Process for polysorbate quantification in a sample involving LC-MS with an internal standard

Abstract

The present application concerns a process for quantifying polysorbates in a sample by implementing a LC-MS analysis with an internal standard, and the process for monitoring degradation of polysorbates in such sample.

Claims

1. A process for quantifying at least one polysorbate derivative in a sample, said process comprising: The step of performing a LC-MS analysis of said sample based on the signal of the dioxolanylium ion; The step of performing an internal calibration with an internal standard of said polysorbate, wherein said internal standard is an ethoxylated fatty acid of formula (I): ##STR00007## wherein: ##STR00008## represents a fatty acid residue; where R represents a C3-C24 linear or branched saturated alkyl; R′ is H or ##STR00009##  where R″ represents a C3-C24 linear or branched saturated alkyl; and n is comprised between 1 and 100; and the mixtures thereof.

2. The process according to claim 1 wherein said polysorbate is PS20 or PS80.

3. The process according to claim 1 wherein said internal standard is chosen from: ##STR00010## where n is comprised between 1 and 100.

4. The process according to claim 1 wherein said internal standard is ##STR00011## n is comprised between 1 and 100.

5. The process according to claim 1 wherein said LC-MS analysis involves a single quadrupole mass detector (QDa).

6. The process according to claim 1 wherein the LC-MS mobile phase comprises a gradient of formic acid.

7. The process according to claim 1 wherein the LC-MS mobile phase is a ternary mobile phase.

8. The process according to claim 7 wherein ternary mobile phase comprises water, acetonitrile and formic acid.

9. The process according to claim 1 wherein said sample is a biopharmaceutical formulation.

10. The process according to claim 1 wherein said sample comprises at least one protein.

11. The process according to claim 1 wherein said sample comprises at least one monoclonal antibody.

12. The process according to claim 1 wherein said process also includes the step of detecting the oxidation and/or hydrolysis of said polysorbate.

13. A process of monitoring the degradation of at least one polysorbate in a sample comprising implementing the process according to claim 1.

Description

FIGURES

(1) FIG. 1 summarizes PS80 theoretical structure and degradation pathways.

(2) FIG. 2 shows the intensity (in arbitrary units) versus the retention time (in minutes). It illustrates the representative total ion current (TIC) profile of PS80 with peaks labeled as 1—non esterified species, 2—mono esters and 3—higher order esters (a), single ion recording (SIR) for m/z 309.3 indicating the elution of oleate species with peaks labeled as 2.1—POE sorbitan monooleate 2.2—POE isosorbide monooleate and 2.3—POE monooleate (b) and SIR for m/z 283.3 indicating the elution of palmitate species (c).

(3) FIG. 3 represents chromatograms of dioxolanylium ions of oleate ester in positive ionization mode with various percentage of formic acid with a CV 50 (a) and with different CV at 0.1% formic acid (b). Peak area extracted from (b) is plotted against cone voltage for mono ester and higher order ester (c).

(4) FIG. 4 represents chromatograms of oleic acid in negative ionization mode with various percentage of formic acid with CV 15 (a) and with different CV at 0.01% formic acid (b).

(5) FIG. 5 illustrates representative chromatograms of a differential diagnosis of PS80 degradation in thermally stressed (top chromatogram) and unstressed (bottom chromatogram) samples. PS80 oleate subspecies are decreased in thermally stressed sample (a—pos m/z 309.3). No increase of oleic acid is observed (b—neg m/z 281.3) and oxidized byproducts increased drastically (c—pos m/z 325.3).

(6) FIG. 6 illustrates representative chromatograms of a differential diagnosis of PS80 degradation in two different batches A (top chromatogram) or B (bottom chromatogram). PS80 oleate subspecies are decreased in batch A sample (a—pos m/z 309.3). Great increase of oleic acid is observed (b—neg m/z 281.3) and oxidized byproducts are no different in the two batches (c—pos m/z 325.3).

(7) FIG. 7 illustrates the SIR signals in positive ionization mode at 309.3 m/z for PS80 sample at 50 μg/mL (dotted line) and internal standard (solid line) for PEG bis C8 (a) and PEG C12 and PEG C14 (b).

(8) FIG. 8 is an extracted ion chromatogram in positive ionization mode at 171 m/z specific signal of PEG bis C8 for PS80 sample (dotted line—panel b) and PEG bis C8 (solid line—panel a). Retention time (Rt) of peak used as internal standard is given.

