- Email: [email protected]
Simple and precise detection of lipid compounds present within liposomal formulations using a charged aerosol detector
Journal of Chromatography A, 1216 (2009) 781–786
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Simple and precise detection of lipid compounds present within liposomal formulations using a charged aerosol detector Carina Schönherr a,b , Sounia Touchene a , Gaston Wilser a , Regine Peschka-Süss b , Giancarlo Francese a,∗ a b
Novartis Pharma AG, Development, Lichtstrasse 35, CH-4058 Basel, Switzerland Department of Pharmaceutical Technology and Biopharmacy, University of Freiburg, Sonnenstrasse 5, D-79104 Freiburg, Germany
a r t i c l e
i n f o
Article history: Received 29 August 2008 Received in revised form 17 November 2008 Accepted 20 November 2008 Available online 3 December 2008 Keywords: Liposomes HPLC Lipid detection Corona aerosol detector (CAD) Phosphatidylcholine
a b s t r a c t In recent decades the use of liposomal preparations as drug delivery systems has become very attractive in pharmaceutical development. Therefore, thorough characterization and quantification of the lipids which form liposomes is wished from both investigators and regulatory authorities when the application in humans is being considered. In this study a new HPLC method for the detection of lipids in liposomal formulations was established using corona charged aerosol detection (CAD) which has the advantage to be independent of the chemical properties of the analytes. The superiority of this method over UV detection was demonstrated. Compared to UV detection no absorption effects of the organic solvent in the mobile phase interfering with the lipid signals were observed with CAD. CAD showed good linearity (R2 > 0.990) for all liposomal compounds. The acceptance criteria for precision including repeatability were met. The average recovery for each of the excipients of the liposomal formulation was in the range of 90.0–110%. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Currently, the use of liposomes as drug delivery vehicles is of growing interest in pharmaceutical drug development. Liposomes are vesicles in the nanometer size range composed of lipid bilayers, primarily phospholipids, which resemble natural membranes [1]. They are able to entrap water-soluble substances inside their aqueous core [2] and to incorporate water-insoluble substances into the lipid bilayer [3]. Due to this versatility and the ability of liposomes to improve the pharmacokinetics and pharmacodynamics of the associated drugs [2,4], pharmaceutical companies are increasingly developing liposomes as drug carrier systems for various drug substances, especially those which are highly toxic and or highly insoluble. The use of several liposomal formulations of anticancer drugs in tumor therapy, such as doxorubicin (Doxil® /Caelyx® ) and daunorubicin (DaunoXome® ), has already been established. Other liposomal drug formulations are currently being studied in clinical phase trials [5,6]. The authorities demand full qualitative and quantitative characterization, not just of the active substance itself, but also of the excipients within a drug formulation. The FDA’s draft guide for industry on liposome drug products also recommends to have an accurate lipid composition characterization (i.e., a set percent-
∗ Corresponding author. Tel.: +41 61 324 1908; fax: +41 61 324 6983. E-mail address: [email protected] (G. Francese). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.11.080
age of each lipid). The lipid components should be assayed, stress tested, as well as profiling of the degradation products related to the lipids [7,8]. In this context high-performance liquid chromatography (HPLC) has been well established as a fast and precise method to analyze both drug substances and excipients. However, for lipids which are the predominant component in liposomal formulations, conventional UV detection is often not adequate and limited to chromophores while other methods, which have been previously described for the detection of lipids, such as flame ionization detection (FID) [9] or evaporative light-scattering detection (ELSD) [10], have significant limitations in precision, sensitivity and dynamic range as demonstrated previously by Loughlin et al. [11–13]. In 2005, ESA (Chelmsford, MA, USA) developed an HPLC detector that allows the universal detection of various substances due to the independence of the analysis from both chemical structures and properties of the substances to be detected [14,15]. A so-called corona charged aerosol detection (CAD) system is compatible with all ordinary HPLC devices. After HPLC separation using a suitable column, the eluent is nebulized with nitrogen and then dried to remove the mobile phase, producing analyte particles. Particles which are too big are separated by impaction and conducted to the waste. A secondary stream of nitrogen is charged by a high-voltage platinum corona wire. The positive charge is transferred to the analyte particles, which then pass a negatively charged ion-trap where high-mobility particles are removed. Arriving at a collector the analyte particles transfer their charge, which is measured by a sensitive
