- Email: [email protected]
Qualitative and quantitative assessment of marketed erythropoiesis-stimulating agents by capillary electrophoresis
Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
Contents lists available at SciVerse ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Short communication
Qualitative and quantitative assessment of marketed erythropoiesis-stimulating agents by capillary electrophoresis Sylvie Boucher, Anita Kane, Michel Girard ∗ Biologics and Genetic Therapies Directorate, Health Canada, 251- Sir Frederick Banting Driveway, Ottawa, ON K1A 0K9, Canada
a r t i c l e
i n f o
Article history: Received 22 May 2012 Received in revised form 16 August 2012 Accepted 19 August 2012 Available online 24 August 2012 Keywords: Erythropoietin Darbepoetin Formulation Capillary electrophoresis Erythropoiesis stimulating agent
a b s t r a c t Formulated erythropoiesis stimulating agents (ESAs) containing erythropoietin (EPO)-alpha, EPO-beta or darbepoetin-alpha were analyzed by capillary electrophoresis with a previously published method requiring no sample pre-treatment [1]. In this study, the method proved to be applicable to all formulations encountered, that is, in the presence of polysorbate 80, polysorbate 20 or human serum albumin as major excipients, thus extending the range of products that can be analyzed without pre-treatment. Method performance was evaluated and showed good linearity, range, precision and sensitivity. No significant matrix effects were observed for the various formulations. The ability of the method to resolve isoforms of each of the three active ingredients enabled comparison of the isoform distribution of finished products with that of the respective drug substance. In general, finished products and their corresponding drug substances showed similar isoform distribution and all were within manufacturer specifications. In addition, the content in active ingredient in the various dosage strengths was found to be in close agreement with the label claims with the exception of 2 out of 131 containers analyzed. Overall, this study demonstrated that the capillary zone electrophoresis method could be successfully applied to the analysis of most of the ESA products currently on the market in North America and Europe and that all products were found to have good batch-to-batch consistency. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved.
1. Introduction Erythropoietin (EPO) was among the first recombinant protein pharmaceuticals to reach the market at the end of the 1980s and it has been one of the most successful, with worldwide annual sales exceeding $10 billion US dollars in 2010 [2]. Early products sold in North America and Europe contained EPO-␣ (Eprex, Epogen, Procrit) or EPO- (Recormon/NeoRecormon). These two forms of human EPO have the same polypeptide sequence but differ in their glycosylation pattern. The product Eprex was originally formulated with human serum albumin (HSA) and, in 1998 the manufacturer changed the formulation to contain polysorbate (PS) 80 as a consequence of the European requirement to remove serum-derived components in pharmaceuticals [3]. A new ESA, darbepoetin-␣ (DPO-␣) (Aranesp), received marketing authorization in 2001. It is an analog of EPO and features a modified polypeptide sequence in addition to an increased number of glycosylation sites. In addition to these products, several EPO-␣ and EPO- biosimilars have received marketing authorization throughout the world since patents of innovator products have expired [4].
There are currently few physicochemical methods that enable the assessment of the active ingredient in biopharmaceutical final formulations. This is due in part to the small amounts of active ingredient present and to the large amounts of excipients required to enhance product stability, to prevent non-specific adsorption or to provide isotonic conditions suitable for injectables. Thus, final formulations often contain a varied mixture of excipients that may include small molecules such as salts, sugars or amino acids, in addition to larger molecules like HSA or PS. In this context, it has become desirable to develop widely applicable methods for the evaluation of products from multiple sources. Efforts from this group have been mainly concerned with the analysis of finished products in a direct manner, that is, without pre-treatment in order to minimize loss of product or generation of artifacts. Herein, we report a wideranging study of ESA products from North America and Europe where the assessment of the integrity of the active ingredient and the isoform distribution have been highlighted.
2. Materials and methods 2.1. Materials
∗ Corresponding author. Tel.: +1 613 952 0399; fax: +1 613 941 8933. E-mail address: [email protected] (M. Girard).
Formulated products and drug substances of four ESAs were generously donated by manufacturers from North America and Europe. They are identified as products A–D throughout this
0731-7085/$ – see front matter. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jpba.2012.08.021
208
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
Table 1 Description of ESA products analyzed during the study. Product
Dosage strength (per vial)
Lots
Excipient
Active ingredient
A
3000 IU 10,000 IU 20,000 IU
A1 A2 A3 A4 A5 A6 A7 A8 A9
Human serum albumin
EPO-␣
B
2000 IU 4000 IU 10,000 IU 40,000 IU
B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12
Polysorbate 80
EPO-␣
C
10,000 IU 20,000 IU 100,000 IU
C1 C2 C3 C4 C5 C6
Polysorbate 20
EPO-
Da
0.04 mg/mL 0.20 mg/mL 0.50 mg/mL
Da1 Da2 Da3 Da4 Da5 Da6 Da7 Da8 Da9
Human serum albumin
DPO-␣
Db
0.04 mg/mL 0.20 mg/mL 0.50 mg/mL
Db10 Db11 Db12 Db13 Db14 Db15 Db16 Db17 Db18
Polysorbate 80
DPO-␣
article. In brief, the study consisted in analyzing vials of multiple lots of three dosage strengths per product as shown in Table 1. Amine regenerator solution and eCap amine capillaries were from Beckman (Fullerton, CA). Sodium dihydrogen phosphate monohydrate and nickel chloride hexahydrate were from Sigma–Aldrich. All solvents were of HPLC quality. 2.2. Methods 2.2.1. Capillary zone electrophoresis setup Capillary zone electrophoresis (CZE) was carried out on a Beckman MDQ P/ACE capillary electrophoresis system (Beckman, Fullerton, CA) fitted with a variable wavelength UV detector and using Beckman 32 KaratTM Software for operation and data analysis. Separations were carried out on eCap amine capillaries, 50.2 cm × 50 m i.d. (effective length 40 cm) fitted in a cassette with an 800 m aperture. The background electrolyte (BGE) consisted of 200 mM NaH2 PO4 containing 1 mM NiCl2 (13.8 g NaH2 PO4 ·H2 O and 0.11885 g NiCl2 ·6H2 O in 500 mL ddH2 O). The pH was adjusted to 4.0 with dilute acetic acid. For analysis of product D, the phosphate concentration in the BGE was increased to 300 mM. 2.2.2. Capillary cleaning and conditioning A capillary cleaning procedure was applied to new capillaries or after running several injection sequences to re-establish proper working conditions. The following rinses were performed consecutively by applying pressure at 20 psi in the forward mode: methanol (5 min), 0.1 M HCl (5 min), doubly deionized water (ddH2 O) (5 min), 1.0 M NaOH (2 min), ddH2 O (5 min) and amine regenerator (5 min). A capillary conditioning procedure was performed after short storage time or upon reduced capillary performance (RSD for migration time greater than 10%). Consecutive rinses were carried out at 20 psi under pressure in the forward mode: 0.25 M NaOH (15 min), ddH2 O (5 min), amine regenerator (30 min) and BGE (15 min). 2.2.3. Separation conditions Electropherograms were monitored at 200 nm. The capillary temperature was maintained at 20 ◦ C and the sample area was at 10 ◦ C. In order to minimize the effect of ion depletion [1], largevolume BGE reservoirs of about 21 mL were used. Separations were carried out as sequences that included automatic pre-injection rinses in the following manner: 0.1 M NaOH (2 min, 20 psi, forward), amine regenerator (5 min, 20 psi, forward), BGE (10 min, 20 psi, forward), water dip (3 s), sample injection (8 s, 0.5 psi, forward), water
dip (3 s) and separate at 8.0 kV for 40 min by reverse polarity with a 0.5 min ramp.
