.; Grace, D.; Dyrstad, K. J. Pharm. Biomed. Anal. 2012, 62, 7−16. (14) Borisov, O. V.; Ji, J. A.; Wang, Y. J.; Vega, F.; Ling, V. T. Anal. Chem. 2011, 83, 3934−3942. (15) Tani, T. H.; Moore, J. M.; Patapoff, T. W. J. Chromatogr., A 1997, 786, 99−106. (16) Adamo, M.; Dick, L. W., Jr.; Qiu, D.; Lee, A.-H.; Devincentis, J.; Cheng, K.-C. J. Chromatogr., B 2010, 878, 1865−1870. (17) Hewitt, D.; Zhang, T.; Kao, Y.-H. J. Chromatogr., A 2008, 1215, 156−160. (18) He, Y.; Friese, O. V.; Schlittler, M. R.; Wang, Q.; Yang, X.; Bass, L. A.; Jones, M. T. J. Chromatogr., A 2012, 1262, 122−129. (19) Kerwin, B. A. J. Pharm. Sci. 2008, 97, 2924−2935. 5157
dx.doi.org/10.1021/ac5009628 | Anal. Chem. 2014, 86, 5150−5157