782
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
electrometer. The resulting signal is directly proportional to the quantity of the analyte present in the sample. This innovative technology offers many advantages compared to conventional lipid detection methods. These include high sensitivity with a low limit of detection in the nanogram range, consistent response which is independent of chemical properties, excellent reproducibility with less than 2% relative standard deviation (RSD) and a dynamic range of up to four orders of magnitude allowing the measurement of quantities in the nanogram to microgram range. Furthermore, this detection method is applicable to a wide variety of applications [15]. In this study, an HPLC–CAD method was established for the analysis of conventional lipid components of liposomal formulations. Analysis by CAD was compared to analysis by conventional UV detection and validation parameters such as linearity, precision, accuracy and repeatability were investigated. 2. Experimental
First, 5 l of the liposome dispersion were diluted with 2 ml of phosphate buffer pH 7.4 and then measured at 20 ◦ C in a 90◦ angle. The average hydrodynamic diameter (z-average) was calculated out of 10 sub-runs. All liposome preparations used for CAD analysis showed a zaverage of approx. 100 nm with a polydispersity index of less than 0.1 (reflecting the narrowness of particle size distribution). 2.4. HPLC 2.4.1. HPLC instrumentation A Thermo Separation Products (TSP) HPLC device (Allschwil, Switzerland) was used with a P4000 pump and an AS 3000 sampler. For detection, a UV 6000 LP detector from TSP (with a linear dynamic range <5% deviation up to 2.0 AU) and a Corona Plus detector (with a dynamic range from 10 ng to 100 g [13]) from ESA was used. The evaluation-software used was Chromeleon (version 6.50) which was obtained from Dionex-Softron GmbH (Olten, Switzerland).
2.1. Chemicals and reagents Methanol, ethanol, acetonitrile and trifluoroacetic acid were of HPLC grade and purchased from Merck (Darmstadt, Germany). Water was purified on a Milli-Q system obtained from Millipore (Zug, Switzerland). For the preparation of liposomes, unsaturated soybean phosphatidylcholine (Lipoid S 100) was purchased from Lipoid (Ludwigshafen, Germany), N-carbonyl-methoxy(polyethyleneglycol 2000)-1,2-distearoyl-sn-glycero-3-phosphoethanolamine sodium salt (mPEG-2000-DSPE) was purchased from Genzyme Pharmaceuticals (Liestal, Switzerland), cholesterol was purchased from Fluka (Buchs, Switzerland) and d,l-␣-tocopherol was purchased from Novartis (Stein, Switzerland). For the hydration buffer potassium phosphate monobasic and disodium hydrogenphosphate dihydrate were obtained from Fluka (Buchs, Switzerland). 2.2. Preparation of PEG-liposomes The lipid composition used for all liposome preparations was as follows: unsaturated soybean phosphatidylcholine (Lipoid S 100), mPEG-2000-DSPE, cholesterol and ␣-tocopherol in a molar ratio of 100:7:20:0.9. All liposome batches were produced according to the thin film hydration method [16]. Briefly, the excipients were dissolved in ethanol and the solvent was evaporated using a rotavapor (Rotavap R 215, Büchi, Switzerland) for 1 h, slowly decreasing the pressure to 30 mbar and were further dried for 2 h to remove the solvent completely. The lipid film was then hydrated with a phosphate buffer pH 7.4 (potassium phosphate monobasic 20 mM, disodium hydrogenphosphate dihydrate 20 mM) under fast magnetic stirring in a 45 ◦ C hot water bath for 15 min to yield a liposome dispersion of approx. 80 mM. The liposome dispersion was extruded through a LIPEX extruder (Northern Lipids, Burnaby, Canada) first using a polycarbonate membrane with 100 nm pores three times (Millipore) then using a polycarbonate membrane with 50 nm pores six times (Millipore). 2.3. Particle size measurement using photon correlation spectroscopy (PCS) The average hydrodynamic diameter of the liposomes and the particle size distribution was analyzed by photon correlation spectroscopy using a Zetasizer 3000HS from Malvern, Worcestershire, UK.