2.2.4. Calibration curves Calibration curves for products A–D were prepared using the corresponding drug substance. At a minimum, six standard solutions encompassing the lowest and highest dosage strengths were prepared by serial dilution from a primary standard solution made from the drug substance: (1) drug substance A (DS A) – primary standard solution: 30 L DS A + 243 L ddH2 O (48,459 IU/mL); standards: 2423, 2908, 4846, 9692, 14,538, 19,384 and 24,230 IU/mL; (2) drug substance B (DS B) – primary standard solution: 200 L DS B + 722.6 L ddH2 O (80,000 IU/mL); standards: 2000, 4000, 6000, 8000, 10,000, 14,000, 20,000, 30,000, 40,000 and 50,000 IU/mL; (3) drug substance C (DS C) – primary standard solution: 759 L DS C + 741 L ddH2 O (187,097 IU/mL); standards: 4677, 9355, 18,710, 37,419, 56,129, 93,548 and 112,258 IU/mL; (4) drug substance D – primary standard solution: 754 L DS D + 749 L ddH2 O (1.00 mg/mL); standards: 0.50, 0.20, 0.10, 0.05, 0.02, 0.01 mg/mL. The linearity of the detector response over the range of concentrations was verified by plotting the corrected peak area percent (%CPA = % peak area/migration time) of the sum of all isoform peaks using isoform I2 as migration time marker.
2.2.5. Analysis of finished products Finished products in liquid form were analyzed directly by injecting the solution without any pre-treatment. Finished products in solid form were first reconstituted to the prescribed concentration with the accompanying diluent and the resulting solution was analyzed without any other treatment. A typical sequence consisted of replicate injections (n = 5) of two standard solutions and all vials of a single lot (e.g., a total of 25 injections for 3 vials). The two standard solutions included in the sequence corresponded to the respective drug substance (see Section 2.2.4) and represented one lower and one higher concentration than that of the vial label content. For content determination, peak areas were measured as the sum of all detected isoforms. Corrected peak areas were calculated by dividing peak areas by the migration time of isoform I2 as marker. The content was determined by interpolation from the plot of %CPA against concentration. For isoform distribution determination, peak areas of individual isoforms were measured. Plots of %CPA against isoform number were prepared.
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
A EP Method A. 4
0.006
2
0.005
0.005
5
0.004
0.004
0.003
0.003
0.002
0.002
0.001
6 8
1 0.001
7
0.000
0.000
-0.001
-0.001
3
Bietlot tlot-Girard Girard B. Bie B Method
0.012
AU
0.006
0.014
0.007
-0.002 -0.002 45.0 47.5 50.0 52.5 55.0 57.5 60.0 62.5 65.0 67.5 70.0 72.5 75.0 77.5 80.0 82.5 85.0
Absorbance @ 200 nm (AU)
Absorbance @ 200 nm (AU)
0.007
0.008
3
0.014 0.012
4
0.010
0.010
2
0.008
0.008
0.006
0.006
5
0.004
0.004
0.002
AU
0.008
209
0.002
0.000
0.000
6 7,8
1
-0.002
-0.002
-0.004
-0.004
-0.006
-0.006 18
19
20
21
Migration time (min)
22
23
24
25
26
27
28
Migration time (min)
Pk #
Bietlot-Girard Method % Corrected Peak Area
EP Method % Corrected Peak Area
1
1.36
1.76
2
16.95
18.57
3
29.23
30.18
4
27.61
27.6
5
17.13
16.06
6
5.49
4.55
7
2.23 (7+8)
1.28
8
-
<0.3
Totals
100
100
Fig. 1. Electropherograms of the European pharmacopoeia biological reference preparation for erythropoietin by (A) the EP monograph method and (B) the method of Bietlot and Girard [1].
3. Results and discussion Samples of finished ESA products A–D, corresponding to a wide range of available dosage strengths, along with their corresponding drug substance (DS A–D) were obtained from sources in Europe and North America (Table 1). Products A and B contained EPO␣ formulated with HSA and polysorbate 80 (PS80), respectively. Product C contained EPO- formulated with polysorbate 20 (PS20). Products D contained DPO-␣ formulated with HSA (product Da) or PS80 (product Db). The primary objectives of this study being the evaluation of the integrity of the active ingredients in terms of isoform distribution and content, capillary electrophoresis was deemed suitable to achieve these objectives as this technique has been frequently used for the separation of intact EPO isoforms [4,5]. 3.1. Capillary zone electrophoresis method The European Pharmacopoeia (EP) monograph for EPO includes an identification test based on CZE [6] that provides baseline separation of 8 isoforms, numbered I1–I8 from the most acidic to the most basic, for the EP biological reference preparation (BRP) for EPO (Fig. 1A). However, the method could not be applied to the direct analysis of formulated products as removal of excipients is required prior to analysis. Furthermore, while efficient removal of small molecule excipients is possible, attempts at removing HSA led to low EPO recovery and apparent loss of resolution (data not shown). For these reasons the use of this method was abandoned. We previously reported a CZE method for the detection and quantification of EPO-␣ and EPO- in HSA-containing formulations in a direct fashion, that is, without the need to remove excipients
[1]. The relatively high ionic strength of the BGE, that is, 200 mM phosphate, provided the necessary conditions for the analysis of final formulations. Analysis of the EP BRP by this method enabled separation of five major isoforms, I2–I6 and 3 minor isoforms I1, I7–I8 (Fig. 1B). As expected when using reversed-polarity, the migration order was reversed compared to that of the EP method, with the most acidic isoform, I1 migrating first and the most basic isoform, I8 migrating last. While only partial isoform resolution was achieved when compared to the EP method, it was deemed sufficient for comparative purposes as demonstrated by the similar relative isoform levels measured by the two methods (see Table in Fig. 1). However, its applicability to the analysis of all of the ESAs and their various formulations, particularly for polysorbate formulations and for DPO-␣ needed to be demonstrated. Typical electropherograms of drug substances, DS A–C are presented in Fig. 2A–C. The EPO-␣ drug substances, DS A and DS B (Fig. 2A and B, respectively) were each separated into 6 isoforms that included 2 minor isoforms, I1 and I6, and 4 major isoforms, I2I5. The EPO- drug substance, DS C (Fig. 2C) was separated into 8 isoforms consisting of the 3 minor isoforms, I1, I7 and I8, and the 5 major isoforms, I2–I6. The analysis of DPO-␣ (DS D) showed a profile indicative of the presence of multiple isoforms that migrated more quickly than those from either EPO-␣ or EPO-. The faster migration was expected and could be easily explained on the basis of the increased number of terminal sialic acids resulting from the additional asparagine-linked glycans in DPO-␣. While the isoform separation was not optimal under the standard separation conditions it could be improved by increasing the BGE phosphate concentration to 300 mM. This resulted in the partial separation of 5 isoforms, numbered I1–I5 from the most acidic to the most basic,
210
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213 0.05
A
0.03
0.02
0.02
0.01
0.01
AU
HSA 0.03
Absorbance @ 200 nm (AU)
0.04
EPO
0.00
0.00
-0.01
0.007
0.006
0.006
B
0.005
0.005
0.004
0.004
0.003
0.003
0.002
0.002
0.001
0.001
0.000
0.000
-0.001
-0.001
-0.002
-0.002
-0.003
-0.01 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
-0.003 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0 42.5 45.0 47.5 50.0
Migration time (min) 0.016
0.016
0.014
0.014
C
0.07
0.010
0.008
0.008
0.006
0.006
0.004
0.004
0.002
0.002
0.000
0.000
-0.002
-0.002
-0.004