(9) FIG. 9 is an extracted ion chromatogram in positive ionization mode at 227.3 m/z specific signal of PEG C12 for PS80 sample (dotted line—panel b) and PEG bis C8 (solid line—panel a). Retention time (Rt) of peak used as internal standard is given.

(10) FIG. 10 shows a single ion recording in positive ionization mode at 255.3 m/z specific signal of PEG C14 for PS80 sample (dotted line) and PEG C14 (solid line). Retention time (Rt) of peak used as internal standard is given.

EXAMPLES

(11) Materials. Polysorbate 80 was obtained from Seppic (Puteaux, France). For LC-MS, 1 g of polysorbate was dissolved in 100 mL of water in volumetric flask to give a 10 g/L stock solution stored at 2-8° C. and protected from light. Acetonitrile LC-MS grade was purchased from Fisher Scientific (Illkirch, France). Purified water from a milliQ system was used. Formic acid, polyethyleneglycol monolaurate (PEG-C12, Mn=±400 g/mol), polyethyleneglycol bis 2 ethyl hexanoate (PEG-bis C8, Mn=±650 g/mol) and oleic acid were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). PEG-C12 was used as an internal standard. PEG-C12 stock solution was prepared by dissolving 500 mg in 100 mL of acetonitrile in volumetric flask to give a 5 g/L stock solution. A working solution at 100 μg/mL was prepared by dilution in acetonitrile. Only one batch of PEG-C12 was used throughout the study. Oleic acid stock solution was prepared by dissolving 100 mg in 100 mL acetonitrile in volumetric flask to give a 1 g/L stock solution. Working standard solution for calibration purpose were prepared in a 20 mL volumetric flask with a PEG-C12 final concentration of 5 μg/mL, with varying PS80 concentration from 5 to 75 μg/mL and varying oleic acid concentration from 1 to 20 μg/mL. Added solvent consisted of a water/acetonitrile mixture (20%/80%).

(12) LC-MS analysis. Reversed-phase separation was performed on an Acquity UPLC system equipped with a QDa mass detector from Waters (Saint Quentin en Yvelines, France). QDa parameters were set as default prior to optimization detailed here after. A zorbax Sb-Aq column (100×2.1 mm; 3.5 μm) from Agilent (Les Ulis, France) was operated at 50° C. with a flow rate of 1 mL/min according to Christiansen et al. (Pharmazie. 2011, 66, 666-671). Mobile phase A and B were respectively water+0.1% formic acid and acetonitrile+0.1% formic acid. Analytical gradient was as followed: 85% A with 15% B for 1 minute followed by linear ramp to 60% B at 6 minutes held until 8 minutes before a linear ramp to 100% B at 10 minutes held for another 3 minutes followed by a return to initial conditions at 13.1 minutes for another 3 minutes. A typical mobile phase gradient is described in following gradient table:

(13) TABLE-US-00001 TABLE 1 Timetable for mobile phase gradient C (acetonitrile + A (water) B (acetonitrile) 0.1% formic acid initial 85% 5% 10% 1 min 85% 5% 10% 6 min 40% 50%  10% 8 min 40% 50%  10% 9 min 20% 70%  10% 9.1 min 18% 0% 82% 10 min  0% 0% 100%  13 min  0% 0% 100%  13.1 min 85% 5% 10%

(14) Analytical conditions such as mobile phase and gradient were adapted according to results obtained.

(15) Sample preparation. Samples from biotherapeutics formulation including mAb and different excipients were used. PS80 concentration was kept at 200 μg/mL. Sample preparation consisted of a protein precipitation step with acetonitrile. To 80 μL of formulated mAb was added 20 μL of internal standard working solution and 300 μL of acetonitrile. After agitation, the mix was submitted to 10 minutes 1500 g centrifugation at 10° C. Supernatant was collected and transferred into HPLC vials. Final concentration is around 40 μg/mL PS80 and 5 μg/mL of internal standard.