2.4.2. HPLC conditions A XBridge C18 3.5 m column (150 mm × 3 mm) from Waters, Milford, MA was kept at 35 ◦ C with a column thermostat Jetstream 2 Plus from TSP (Allschwil, Switzerland) for all of the HPLC–CAD experiments performed, while the flow rate was 0.5 ml/min for the mobile phases (mobile phase A: acetonitrile/Millipore water (90/10, v/v) + 0.05% trifluoroacetic acid (TFA) and mobile phase B: methanol + 0.05% TFA). The injected volume was 10 l and the cooling temperature for the samples was 6 ◦ C, unless stated otherwise. 2.4.3. HPLC gradient HPLC was started at time point zero using 60% of mobile phase A and 40% of mobile phase B. After 25 min the mobile phase B was increased to 100%. Those conditions were maintained for another 10 min. Each HPLC run was stopped after 35 min. 2.4.4. Dilution of single lipid compounds For all excipients (phosphatidylcholine, mPEG-2000-DSPE, cholesterol and ␣-tocopherol) stock solutions were prepared with methanol/Millipore water (90/10, v/v) and then diluted with methanol/Millipore water (90/10, v/v) to reach the concentrations listed in Tables 1 and 2. 2.4.5. Dilution of liposome samples All liposomal samples were diluted 1:1000 with methanol/ Millipore water (90/10, v/v) which resulted in a final lipid concentration of approx. 80 M. This brought the lipid concentration into an appropriate range for CAD. 2.5. Identification of the phosphatidylcholine peaks by HPLC–MS Unsaturated soy phosphatidylcholine (PC) is a mixture of lipids with different fatty acid moieties. For Lipoid S 100 the chromatogram consists of five different peaks (Fig. 1). In addition an analysis with HPLC–mass spectrometry (MS) was performed with a Hewlett–Packard 1100 MSD device, to determine the main peaks corresponding to the different fatty acid compositions of phosphatidylcholine. The conditions used for HPLC were the same as those used for the HPLC–CAD method except for the sample temperature, which remained at room temperature. The detection conditions were the following: the gas temperature was set at 350 ◦ C, the volume of the drying gas was 12 l/min with a nebulizing pressure of 60 psig. The ionization mode was
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
783
Table 1 Correlation of the concentration of liposomal components to HPLC–CAD peak area. The correlation coefficient is taken as a parameter for linearity, the residual standard deviation (RSD) (≤3.5%, relative to the highest response) and the relative y-intercept (≤10.0%, relative to the highest response) were used as parameters for method precision (n = 2). Liposomal component
Concentration (g/ml)
Average area (mV s)
Residual standard deviation (RSD) (%)a
Correlation coefficient
y-Intercept (%)a
LODb (g/ml)
LOQc (g/ml)
Phosphatidylcholine (Main peak 1)
80 160 320 480 640
51.6 94.0 164.3 224.7 276.4
3.12
R2 = 0.9934
10.0
0.42
1.28
Phosphatidylcholine (Main peak 2)
80 160 320 480 640
48.5 91.3 163.6 229.1 289.5
2.15
R2 = 0.9970
7.1
0.28
0.86
mPEG-2000-DSPE
15 30 60 120 180
32.6 81.8 163. 298.1 418.5
3.15
R2 = 0.9948
2.7
0.33
0.99
Cholesterol
6 12 24 36 48 60
10.0 20.5 41.2 58.0 73.6 89.5
2.42
R2 = 0.9960
3.8
0.34
1.03
␣-Tocopherol
6 12 24 36 48 60
8.3 19.0 35.2 54.1 70.2 85.1
1.67
R2 = 0.9981
1.4
0.23
0.70
a b c
Calculated relative to the highest response. LOD = 3.3/S (slope S and as standard deviation of the y-intercept calculated from the calibration curve). LOQ = 10/S (slope S and as standard deviation of the y-intercept calculated from the calibration curve).