-0.004
16
18
20
22
24
26
28
30
32
34
36
38
40
Migration time (min)
Absorbance @ 200 nm (AU)
0.012
0.010
0.07
D
AU
0.012
Absorbance @ 200 nm (AU)
Migration time (min)
0.06
0.06
0.05
0.05
0.04
0.04
0.03
0.03
0.02
0.02
0.01
0.01
2.5
5.0
AU
Absorbance @ 200 nm (AU)
0.04
0.007
AU
0.05
7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
Migration time (min)
Fig. 2. Typical electropherograms of finished products A–D (red and blue traces) with their respective drug substance (DS A–D, black traces): (A) product A4 and DS A; (B) product B7 and DS B; (C) product C2 and DS C and (D) products Da7, Db16 and DS D. Insets: enlarged view of separated isoforms labeled from the most acidic to the most basic. Separation conditions: eCap Amine capillary; 200 mM sodium phosphate, 1 mM nickel chloride, pH 4.0; voltage: −8 kV (for samples D, BGE was 300 mM phosphate, 1 mM nickel chloride, pH 4.0). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
with 4 major isoforms, I1–I4, and one minor isoform, I5 (Fig. 2D). The 300 mM phosphate buffer concentration was chosen to perform all subsequent analyses of DS D and product D. As mentioned above, it was essential to verify that all types of formulations would be amenable to the detection of the active ingredient and to the separation of isoforms. Typical electropherograms of finished products A–D are shown in Fig. 2 alongside their respective drug substance. In all cases no interference from excipients were observed and the isoform separation and distribution was similar to that of the corresponding drug substance. These results provided evidence that the CZE method could be successfully extended to the direct analysis of EPO polysorbate formulations. This was also the case for products Da and Db, which were formulated with HSA and PS80, respectively. Overall, these results showed that the range of products that can be directly analyzed using this CZE method covered a large part of the ESA products currently on the market in North America and Europe. 3.2. Method performance for drug substances The performance of the method in terms of matrix effect, linearity, range, precision and sensitivity was evaluated for each of the drug substances, DS A–D. Matrix effect on the detector response was carried out using simulated formulation solutions obtained by mixing the drug substance and either HSA, PS80 or PS20 at concentrations close to those in finished products. In all cases, %CPA
for simulated solutions were within 2–3% of their corresponding drug substance solutions prepared at the same concentration, an indication that no substantial matrix effect was present (data not shown). Table 2 presents performance data for each of the four drug substances. Linearity of the detector response was assessed for concentration ranges that generally encompassed at least 80–120% of all of the available dosages to be examined. Plots of the %CPA for the sum of all isoforms, using isoform I2 as peak marker, against the concentration showed linear relationships with coefficients of determination, R2 , better than 0.996 over the entire range. Good migration time and %CPA repeatability were obtained on replicate injections of standard solutions at three concentrations (low, medium and high) within the same sequence. Typically, migration time repeatability was found to be very good for all four drug substances with %RSD values below 1%. Good repeatability was also obtained for %CPA with %RSD values below 7% for the average %RSD values at three concentrations. Intermediate precision from replicate injections of standards at the same three concentrations in separate sequences showed, as anticipated, greater variability for migration time, with %RSD values of less than 5%. Similarly, increased variability was found for %CPA, with average %RSD for three concentrations between 4 and 11%. These values were considered acceptable for the purpose of the study. Limits of detection (LODs) were set for the lowest concentration giving a signal-to-noise ratio greater than 3. For DS A and B, LODs
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
211
Table 2 Method performance for drug substances A–D. Drug substance
A B C D a b c d
Repeatability
Intermediate precision
Limit of detectiond
MTa (Avg %RSDb )
CPAa (Avg %RSDb )
MT (Avg %RSDc )
CPA (Avg %RSDc )
0.11 0.62 0.18 0.17
5.54 4.21 6.47 6.53
1.67 4.62 1.05 1.82
10.64 10.36 4.26 4.56
2000 IU/mL 2000 IU/mL 2500 IU/mL 0.010 mg/mL
Limit of quantificationd
2500 IU/mL 3000 IU/mL 4000 IU/mL 0.020 mg/mL
Linearity R2
Slope
Y intercept
0.9999 0.9963 0.9991 1.0000
0.3216 0.6164 0.613 86,653.2
−211.3 −1367.2 −2359.0 −342.1
MT, migration time; CPA, corrected peak area. Avg %RSD, average %RSD from replicate injections (n = 5) of standards at three concentrations in the same sequence (see text). Avg %RSD, average %RSD from replicate injections (n = 5) of standards at three concentrations in different sequences (see text). Limit of detection was set for the lowest concentration giving a S/N ≥ 3 (n = 5); limit of quantification was for the lowest concentration giving a RSD < 10% (n = 5).
were found at 2000 IU/mL while for DS C a value of 2500 IU/mL was obtained. The higher LOD value for product C could be explained on the fact that there was a greater number of minor isoforms (I1, I7 and I8) contributing to the total peak area and that these were more difficult to detect at very low concentrations. The LOD for DS D was found to be at 0.01 mg/mL. The limits of quantification (LOQs) were determined for the lowest concentration giving %RSD values lower than 10% for replicate injections (n = 5). Again, the LOQ for DS C was found to be at a higher concentration than for both DS A and B for the same reasons. The LOQ for DS-D was found to be 0.02 mg/mL. Overall, results from the performance assessment demonstrated that the CZE method met the usual criteria of acceptability for the qualitative and quantitative analysis of the four types of ESA products under investigation. In addition, these results showed that all of the finished products available, with the exception of lots B1–B3 of product B, where the label dosage was 2000 IU/mL, could be reasonably expected to be amenable to quantification since their label content was at or greater than the LOQ of the respective drug substance.
product, a comparison of replicate injections of the drug substance (panel A), replicate injections of a given dosage strength of the finished product (panel B), different vials of a given dosage strength of the finished product (panel C) and different lots of a given dosage strength of the finished product (panel D) is presented. Products A and B (Figs. S1 and S2, respectively) showed a high degree of consistency between drug substances and finished products, with only minor variations (less than 3%) in the relative amounts of the major isoforms I3–I5. For product C (Fig. S3), slightly larger variations were apparent between the drug substance and the finished product, especially with regards to isoforms 5 and 7 + 8. This was likely a consequence of the difficulty in obtaining highly reproducible peak integration for these isoforms since they were not always well resolved. The two types of formulation for product D (Figs. S4 and S5) showed very similar isoform distributions as well as a high degree of consistency between finished products and the drug substance. Despite the minor variations noted above all products analyzed were within the manufacturer specifications with regards to the relative isoform content.