(16) Results and Discussion

(17) Separation of PS80 components and identification of fatty acids esters with “in source” dissociation. According to a large number of previous publications, polysorbates are heterogeneous mixture with a great diversity in chemical structure and concentration of esterified species. Moreover, nature of fatty acids involved in ester bond can vary. According to EU pharmacopeia oleic acid and palmitic acid (respectively mono unsaturated and saturated fatty acids) are the two major fatty acids of PS80. Series of low m/z ions characteristic of polyethylene glycol fatty acids ester—called dioxolanylium ion—were used to identify esterified species of PS80. Typical profile of PS80 exhibits different fatty acids ester from mono to tetra ester (FIG. 2a) while single ion recording (SIR) of m/z 309.3 and 283.3 are specific of oleate ester and palmitate ester species (FIGS. 2b and 2c). Chromatograms obtained by SIR were much easier to interpret and integrate due to the artificial gain of chromatographic resolution. Identification of the different ester species was based on work published by Borisov et al, 2015 (supra) on elution order of polysorbate species in reversed phase chromatography (FIGS. 2a and 2b).

(18) LC-MS method development. QDa is able to perform simultaneous analysis in negative and positive ionization mode for the ESI source. To gain advantage of this capability, LC-MS conditions were carefully optimized. Systematic evaluation of formic acid percentage in mobile phase and cone voltage (CV) of ESI source allowed to obtain simultaneous separation of PS80 and its degradation product with a high sensitivity. Formic acid percentage from 0.01% to 0.1% had little influence on oleate mono esters detection but was critical for higher order esters (FIG. 3a). Increased cone voltage resulted in increased signal represented by peak area of mono and higher order ester (FIGS. 3b and 3c) up to CV50. Above CV50, dramatic loss of signal was experienced. This phenomenon was explained by in source fragmentation. Increased cone voltage induced increased in source fragmentation which resulted in formation of more dioxolanylium ions up to CV50. For higher CV, intense in source fragmentation resulted in formation of secondary fragment explaining loss of signal.

(19) Detection in negative ionization mode of oleic acid required little proportion of formic acid in mobile phase—not more than 0.01% (FIG. 4a). Detection of unfragmented molecular ion of oleic acid required low cone voltage to minimize in source fragmentation (FIG. 4b). Order of elution between mono esters, oleic acid and higher order esters made it possible to put in place a ternary mobile phase gradient in order to vary acetonitrile and formic acid percentage independently throughout the analysis. LC-MS method was adapted from experimental section as follows. Mobile phase gradient with 85% A (water), 5% B (acetonitrile) and 10% C (acetonitrile+0.1% formic acid) for 1 minute followed by a linear ramp to 40% A, 50% B and 10% C at 6 minutes held for 2 minutes. Another ramp to 20% A 70% B and 10% Cat 9 minutes was added. At 9.1 minutes, a switch to 18% A with 82% C was added to increase formic acid percentage to 0.1% and the ramp continued to 100% C at 10 minutes and held for 3 minutes before returning to initial conditions at 13.1 minutes. The switch between mobile phase B and C at 9 minutes was added to change formic acid percentage in mobile phase from 0.01% to 0.1% while keeping acetonitrile percentage in mobile phase on the same gradient ramp. SIR signals in positive ionization mode were recorded with CV50 while SIR signals in negative ionization mode were recorded with CV15.

(20) Polysorbate quantification by external calibration. In previously published methods, PS80 or PS20 is quantified by using a suitable surrogate with an external calibration curve. Well known examples are methods based on hydrolysis and subsequent esterification in methanol to form fatty acid methyl ester like methyl oleate. This methyl oleate is then considered a surrogate of PS80 and quantified. Here surrogate for quantification was peak 2.1 (FIG. 2b) as this peak corresponds to POE sorbitan mono oleate which is a structure close to the theoretical structure of PS80. Based on this peak, a seven concentration levels calibration curve was constructed with concentration ranging from 5 to 75 μg/mL of PS80 in aqueous solution. Linearity was evaluated with unweighted linear regression of peak area versus concentration. Linearity was poor with r.sup.2 of 0.977. Residuals exhibited a quadratic behavior explaining poor linearity performance. Repeatability evaluated by injection of 6 replicates of every calibration level was bad with RSD % of peak area above 22%. It was hypothesized that these poor results were due to competition during “in-source” fragmentation as this fragmentation is not well controlled in ESI source of the QDa, a single quadrupole.