Table 2 The recovery rate of the liposomal lipid concentration is taken as a parameter for the accuracy of the HPLC–CAD detection method (n = 6). Liposomal component
Theoretical conc. (g/ml)
Recovered average conc. (g/ml)
Recovery (%)
Cholesterol ␣-Tocopherol Phosphatidylcholine (Main peak 1) Phosphatidylcholine (Main peak 2) mPEG-2000-DSPE
49.8 29.4 478.0 478.0 137.0
51.0 29.5 485.6 489.3 134.0
102.5 100.3 101.6 102.4 97.8
Fig. 1. HPLC–CAD chromatograms of every single liposomal compound (Toc: ␣-tocopherol; PC: phosphatidylcholine; Chol: cholesterol; PEG: mPEG-2000-DSPE) and a mixture of all excipients (Mix). The concentration of each compound is 50 g/ml.
784
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
atmospheric pressure ionization (API)–electrospray ionization (ESI) with positive polarity [17].
Table 3 Repeatability of the detection of single liposome compounds in solution, considering the mean retention time and peak area (n = 5).
3. Results and discussion
Liposomal component
3.1. Linearity The linearity of analysis using CAD was determined by analyzing different concentrations of each liposomal compound (Table 1). All solutions were injected twice for HPLC–CAD analysis. From MS results the main peak (Main peak 1) for phosphatidylcholine (758.6 g/mol [M+H]+ ) was identified having palmitic acid and linoleic acid as fatty acid chains (PC 16:0/18:2[9Z,12Z]) (retention time approx. 17.5 min). Another peak with a retention time of approx. 12 min, identified by MS as PC (18:1[9Z]/18:3[6Z, 9Z, 12Z]) (782.6 g/mol [M+H]+ ), was also examined in order to prove the linearity of the measurements (Main peak 2). The relation of peak area to concentration of excipients resulted in a good linearity with high correlation coefficients (>0.990) for all liposomal compounds (Table 1). 3.2. Precision As a criterion for determining the precision of the method, the residual standard deviation and y-intercept values were evaluated from linearity measurements taken relative to the highest response. It was requested that the residual standard deviation be lower than 3.5%, and that the y-intercept be lower than 10.0%. Therefore, the acceptance criteria were met (Table 1). 3.3. Accuracy In order to determine the accuracy of the measurements, a stock solution for each excipient was diluted with methanol/Millipore water (90/10, v/v) to obtain the theoretical concentration values listed in Table 2. Six diluted solutions were prepared equally and injected individually. The resulting average concentrations which were measured were then evaluated, in addition to an average recovery percentage which was used as quality parameters for determining the accuracy of the method. The average recovery was requested by the FDA’s guide for liposome drug products [7] according to the International Conference on Harmonisation (ICH) [18] to
Cholesterol ␣-Tocopherol Phosphatidylcholine (Main peak 1) Phosphatidylcholine (Main peak 2) mPEG-2000-DSPE
Retention time (min)
Area (mV min)
Mean (min)
RSD (%)
Mean (mV min)
RSD (%)
17.0 16.1 20.2
0.015 0.029 0.043
73.0 43.2 210.5
0.801 0.576 3.745
14.2
0.039
206.6
1.589
29.2
0.021
308.4
3.322
be in the range of 90.0–110.0%. This criterion was fulfilled for each of the excipients (Table 2). 3.4. Repeatability for each single excipient As evaluation-data for the repeatability of the method, both the retention time and the area under the peaks were evaluated (Table 3). The relative standard deviation (RSD) was evaluated as a criterion for the repeatability of the measurements. For the mean retention time all values of RSD were lower than 0.05%. Considering the peak area, the RSD was higher due to inexactness in the integration. 3.5. Repeatability for the measurement of liposome solution (mixture of lipids) The self-assembled liposomal structure only exists in an aqueous environment. After solution in an organic solvent, the liposomal configuration is destroyed. All lipids are separately dissolved in methanol/Millipore water (90/10, v/v). Hence, this lipid solution from liposomal dispersion (dilution 1:1000, refer to 2.3.4) was investigated by CAD analysis to ensure that there were no changes in retention time due to possible interactions between the single lipids. The diluted liposomal solution was injected six times. For retention time all of the measured values of RSD were lower than 0.08%, demonstrating high repeatability in the determination of lipid compounds by HPLC–CAD (Table 4). When peak area was examined the RSD was again higher due to inexactness in the integration.