3.3. Analysis of finished products As indicated above, the assessment of the integrity of the active ingredient in finished products used a two-prong approach, namely, the verification of the isoform distribution and the estimation of the content in active ingredient. All of the products analyzed were within their expiry date. 3.3.1. Isoform distribution The relationship between in vivo biological activity and the type of N-glycans and their degree of sialylation in EPO has been described [7,8]. Variable terminal sialic acid content of the glycans is responsible for the overall charge distribution of the molecule resulting in the presence of isoforms. Thus, knowledge of the isoform distribution is an important issue for ESAs. Furthermore, a quantitative method which assessed glycan characteristics such as isoform percentages and total sialic acid content, as well as glycan branching and branch repeats, to determine such bioactivity, was found to be reducible to the isoform profile and total sialic acid content [9]. In addition, changes in the isoform distribution may be acceptable as was recently demonstrated between batches of the same ESA product [10]. In that case, isoform distribution changes purportedly coincided with a process modification and may be reflective of this modification. In particular, very good batch-tobatch consistencies were observed in each set of batches prior to and after the isoform distribution change, which was deemed to be within acceptable limits. In this study, the isoform distribution was evaluated by determining the relative amount of each isoform in a given product. Graphical representations of results are shown in Supplementary Figs. S1–S5 for products A, B, C, Da and Db, respectively. For each
3.3.2. Content in active ingredient In view of the absence of matrix effects and that linearity was achieved over the entire range of concentrations, quantification of ESA content in finished products was determined by interpolation from a two-point standard calibration plot. In most cases, the two standards used in the calibration plots represented concentrations below and above the expected finished product concentration stated on the container label. Fig. 3 shows a graphical representation of the overall results from the assessment of the content in active ingredient. The content was determined by interpolation (and in a few instances, extrapolation) on plots of the %CPA against the concentration prepared from two standard solutions injected during the same sequence. In all cases, the sum of the corrected peak area of all detected isoforms was used and isoform I2 was used as peak marker for the calculation of the corrected peak area. All vials from a single lot were analyzed during the same sequence and the average vial content was used for plotting in Fig. 3. Thus, each rectangular bar corresponds to the average vial content from a single lot. The round bar represents the average content of all lots. The x axis represents the label dosage strength. It must be noted here that differences in content compared to the label dosage strengths were expected taking into account that the content was determined in relation to drug substance samples that were not the same as those used in the formulated products. As such, differences may result due to the fact that the specific activity of the drug substances may vary and that manufacturers may also overfill in order to compensate for losses from adsorption. In addition, adsorption is a time-dependent process and the different lots of different products varied in time since filling and time to expiry date.
212
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
Fig. 3. Graphical representation of the content in active ingredient in finished products determined by CZE. In each panel, the measured content of each lot (square bar) of the various dosage strengths are plotted. A square bar represents the average lot content measured from three vials. A round bar represents the average of the lots shown on its left. For the dash blue square bar, see text. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Overall results showed that the measured content in active ingredient for the vast majority of products was close to the expected value, generally within 15% of the label claim. This was particularly evident for products A, B and D. For example, the average content found for the three 3000 IU vials of product A (Fig. 3 – product A) was 2875 IU, which corresponded to 95.0% of the label claim. The RSD values were also found to be good, with values within 10% for the average of three lots. This was an indication of the satisfactory lot-to-lot consistency of the various products. Results for product C require specific comments with regards to the measured content. For the 10,000 IU and 20,000 IU dosage strengths,
only 2 lots and 1 lot, respectively, were available for analysis and, as such, results may not be viewed in the same manner as for the other products and are presented here for guidance purposes only. For the 100,000 IU dosage strength, a single vial of lot C6 was available. The electropherogram showed a wide peak migrating at the expected time for EPO but without distinct isoform peaks (Fig. 4). In addition, the measured content was lower than that of the other two lots as shown by the hatched bar in Fig. 3 – product C. These results suggested that the integrity of EPO in these samples may have been compromised to some extent; however, further investigation on the possible cause could not be carried out at the time
0.040 0.035 0.030
PS20
0.025
AU
0.020
EPO
0.015 0.010 0.005 0.000 -0.005
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Minutes Fig. 4. Electropherograms of vials of product C at 100,000 IU with (red trace) and without (blue trace) resolved isoform peaks (see text for details). Separation conditions as in Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
of these experiments. A similar situation occurred for one of the 0.5 mg/mL lot of product Da where no active ingredient could be detected in any of the three vials. Again, it was not possible to investigate the possible cause at the time of these experiments. As such, the average content of 0.587 mg/mL was based on the average of two lots. Overall, these results showed that marketed products generally showed good lot-to-lot consistency with regards to isoform distribution and active ingredient content as measured by capillary electrophoresis. 4. Conclusion Results in this study have demonstrated that capillary zone electrophoresis was adequate for the monitoring of the active ingredient in all of the ESA products analyzed no matter the type of formulation encountered. This expands the types of products that can be analyzed directly, that is, without pretreatment. This method also enabled to examine major issues with regards to finished products, that is, the integrity of the active ingredient and its content. Products analyzed showed high level of consistency between vials of the same lot and between lots. Acknowledgments We are grateful to Drs. Mary Alice Hefford and Jeremy Kunkel (Health Canada) for fruitful discussions during the preparation of this article.
213
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jpba.2012.08.021. References [1] H.P. Bietlot, M. Girard, Analysis of recombinant human erythropoietin in drug formulations by high performance capillary electrophoresis, J. Chromatogr. A 759 (1997) 177–184. [2] G. Walsh, Biopharmaceutical benchmarks 2010, Nat. Biotechnol. 28 (2010) 917–924. [3] K. Boven, S. Stryker, J. Knight, A. Thomas, M. van Regenmortel, D.M. Kemeny, D. Power, J. Rossert, N. Casadevall, The increased incidence of pure red cell aplasia with an Eprex formulation in uncoated rubber stopper syringes, Kidney Int. 67 (2005) 2346–2353. [4] M. Girard, J.C. Diez-Masa, A. Puerta, M. de Frutos, High resolution separation methods for the determination of intact human erythropoiesis stimulating agents: a review, Anal. Chim. Acta 713 (2012) 7–22. [5] M. Girard, I. Lacunza, J.C. Diez-Masa, M. de Frutos, Glycoprotein analysis by capillary electrophoresis, in: J.P. Landers (Ed.), Handbook of Capillary and Microchip Electrophoresis and Associated Microtechniques, 3rd ed., CRC Press, 2008, pp. 631–705. [6] Erythropoietin Concentrated Solution, European Pharmacopoeia 7.0, 01/2008:1316 (2010), pp. 1946–1950. [7] M. Fukuda, H. Sasaki, M.N. Fukuda, Erythropoietin metabolism and the influence of carbohydrate structure, Contrib. Nephrol. 76 (1989) 78–89. [8] M.N. Fukuda, H. Sasaki, L. Lopez, M. Fukuda, Survival of recombinant erythropoietin in the circulation: the role of carbohydrates, Blood 73 (1989) 84–89. [9] H. Zimmerman, D. Gerhard, L.A. Hothorn, T. Dingermann, An alternative to animal testing in the quality control of erythropoietin, Pharmeur. Bio Sci. Notes 2011 (2011) 66–80. [10] M. Schiestl, T. Stangler, C. Torella, T. Cepeljnik, H. Toll, R. Grau, Acceptable changes in quality attributes of glycosylated biopharmaceuticals, Nat. Biotechnol. 29 (2011) 310–312.