(21) Comparison between external and internal calibration. To overcome this issue, an internal standard (IS) was used in order to perform internal calibration instead of external calibration. Most commonly used internal standard in LC-MS for quantification purpose are deuterated compound. Given the heterogeneity of PS80 mixture this approach was not considered. Compounds with similar chemical structures and thus similar in source fragmentation pathways were chosen: polyethyleneglycol monomyristate (PEG-C14), polyethyleneglycol monolaurate (PEG-C12) and polyethyleneglycol bis 2 ethyl hexanoate (PEG-bis C8). PEG-C12 and PEG-bis C8 are commercially available (Sigma-Aldrich/Merck). PEG-C14 could not be purchased from classical chemical compounds manufacturers but was found as an impurity of PEG-C12.

(22) As this PEG-C14 is made of PEG esterified with a C14 saturated fatty acids it was subjected to the same fragmentation resulting in formation of characteristic dioxolanylium ions at m/z 255.3 in positive ionization mode. Interferences between PS80 and PEG-C14/PEG-C12/PEG bis C8 were checked prior to using it as an internal standard. No signal from PS80 was interfering with PEG-C14/PEG-C12/PEG bis C8 signal and vice versa. Retention times of PEG-C14/PEG-C12/PEG bis C8 and PS80 surrogate peak were close, so that they underwent the same competition during in source fragmentation.

(23) The lack of interferences of signals from PEG-C14/PEG-C12/PEG bis C8 with the signals of PS80 is illustrated by FIG. 7.

(24) The lack of interference of signals from PS80 with the signals of PEG-C14/PEG-C12/PEG bis C8 as well as retention times of peak of interest from PEG-C14/PEG-C12/PEG bis C8 is illustrated by FIGS. 8 to 10.

(25) For comparison purpose, internal calibration was evaluated with the same set of experiment as external calibration. Linearity was evaluated with unweighted linear regression of peak area ratio of PS80 surrogate peak over IS versus concentration. Linearity was good with r.sup.2 greater than 0.999. Residuals were more or less randomly distributed and relative bias was kept below 10%. Repeatability evaluated by injection of 6 replicates of every calibration level was again good with RSD % of peak area ratio below 6% for all calibration levels. Table 2 compares these results about repeatability for both external and internal calibration with: polyethyleneglycol monomyristate (PEG-C14), polyethyleneglycol monolaurate (PEG-C12), polyethyleneglycol bis 2 ethyl hexanoate (PEG-bis C8). Comparison between these IS is based on linearity (r.sup.2) and repeatability (RSD %).

(26) TABLE-US-00002 TABLE 2 RSD % value of peak area (external calibration) and peak area ratio (internal calibration) for 6 replicates (except for PEG bis C8, only 4 replicates). Linearity Target concentration (μg/mL) r.sup.2 5 7.5 10 20 30 50 75 External calibration 0.977 27.0 27.8 27.1 24.8 23.4 22.8 23.1 Internal PEG C14 0.999 4.8 5.4 4.3 4.3 4.6 5.1 5.5 calibration PEG C12 0.991 6.1 5.5 4.3 3.0 3.4 4.4 5.9 PEG bis C8 0.988 6.1 3.8 3.5 1.8 2.2 1.7 1.3

(27) On the basis of these data, internal calibration such as PEG-C12, PEG-C14 and PEG bis C8 was shown to exhibit higher repeatability over external calibration. These internal standards exhibit similar repeatability behavior.

(28) Oleic acid quantification. As the major byproducts of PS80 hydrolysis, oleic acid is a parameter to follow. Several methods as reported in Martos et al (2017, supra) were developed to quantify free fatty acids—including oleic acid—released by PS20 or PS80 hydrolysis.

(29) Thanks to QDa features, SIR signal of free oleic acid (m/z 281.3 in negative ionization mode) was recorded in the same analytical run along SIR signals of intact and oxidized PS80. Oleic acid was quantified by external calibration with six calibration levels—from 1 μg/mL to 20 μg/mL—added to calibration level of PS80. For example a typical calibration level with 20 μg/mL of PS80 contained also 5 μg/mL internal standard and 5 μg/mL oleic acid. Linearity was evaluated with unweighted linear regression of peak area against concentration. Linearity was good with r.sup.2 greater than 0.997 and residuals randomly distributed. Repeatability was assessed in a similar manner to that of PS80. RSD % values were between 6.0 to 13.1% with a maximum value observed for a 2 μg/mL concentration. Although RSD % values were higher than expected the assay was deemed satisfactory since this information would be used in case of investigation only. It was hypothesized that these high values were due to high baseline noise in negative ionization mode because of mobile phase composition. Indeed a compromise was made during development in favor of PS80 oleate ester species detection.