Fig. 2. HPLC–CAD detection of PEG-liposomes. In total seven peaks from lipid compounds were detected: five peaks from phosphatidylcholine (conc. 50 g/ml, PC1 : 9.2 min; PC2 : 12.5 min; PC3 : 18.1 min; PC4 : 23.8 min; PC5 : 24.3 min) and one peak each for cholesterol (conc. 5 g/ml, Chol: 16.9 min) and mPEG-2000-DSPE (conc. 13 g/ml, PEG: 28.5 min). The peaks in the beginning of the chromatogram are from phosphate buffer (buffer). The concentration of the antioxidant ␣-tocopherol in the liposomal sample (conc. 0.25 g/ml) is at the limit of detection and therefore not visible.
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
785
Fig. 3. Comparison of CAD and UV detection for all liposomal compounds. (A) Cholesterol (conc. 30 g/ml), (B) ␣-tocopherol (conc. 18 g/ml), (C) phosphatidylcholine (Lipoid S 100) (conc. 160 g/ml), (D) mPEG-2000-DSPE (conc. 80 g/ml).
It has to be mentioned that the amount of the antioxidant ␣tocopherol in the liposomal formulation was below the limit of detection (Fig. 2). 3.6. Comparison CAD to UV detection A direct comparison between UV detection and the CAD for the same HPLC analysis was done to emphasize the need for a detection method for lipids which is more precise and sensitive than UV. Higher and sharper peaks were obtained using CAD compared to UV detection (Fig. 3). Additionally, the baseline increased with time for UV detection using the HPLC gradient elution. This was due to the increasing amount of methanol in the mobile phase during gradient elution. The absorption of the organic solvent overlapped with the lipid signals Therefore, some essential lipid peaks could not be precisely detected by UV. 3.7. Helpful observations made during method development During method development we found that the addition of TFA to the mobile phase was necessary to optimize the peak sharpness for mPEG-2000-DSPE. Another parameter that was found to
Table 4 Repeatability for the excipient mixture used in liposome dispersion, considering the mean retention time and peak area (n = 6). The amount of ␣-tocopherol is below the limit of detection. Liposomal component
Cholesterol ␣-Tocopherol Phosphatidylcholine (Main peak 1) Phosphatidylcholine (Main peak 2) mPEG-2000-DSPE
Retention time (min)
Area (mV min)
Mean (min)
RSD (%)
Mean (mV min)
RSD (%)
16.9 n.a. 18.3
0.029 n.a. 0.065
6.1 n.a. 19.6
0.575 n.a. 0.462
12.5
0.026
24.9
1.568
28.4
0.075
19.4
0.785
influence the sharpness of the peak for mPEG-2000-DSPE was the amount of methanol in the mobile phase. The use of an organic solvent in the mobile phase caused an increasing baseline following the composition of the gradient when using UV detection, but had no influence on the baseline flatness when using CAD detection. 4. Conclusion During this study a suitable HPLC–CAD method for the detection of lipids from liposomal preparations was established. It could be shown that this method is reproducible and significant for the most important evaluation parameters such as linearity, precision and accuracy. CAD was compared to measurements using a conventional UV detection and an improvement in the sharpness and height of the lipid peaks was seen with CAD. In any case, the charged aerosol detector appears to function as a universal detector which has many advantages in the field of liposome analysis where a precise and fast method of characterizing lipid excipients in detail is strongly needed. In future studies other lipids which are typically used in the process of preparing liposomes should be investigated for their detection behavior using HPLC–CAD based on the current data. Furthermore, a liposomal preparation bearing a drug substance should be analyzed by CAD in order to see if there are any difficulties in the parallel detection of lipids and a drug molecule. Acknowledgements We would like to thank our laboratory technicians Evelyne Scossa and Karin Mettenberger for the excellent assistance in the analyses and Marie Follo for proof-reading the manuscript. References [1] A.D. Bangham, M.M. Standish, J.C. Watkins, J. Mol. Biol. 13 (1965) 238. [2] A. Gabizon, H. Shmeeda, Y. Barenholz, Clin. Pharmacokinet. 42 (2003) 419.