Contents lists available at SciVerse ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Short communication
Qualitative and quantitative assessment of marketed erythropoiesis-stimulating agents by capillary electrophoresis Sylvie Boucher, Anita Kane, Michel Girard ∗ Biologics and Genetic Therapies Directorate, Health Canada, 251- Sir Frederick Banting Driveway, Ottawa, ON K1A 0K9, Canada
a r t i c l e
i n f o
Article history: Received 22 May 2012 Received in revised form 16 August 2012 Accepted 19 August 2012 Available online 24 August 2012 Keywords: Erythropoietin Darbepoetin Formulation Capillary electrophoresis Erythropoiesis stimulating agent
a b s t r a c t Formulated erythropoiesis stimulating agents (ESAs) containing erythropoietin (EPO)-alpha, EPO-beta or darbepoetin-alpha were analyzed by capillary electrophoresis with a previously published method requiring no sample pre-treatment [1]. In this study, the method proved to be applicable to all formulations encountered, that is, in the presence of polysorbate 80, polysorbate 20 or human serum albumin as major excipients, thus extending the range of products that can be analyzed without pre-treatment. Method performance was evaluated and showed good linearity, range, precision and sensitivity. No significant matrix effects were observed for the various formulations. The ability of the method to resolve isoforms of each of the three active ingredients enabled comparison of the isoform distribution of finished products with that of the respective drug substance. In general, finished products and their corresponding drug substances showed similar isoform distribution and all were within manufacturer specifications. In addition, the content in active ingredient in the various dosage strengths was found to be in close agreement with the label claims with the exception of 2 out of 131 containers analyzed. Overall, this study demonstrated that the capillary zone electrophoresis method could be successfully applied to the analysis of most of the ESA products currently on the market in North America and Europe and that all products were found to have good batch-to-batch consistency. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved.
1. Introduction Erythropoietin (EPO) was among the first recombinant protein pharmaceuticals to reach the market at the end of the 1980s and it has been one of the most successful, with worldwide annual sales exceeding $10 billion US dollars in 2010 [2]. Early products sold in North America and Europe contained EPO-␣ (Eprex, Epogen, Procrit) or EPO- (Recormon/NeoRecormon). These two forms of human EPO have the same polypeptide sequence but differ in their glycosylation pattern. The product Eprex was originally formulated with human serum albumin (HSA) and, in 1998 the manufacturer changed the formulation to contain polysorbate (PS) 80 as a consequence of the European requirement to remove serum-derived components in pharmaceuticals [3]. A new ESA, darbepoetin-␣ (DPO-␣) (Aranesp), received marketing authorization in 2001. It is an analog of EPO and features a modified polypeptide sequence in addition to an increased number of glycosylation sites. In addition to these products, several EPO-␣ and EPO- biosimilars have received marketing authorization throughout the world since patents of innovator products have expired [4].
There are currently few physicochemical methods that enable the assessment of the active ingredient in biopharmaceutical final formulations. This is due in part to the small amounts of active ingredient present and to the large amounts of excipients required to enhance product stability, to prevent non-specific adsorption or to provide isotonic conditions suitable for injectables. Thus, final formulations often contain a varied mixture of excipients that may include small molecules such as salts, sugars or amino acids, in addition to larger molecules like HSA or PS. In this context, it has become desirable to develop widely applicable methods for the evaluation of products from multiple sources. Efforts from this group have been mainly concerned with the analysis of finished products in a direct manner, that is, without pre-treatment in order to minimize loss of product or generation of artifacts. Herein, we report a wideranging study of ESA products from North America and Europe where the assessment of the integrity of the active ingredient and the isoform distribution have been highlighted.
2. Materials and methods 2.1. Materials
∗ Corresponding author. Tel.: +1 613 952 0399; fax: +1 613 941 8933. E-mail address: [email protected] (M. Girard).
Formulated products and drug substances of four ESAs were generously donated by manufacturers from North America and Europe. They are identified as products A–D throughout this
0731-7085/$ – see front matter. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jpba.2012.08.021
208
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
Table 1 Description of ESA products analyzed during the study. Product
Dosage strength (per vial)
Lots
Excipient
Active ingredient
A
3000 IU 10,000 IU 20,000 IU
A1 A2 A3 A4 A5 A6 A7 A8 A9
Human serum albumin
EPO-␣
B
2000 IU 4000 IU 10,000 IU 40,000 IU
B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12
Polysorbate 80
EPO-␣
C
10,000 IU 20,000 IU 100,000 IU
C1 C2 C3 C4 C5 C6
Polysorbate 20
EPO-
Da
0.04 mg/mL 0.20 mg/mL 0.50 mg/mL
Da1 Da2 Da3 Da4 Da5 Da6 Da7 Da8 Da9
Human serum albumin
DPO-␣
Db
0.04 mg/mL 0.20 mg/mL 0.50 mg/mL
Db10 Db11 Db12 Db13 Db14 Db15 Db16 Db17 Db18
Polysorbate 80
DPO-␣
article. In brief, the study consisted in analyzing vials of multiple lots of three dosage strengths per product as shown in Table 1. Amine regenerator solution and eCap amine capillaries were from Beckman (Fullerton, CA). Sodium dihydrogen phosphate monohydrate and nickel chloride hexahydrate were from Sigma–Aldrich. All solvents were of HPLC quality. 2.2. Methods 2.2.1. Capillary zone electrophoresis setup Capillary zone electrophoresis (CZE) was carried out on a Beckman MDQ P/ACE capillary electrophoresis system (Beckman, Fullerton, CA) fitted with a variable wavelength UV detector and using Beckman 32 KaratTM Software for operation and data analysis. Separations were carried out on eCap amine capillaries, 50.2 cm × 50 m i.d. (effective length 40 cm) fitted in a cassette with an 800 m aperture. The background electrolyte (BGE) consisted of 200 mM NaH2 PO4 containing 1 mM NiCl2 (13.8 g NaH2 PO4 ·H2 O and 0.11885 g NiCl2 ·6H2 O in 500 mL ddH2 O). The pH was adjusted to 4.0 with dilute acetic acid. For analysis of product D, the phosphate concentration in the BGE was increased to 300 mM. 2.2.2. Capillary cleaning and conditioning A capillary cleaning procedure was applied to new capillaries or after running several injection sequences to re-establish proper working conditions. The following rinses were performed consecutively by applying pressure at 20 psi in the forward mode: methanol (5 min), 0.1 M HCl (5 min), doubly deionized water (ddH2 O) (5 min), 1.0 M NaOH (2 min), ddH2 O (5 min) and amine regenerator (5 min). A capillary conditioning procedure was performed after short storage time or upon reduced capillary performance (RSD for migration time greater than 10%). Consecutive rinses were carried out at 20 psi under pressure in the forward mode: 0.25 M NaOH (15 min), ddH2 O (5 min), amine regenerator (30 min) and BGE (15 min). 2.2.3. Separation conditions Electropherograms were monitored at 200 nm. The capillary temperature was maintained at 20 ◦ C and the sample area was at 10 ◦ C. In order to minimize the effect of ion depletion [1], largevolume BGE reservoirs of about 21 mL were used. Separations were carried out as sequences that included automatic pre-injection rinses in the following manner: 0.1 M NaOH (2 min, 20 psi, forward), amine regenerator (5 min, 20 psi, forward), BGE (10 min, 20 psi, forward), water dip (3 s), sample injection (8 s, 0.5 psi, forward), water
dip (3 s) and separate at 8.0 kV for 40 min by reverse polarity with a 0.5 min ramp.