(30) Application to formulated mAb samples. The method was applied to different cases of mAb formulation with different mAb properties and excipients but each time with 200 μg/mL PS80.

(31) mAb1 formulated at 5 mg/mL was used to evaluate repeatability of sample preparation as well as method accuracy. Three preparations were made and analyzed on three consecutive days resulting in nine determination of PS80 concentration. Overall mean was 182 μg/mL with 3.1% RSD value. These results were in line with previous measurement with a classical mixed-mode LC-CAD analysis where a mean value of 188 μg/mL with 1.2% RSD on six measurements was found.

(32) mAb2 formulated at 20 mg/mL was submitted to thermal stress conditions two weeks at 40° C. Results showed a drastic decrease of PS80 content from 198 to less than 25 μg/mL. More information was retrieved from other recorded signals. Chromatogram of palmitate ester species showed no difference between stressed and unstressed samples which indicates that degradation affects only oleate ester species typically observed in case of oxidation. No trace of free oleic acid was found ruling out hydrolysis as degradation pathways. SIR signals m/z 325.3 (in positive ionization mode) showed great differences between stressed and unstressed samples (FIG. 5). Ions at m/z 325.3 were described as characteristic byproducts of PS80 oxidation either as an epoxy-stearate or hydroxy-oleate modified polysorbate. Here the whole picture of PS80 in this degraded sample was obtained with three different chromatograms recorded in one analysis allowing to conclude unambiguously that PS80 decrease is due to an oxidation.

(33) mAb3 formulated at 50 mg/mL was another example. Samples from different batches were kept at 5° C. for almost a month before analysis. Results showed differences in PS80 content between batch A and batch B respectively 100 and 190 μg/mL instead of 200 μg/mL. This is illustrated in FIG. 6a when looking at peak 2.1, surrogate for PS80 quantification. SIR signals of oxidized PS80 revealed no difference between the batches which meant no increase of PS80 oxidation byproducts especially when compared to intensity observed in case of mAb2. Free oleic acid in solution was found in the problematic batch A at a concentration around 25 μg/mL. Only small traces of oleic acid were found in batch B (FIG. 6). Presence of free oleic acid is an evidence of PS80 hydrolysis. Given all information retrieved from this analysis, it was concluded that PS80 degradation was due to hydrolysis. As higher order esters were not impacted, it was hypothesized that this hydrolysis was from enzymatical origin.

(34) Conclusion. PS80 is widely used in biotherapeutics formulation. Recently, growing concerns were raised about polysorbates stability in drug product and its ability to maintain its protective role against protein aggregation. Efficient methods to measure PS80 content and monitor its degradation were reported over the past few years (Martos et al, supra). Among them, LC-MS based methods showed promising results in terms of characterization and semi-quantitative information. Characteristic signals of dioxolanylium ions of fatty acids esters were used following “in source” CID to significantly simplify chromatograms and identify more easily PS80 subclasses. In previous work mentioned, their method was not verified for being quantitative. In this study, the same methodology was applied with the aim to use it for quantification at a QC level with a single quadrupole mass detector. Unfortunately first results of quantification confirmed Borisov (supra) conclusion. This issue was overcome by using a carefully chosen internal standard with similar chemical properties. It was shown that the formic acid gradient allowed for more sensitive and simultaneous detection in both positive and negative ionization mode of all compounds of interest from PS80 ester subclasses to free fatty acids.

(35) This method was successfully applied to different cases of PS80 monitoring and especially two of them with distinctive features of PS80 degradation. In these two cases, root cause of PS80 degradation was identified using only one 16 minutes analysis. To reach out the same goal without this method it would have required three to four different methods: one for PS80 quantification (mixed-mode LC-CAD) plus one for PS80 profiling (reversed-phase CAD) along with other methods for identification of characteristic byproducts (oxidized PS80 and free oleic acid). The method presented here can efficiently replace the extensive analytical toolbox needed up to now.

(36) The aim of this method is to provide accurate and stability indicating measurement of PS80 as well as valuable information in order to identify unambiguously root cause of PS80 degradation.