786 [3] [4] [5] [6] [7]
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
P. Stano, S. Bufali, C. Pisano, J. Lipos. Res. 14 (1) (2004) 87. A. Gabizon, A. Meshorer, Y. Barenholz, J. Natl. Cancer Inst. 77 (1986) 459. T.M. Allen, P.R. Cullis, Science 303 (2004) 1818. K.J. Harrington, Expert Opin. Invest. Drugs 10 (6) (2001) 1045. L. Zhou, Draft Guidance for Industry on Liposome Drug Products, Food and Drug Administration, Center for Drug Evaluation and Research, 2002, http://www.fda.gov/cder/guidance/index.htm. [8] S.H. Lee, I.A. Blair, Methods Enzymol. 433 (2007) 159. [9] R.A. Moreau, P.T. Asmann, H.A. Norman, Phytochemistry 29 (1990) 2461. [10] R.A. Moreau, in: J.T. Lin, T.A. McKeon (Eds.), The Evaporative Light Scattering Detector, in HPLC of Acryl Lipids, HNB Publ., New York, p. 93.
[11] [12] [13] [14] [15] [16] [17] [18]
J. Loughlin, H. Phan, M. Wan, Am. Lab. 39 (2007) 24. T. Gorecki, F. Lynen, R. Szucs, Anal. Chem. 78 (2006) 3186. Q. Lu, J.A. Koropchak, Anal. Chem. 76 (2004) 5539. P.H. Gamache, R.S. McCarthy, S.M. Freeto, LC-GC Eur. 18 (2005) 345. P.H. Gamache, R.S. McCarthy, S.M. Freeto, LC-GC N. Am. Suppl. S (2005) 101. V.P. Torchilin, V. Weissig, Liposomes, Oxford Univ. Press, p. 4. M.C. Thomas, T.W. Mitchell, J. Am. Chem. Soc. 128 (2006) 58. International Conference on Harmonisation (ICH), Guidance for Industry: Q6A Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products: Chemical Substances, Federal Register 65 (251) (2000) Notices, pp. 83041–83063.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Simple and precise detection of lipid compounds present within liposomal formulations using a charged aerosol detector Carina Schönherr a,b , Sounia Touchene a , Gaston Wilser a , Regine Peschka-Süss b , Giancarlo Francese a,∗ a b
Novartis Pharma AG, Development, Lichtstrasse 35, CH-4058 Basel, Switzerland Department of Pharmaceutical Technology and Biopharmacy, University of Freiburg, Sonnenstrasse 5, D-79104 Freiburg, Germany
a r t i c l e
i n f o
Article history: Received 29 August 2008 Received in revised form 17 November 2008 Accepted 20 November 2008 Available online 3 December 2008 Keywords: Liposomes HPLC Lipid detection Corona aerosol detector (CAD) Phosphatidylcholine
a b s t r a c t In recent decades the use of liposomal preparations as drug delivery systems has become very attractive in pharmaceutical development. Therefore, thorough characterization and quantification of the lipids which form liposomes is wished from both investigators and regulatory authorities when the application in humans is being considered. In this study a new HPLC method for the detection of lipids in liposomal formulations was established using corona charged aerosol detection (CAD) which has the advantage to be independent of the chemical properties of the analytes. The superiority of this method over UV detection was demonstrated. Compared to UV detection no absorption effects of the organic solvent in the mobile phase interfering with the lipid signals were observed with CAD. CAD showed good linearity (R2 > 0.990) for all liposomal compounds. The acceptance criteria for precision including repeatability were met. The average recovery for each of the excipients of the liposomal formulation was in the range of 90.0–110%. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Currently, the use of liposomes as drug delivery vehicles is of growing interest in pharmaceutical drug development. Liposomes are vesicles in the nanometer size range composed of lipid bilayers, primarily phospholipids, which resemble natural membranes [1]. They are able to entrap water-soluble substances inside their aqueous core [2] and to incorporate water-insoluble substances into the lipid bilayer [3]. Due to this versatility and the ability of liposomes to improve the pharmacokinetics and pharmacodynamics of the associated drugs [2,4], pharmaceutical companies are increasingly developing liposomes as drug carrier systems for various drug