2.2.4. Calibration curves Calibration curves for products A–D were prepared using the corresponding drug substance. At a minimum, six standard solutions encompassing the lowest and highest dosage strengths were prepared by serial dilution from a primary standard solution made from the drug substance: (1) drug substance A (DS A) – primary standard solution: 30 L DS A + 243 L ddH2 O (48,459 IU/mL); standards: 2423, 2908, 4846, 9692, 14,538, 19,384 and 24,230 IU/mL; (2) drug substance B (DS B) – primary standard solution: 200 L DS B + 722.6 L ddH2 O (80,000 IU/mL); standards: 2000, 4000, 6000, 8000, 10,000, 14,000, 20,000, 30,000, 40,000 and 50,000 IU/mL; (3) drug substance C (DS C) – primary standard solution: 759 L DS C + 741 L ddH2 O (187,097 IU/mL); standards: 4677, 9355, 18,710, 37,419, 56,129, 93,548 and 112,258 IU/mL; (4) drug substance D – primary standard solution: 754 L DS D + 749 L ddH2 O (1.00 mg/mL); standards: 0.50, 0.20, 0.10, 0.05, 0.02, 0.01 mg/mL. The linearity of the detector response over the range of concentrations was verified by plotting the corrected peak area percent (%CPA = % peak area/migration time) of the sum of all isoform peaks using isoform I2 as migration time marker.
2.2.5. Analysis of finished products Finished products in liquid form were analyzed directly by injecting the solution without any pre-treatment. Finished products in solid form were first reconstituted to the prescribed concentration with the accompanying diluent and the resulting solution was analyzed without any other treatment. A typical sequence consisted of replicate injections (n = 5) of two standard solutions and all vials of a single lot (e.g., a total of 25 injections for 3 vials). The two standard solutions included in the sequence corresponded to the respective drug substance (see Section 2.2.4) and represented one lower and one higher concentration than that of the vial label content. For content determination, peak areas were measured as the sum of all detected isoforms. Corrected peak areas were calculated by dividing peak areas by the migration time of isoform I2 as marker. The content was determined by interpolation from the plot of %CPA against concentration. For isoform distribution determination, peak areas of individual isoforms were measured. Plots of %CPA against isoform number were prepared.
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
A EP Method A. 4
0.006
2
0.005
0.005
5
0.004
0.004
0.003
0.003
0.002
0.002
0.001
6 8
1 0.001
7
0.000
0.000
-0.001
-0.001
3
Bietlot tlot-Girard Girard B. Bie B Method
0.012
AU
0.006
0.014
0.007
-0.002 -0.002 45.0 47.5 50.0 52.5 55.0 57.5 60.0 62.5 65.0 67.5 70.0 72.5 75.0 77.5 80.0 82.5 85.0
Absorbance @ 200 nm (AU)
Absorbance @ 200 nm (AU)
0.007
0.008
3
0.014 0.012
4
0.010
0.010
2
0.008
0.008
0.006
0.006
5
0.004
0.004
0.002
AU
0.008
209
0.002
0.000
0.000
6 7,8
1
-0.002
-0.002
-0.004
-0.004
-0.006
-0.006 18
19
20
21
Migration time (min)
22
23
24
25
26
27
28
Migration time (min)
Pk #
Bietlot-Girard Method % Corrected Peak Area
EP Method % Corrected Peak Area
1
1.36
1.76
2
16.95
18.57
3
29.23
30.18
4
27.61
27.6
5
17.13
16.06
6
5.49
4.55
7
2.23 (7+8)
1.28
8
-
<0.3
Totals
100
100
Fig. 1. Electropherograms of the European pharmacopoeia biological reference preparation for erythropoietin by (A) the EP monograph method and (B) the method of Bietlot and Girard [1].
3. Results and discussion Samples of finished ESA products A–D, corresponding to a wide range of available dosage strengths, along with their corresponding drug substance (DS A–D) were obtained from sources in Europe and North America (Table 1). Products A and B contained EPO␣ formulated with HSA and polysorbate 80 (PS80), respectively. Product C contained EPO- formulated with polysorbate 20 (PS20). Products D contained DPO-␣ formulated with HSA (product Da) or PS80 (product Db). The primary objectives of this study being the evaluation of the integrity of the active ingredients in terms of isoform distribution and content, capillary electrophoresis was deemed suitable to achieve these objectives as this technique has been frequently used for the separation of intact EPO isoforms [4,5]. 3.1. Capillary zone electrophoresis method The European Pharmacopoeia (EP) monograph for EPO includes an identification test based on CZE [6] that provides baseline separation of 8 isoforms, numbered I1–I8 from the most acidic to the most basic, for the EP biological reference preparation (BRP) for EPO (Fig. 1A). However, the method could not be applied to the direct analysis of formulated products as removal of excipients is required prior to analysis. Furthermore, while efficient removal of small molecule excipients is possible, attempts at removing HSA led to low EPO recovery and apparent loss of resolution (data not shown). For these reasons the use of this method was abandoned. We previously reported a CZE method for the detection and quantification of EPO-␣ and EPO- in HSA-containing formulations in a direct fashion, that is, without the need to remove excipients
[1]. The relatively high ionic strength of the BGE, that is, 200 mM phosphate, provided the necessary conditions for the analysis of final formulations. Analysis of the EP BRP by this method enabled separation of five major isoforms, I2–I6 and 3 minor isoforms I1, I7–I8 (Fig. 1B). As expected when using reversed-polarity, the migration order was reversed compared to that of the EP method, with the most acidic isoform, I1 migrating first and the most basic isoform, I8 migrating last. While only partial isoform resolution was achieved when compared to the EP method, it was deemed sufficient for comparative purposes as demonstrated by the similar relative isoform levels measured by the two methods (see Table in Fig. 1). However, its applicability to the analysis of all of the ESAs and their various formulations, particularly for polysorbate formulations and for DPO-␣ needed to be demonstrated. Typical electropherograms of drug substances, DS A–C are presented in Fig. 2A–C. The EPO-␣ drug substances, DS A and DS B (Fig. 2A and B, respectively) were each separated into 6 isoforms that included 2 minor isoforms, I1 and I6, and 4 major isoforms, I2I5. The EPO- drug substance, DS C (Fig. 2C) was separated into 8 isoforms consisting of the 3 minor isoforms, I1, I7 and I8, and the 5 major isoforms, I2–I6. The analysis of DPO-␣ (DS D) showed a profile indicative of the presence of multiple isoforms that migrated more quickly than those from either EPO-␣ or EPO-. The faster migration was expected and could be easily explained on the basis of the increased number of terminal sialic acids resulting from the additional asparagine-linked glycans in DPO-␣. While the isoform separation was not optimal under the standard separation conditions it could be improved by increasing the BGE phosphate concentration to 300 mM. This resulted in the partial separation of 5 isoforms, numbered I1–I5 from the most acidic to the most basic,
210
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213 0.05
A
0.03
0.02
0.02
0.01
0.01
AU
HSA 0.03
Absorbance @ 200 nm (AU)
0.04
EPO
0.00
0.00
-0.01
0.007
0.006
0.006
B
0.005
0.005
0.004
0.004
0.003
0.003
0.002
0.002
0.001
0.001
0.000
0.000
-0.001
-0.001
-0.002
-0.002
-0.003
-0.01 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
-0.003 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0 42.5 45.0 47.5 50.0
Migration time (min) 0.016
0.016
0.014
0.014
C
0.07
0.010
0.008
0.008
0.006
0.006
0.004
0.004
0.002
0.002
0.000
0.000
-0.002
-0.002
-0.004
-0.004
16
18
20
22
24
26
28
30
32
34
36
38
40
Migration time (min)
Absorbance @ 200 nm (AU)
0.012
0.010
0.07
D
AU
0.012
Absorbance @ 200 nm (AU)
Migration time (min)
0.06
0.06
0.05
0.05
0.04
0.04
0.03
0.03
0.02
0.02
0.01
0.01
2.5
5.0
AU
Absorbance @ 200 nm (AU)
0.04
0.007
AU
0.05
7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.0 37.5 40.0
Migration time (min)
Fig. 2. Typical electropherograms of finished products A–D (red and blue traces) with their respective drug substance (DS A–D, black traces): (A) product A4 and DS A; (B) product B7 and DS B; (C) product C2 and DS C and (D) products Da7, Db16 and DS D. Insets: enlarged view of separated isoforms labeled from the most acidic to the most basic. Separation conditions: eCap Amine capillary; 200 mM sodium phosphate, 1 mM nickel chloride, pH 4.0; voltage: −8 kV (for samples D, BGE was 300 mM phosphate, 1 mM nickel chloride, pH 4.0). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
with 4 major isoforms, I1–I4, and one minor isoform, I5 (Fig. 2D). The 300 mM phosphate buffer concentration was chosen to perform all subsequent analyses of DS D and product D. As mentioned above, it was essential to verify that all types of formulations would be amenable to the detection of the active ingredient and to the separation of isoforms. Typical electropherograms of finished products A–D are shown in Fig. 2 alongside their respective drug substance. In all cases no interference from excipients were observed and the isoform separation and distribution was similar to that of the corresponding drug substance. These results provided evidence that the CZE method could be successfully extended to the direct analysis of EPO polysorbate formulations. This was also the case for products Da and Db, which were formulated with HSA and PS80, respectively. Overall, these results showed that the range of products that can be directly analyzed using this CZE method covered a large part of the ESA products currently on the market in North America and Europe. 3.2. Method performance for drug substances The performance of the method in terms of matrix effect, linearity, range, precision and sensitivity was evaluated for each of the drug substances, DS A–D. Matrix effect on the detector response was carried out using simulated formulation solutions obtained by mixing the drug substance and either HSA, PS80 or PS20 at concentrations close to those in finished products. In all cases, %CPA
for simulated solutions were within 2–3% of their corresponding drug substance solutions prepared at the same concentration, an indication that no substantial matrix effect was present (data not shown). Table 2 presents performance data for each of the four drug substances. Linearity of the detector response was assessed for concentration ranges that generally encompassed at least 80–120% of all of the available dosages to be examined. Plots of the %CPA for the sum of all isoforms, using isoform I2 as peak marker, against the concentration showed linear relationships with coefficients of determination, R2 , better than 0.996 over the entire range. Good migration time and %CPA repeatability were obtained on replicate injections of standard solutions at three concentrations (low, medium and high) within the same sequence. Typically, migration time repeatability was found to be very good for all four drug substances with %RSD values below 1%. Good repeatability was also obtained for %CPA with %RSD values below 7% for the average %RSD values at three concentrations. Intermediate precision from replicate injections of standards at the same three concentrations in separate sequences showed, as anticipated, greater variability for migration time, with %RSD values of less than 5%. Similarly, increased variability was found for %CPA, with average %RSD for three concentrations between 4 and 11%. These values were considered acceptable for the purpose of the study. Limits of detection (LODs) were set for the lowest concentration giving a signal-to-noise ratio greater than 3. For DS A and B, LODs
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
211
Table 2 Method performance for drug substances A–D. Drug substance
A B C D a b c d
Repeatability
Intermediate precision
Limit of detectiond
MTa (Avg %RSDb )
CPAa (Avg %RSDb )
MT (Avg %RSDc )
CPA (Avg %RSDc )
0.11 0.62 0.18 0.17
5.54 4.21 6.47 6.53
1.67 4.62 1.05 1.82
10.64 10.36 4.26 4.56
2000 IU/mL 2000 IU/mL 2500 IU/mL 0.010 mg/mL
Limit of quantificationd
2500 IU/mL 3000 IU/mL 4000 IU/mL 0.020 mg/mL
Linearity R2
Slope
Y intercept
0.9999 0.9963 0.9991 1.0000
0.3216 0.6164 0.613 86,653.2
−211.3 −1367.2 −2359.0 −342.1
MT, migration time; CPA, corrected peak area. Avg %RSD, average %RSD from replicate injections (n = 5) of standards at three concentrations in the same sequence (see text). Avg %RSD, average %RSD from replicate injections (n = 5) of standards at three concentrations in different sequences (see text). Limit of detection was set for the lowest concentration giving a S/N ≥ 3 (n = 5); limit of quantification was for the lowest concentration giving a RSD < 10% (n = 5).
were found at 2000 IU/mL while for DS C a value of 2500 IU/mL was obtained. The higher LOD value for product C could be explained on the fact that there was a greater number of minor isoforms (I1, I7 and I8) contributing to the total peak area and that these were more difficult to detect at very low concentrations. The LOD for DS D was found to be at 0.01 mg/mL. The limits of quantification (LOQs) were determined for the lowest concentration giving %RSD values lower than 10% for replicate injections (n = 5). Again, the LOQ for DS C was found to be at a higher concentration than for both DS A and B for the same reasons. The LOQ for DS-D was found to be 0.02 mg/mL. Overall, results from the performance assessment demonstrated that the CZE method met the usual criteria of acceptability for the qualitative and quantitative analysis of the four types of ESA products under investigation. In addition, these results showed that all of the finished products available, with the exception of lots B1–B3 of product B, where the label dosage was 2000 IU/mL, could be reasonably expected to be amenable to quantification since their label content was at or greater than the LOQ of the respective drug substance.
product, a comparison of replicate injections of the drug substance (panel A), replicate injections of a given dosage strength of the finished product (panel B), different vials of a given dosage strength of the finished product (panel C) and different lots of a given dosage strength of the finished product (panel D) is presented. Products A and B (Figs. S1 and S2, respectively) showed a high degree of consistency between drug substances and finished products, with only minor variations (less than 3%) in the relative amounts of the major isoforms I3–I5. For product C (Fig. S3), slightly larger variations were apparent between the drug substance and the finished product, especially with regards to isoforms 5 and 7 + 8. This was likely a consequence of the difficulty in obtaining highly reproducible peak integration for these isoforms since they were not always well resolved. The two types of formulation for product D (Figs. S4 and S5) showed very similar isoform distributions as well as a high degree of consistency between finished products and the drug substance. Despite the minor variations noted above all products analyzed were within the manufacturer specifications with regards to the relative isoform content.
3.3. Analysis of finished products As indicated above, the assessment of the integrity of the active ingredient in finished products used a two-prong approach, namely, the verification of the isoform distribution and the estimation of the content in active ingredient. All of the products analyzed were within their expiry date. 3.3.1. Isoform distribution The relationship between in vivo biological activity and the type of N-glycans and their degree of sialylation in EPO has been described [7,8]. Variable terminal sialic acid content of the glycans is responsible for the overall charge distribution of the molecule resulting in the presence of isoforms. Thus, knowledge of the isoform distribution is an important issue for ESAs. Furthermore, a quantitative method which assessed glycan characteristics such as isoform percentages and total sialic acid content, as well as glycan branching and branch repeats, to determine such bioactivity, was found to be reducible to the isoform profile and total sialic acid content [9]. In addition, changes in the isoform distribution may be acceptable as was recently demonstrated between batches of the same ESA product [10]. In that case, isoform distribution changes purportedly coincided with a process modification and may be reflective of this modification. In particular, very good batch-tobatch consistencies were observed in each set of batches prior to and after the isoform distribution change, which was deemed to be within acceptable limits. In this study, the isoform distribution was evaluated by determining the relative amount of each isoform in a given product. Graphical representations of results are shown in Supplementary Figs. S1–S5 for products A, B, C, Da and Db, respectively. For each
3.3.2. Content in active ingredient In view of the absence of matrix effects and that linearity was achieved over the entire range of concentrations, quantification of ESA content in finished products was determined by interpolation from a two-point standard calibration plot. In most cases, the two standards used in the calibration plots represented concentrations below and above the expected finished product concentration stated on the container label. Fig. 3 shows a graphical representation of the overall results from the assessment of the content in active ingredient. The content was determined by interpolation (and in a few instances, extrapolation) on plots of the %CPA against the concentration prepared from two standard solutions injected during the same sequence. In all cases, the sum of the corrected peak area of all detected isoforms was used and isoform I2 was used as peak marker for the calculation of the corrected peak area. All vials from a single lot were analyzed during the same sequence and the average vial content was used for plotting in Fig. 3. Thus, each rectangular bar corresponds to the average vial content from a single lot. The round bar represents the average content of all lots. The x axis represents the label dosage strength. It must be noted here that differences in content compared to the label dosage strengths were expected taking into account that the content was determined in relation to drug substance samples that were not the same as those used in the formulated products. As such, differences may result due to the fact that the specific activity of the drug substances may vary and that manufacturers may also overfill in order to compensate for losses from adsorption. In addition, adsorption is a time-dependent process and the different lots of different products varied in time since filling and time to expiry date.
212
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
Fig. 3. Graphical representation of the content in active ingredient in finished products determined by CZE. In each panel, the measured content of each lot (square bar) of the various dosage strengths are plotted. A square bar represents the average lot content measured from three vials. A round bar represents the average of the lots shown on its left. For the dash blue square bar, see text. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Overall results showed that the measured content in active ingredient for the vast majority of products was close to the expected value, generally within 15% of the label claim. This was particularly evident for products A, B and D. For example, the average content found for the three 3000 IU vials of product A (Fig. 3 – product A) was 2875 IU, which corresponded to 95.0% of the label claim. The RSD values were also found to be good, with values within 10% for the average of three lots. This was an indication of the satisfactory lot-to-lot consistency of the various products. Results for product C require specific comments with regards to the measured content. For the 10,000 IU and 20,000 IU dosage strengths,
only 2 lots and 1 lot, respectively, were available for analysis and, as such, results may not be viewed in the same manner as for the other products and are presented here for guidance purposes only. For the 100,000 IU dosage strength, a single vial of lot C6 was available. The electropherogram showed a wide peak migrating at the expected time for EPO but without distinct isoform peaks (Fig. 4). In addition, the measured content was lower than that of the other two lots as shown by the hatched bar in Fig. 3 – product C. These results suggested that the integrity of EPO in these samples may have been compromised to some extent; however, further investigation on the possible cause could not be carried out at the time
0.040 0.035 0.030
PS20
0.025
AU
0.020
EPO
0.015 0.010 0.005 0.000 -0.005
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Minutes Fig. 4. Electropherograms of vials of product C at 100,000 IU with (red trace) and without (blue trace) resolved isoform peaks (see text for details). Separation conditions as in Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
S. Boucher et al. / Journal of Pharmaceutical and Biomedical Analysis 71 (2012) 207–213
of these experiments. A similar situation occurred for one of the 0.5 mg/mL lot of product Da where no active ingredient could be detected in any of the three vials. Again, it was not possible to investigate the possible cause at the time of these experiments. As such, the average content of 0.587 mg/mL was based on the average of two lots. Overall, these results showed that marketed products generally showed good lot-to-lot consistency with regards to isoform distribution and active ingredient content as measured by capillary electrophoresis. 4. Conclusion Results in this study have demonstrated that capillary zone electrophoresis was adequate for the monitoring of the active ingredient in all of the ESA products analyzed no matter the type of formulation encountered. This expands the types of products that can be analyzed directly, that is, without pretreatment. This method also enabled to examine major issues with regards to finished products, that is, the integrity of the active ingredient and its content. Products analyzed showed high level of consistency between vials of the same lot and between lots. Acknowledgments We are grateful to Drs. Mary Alice Hefford and Jeremy Kunkel (Health Canada) for fruitful discussions during the preparation of this article.
213
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jpba.2012.08.021. References [1] H.P. Bietlot, M. Girard, Analysis of recombinant human erythropoietin in drug formulations by high performance capillary electrophoresis, J. Chromatogr. A 759 (1997) 177–184. [2] G. Walsh, Biopharmaceutical benchmarks 2010, Nat. Biotechnol. 28 (2010) 917–924. [3] K. Boven, S. Stryker, J. Knight, A. Thomas, M. van Regenmortel, D.M. Kemeny, D. Power, J. Rossert, N. Casadevall, The increased incidence of pure red cell aplasia with an Eprex formulation in uncoated rubber stopper syringes, Kidney Int. 67 (2005) 2346–2353. [4] M. Girard, J.C. Diez-Masa, A. Puerta, M. de Frutos, High resolution separation methods for the determination of intact human erythropoiesis stimulating agents: a review, Anal. Chim. Acta 713 (2012) 7–22. [5] M. Girard, I. Lacunza, J.C. Diez-Masa, M. de Frutos, Glycoprotein analysis by capillary electrophoresis, in: J.P. Landers (Ed.), Handbook of Capillary and Microchip Electrophoresis and Associated Microtechniques, 3rd ed., CRC Press, 2008, pp. 631–705. [6] Erythropoietin Concentrated Solution, European Pharmacopoeia 7.0, 01/2008:1316 (2010), pp. 1946–1950. [7] M. Fukuda, H. Sasaki, M.N. Fukuda, Erythropoietin metabolism and the influence of carbohydrate structure, Contrib. Nephrol. 76 (1989) 78–89. [8] M.N. Fukuda, H. Sasaki, L. Lopez, M. Fukuda, Survival of recombinant erythropoietin in the circulation: the role of carbohydrates, Blood 73 (1989) 84–89. [9] H. Zimmerman, D. Gerhard, L.A. Hothorn, T. Dingermann, An alternative to animal testing in the quality control of erythropoietin, Pharmeur. Bio Sci. Notes 2011 (2011) 66–80. [10] M. Schiestl, T. Stangler, C. Torella, T. Cepeljnik, H. Toll, R. Grau, Acceptable changes in quality attributes of glycosylated biopharmaceuticals, Nat. Biotechnol. 29 (2011) 310–312.