substances, especially those which are highly toxic and or highly insoluble. The use of several liposomal formulations of anticancer drugs in tumor therapy, such as doxorubicin (Doxil® /Caelyx® ) and daunorubicin (DaunoXome® ), has already been established. Other liposomal drug formulations are currently being studied in clinical phase trials [5,6]. The authorities demand full qualitative and quantitative characterization, not just of the active substance itself, but also of the excipients within a drug formulation. The FDA’s draft guide for industry on liposome drug products also recommends to have an accurate lipid composition characterization (i.e., a set percent-
∗ Corresponding author. Tel.: +41 61 324 1908; fax: +41 61 324 6983. E-mail address: [email protected] (G. Francese). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.11.080
age of each lipid). The lipid components should be assayed, stress tested, as well as profiling of the degradation products related to the lipids [7,8]. In this context high-performance liquid chromatography (HPLC) has been well established as a fast and precise method to analyze both drug substances and excipients. However, for lipids which are the predominant component in liposomal formulations, conventional UV detection is often not adequate and limited to chromophores while other methods, which have been previously described for the detection of lipids, such as flame ionization detection (FID) [9] or evaporative light-scattering detection (ELSD) [10], have significant limitations in precision, sensitivity and dynamic range as demonstrated previously by Loughlin et al. [11–13]. In 2005, ESA (Chelmsford, MA, USA) developed an HPLC detector that allows the universal detection of various substances due to the independence of the analysis from both chemical structures and properties of the substances to be detected [14,15]. A so-called corona charged aerosol detection (CAD) system is compatible with all ordinary HPLC devices. After HPLC separation using a suitable column, the eluent is nebulized with nitrogen and then dried to remove the mobile phase, producing analyte particles. Particles which are too big are separated by impaction and conducted to the waste. A secondary stream of nitrogen is charged by a high-voltage platinum corona wire. The positive charge is transferred to the analyte particles, which then pass a negatively charged ion-trap where high-mobility particles are removed. Arriving at a collector the analyte particles transfer their charge, which is measured by a sensitive
782
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
electrometer. The resulting signal is directly proportional to the quantity of the analyte present in the sample. This innovative technology offers many advantages compared to conventional lipid detection methods. These include high sensitivity with a low limit of detection in the nanogram range, consistent response which is independent of chemical properties, excellent reproducibility with less than 2% relative standard deviation (RSD) and a dynamic range of up to four orders of magnitude allowing the measurement of quantities in the nanogram to microgram range. Furthermore, this detection method is applicable to a wide variety of applications [15]. In this study, an HPLC–CAD method was established for the analysis of conventional lipid components of liposomal formulations. Analysis by CAD was compared to analysis by conventional UV detection and validation parameters such as linearity, precision, accuracy and repeatability were investigated. 2. Experimental
First, 5 l of the liposome dispersion were diluted with 2 ml of phosphate buffer pH 7.4 and then measured at 20 ◦ C in a 90◦ angle. The average hydrodynamic diameter (z-average) was calculated out of 10 sub-runs. All liposome preparations used for CAD analysis showed a zaverage of approx. 100 nm with a polydispersity index of less than 0.1 (reflecting the narrowness of particle size distribution). 2.4. HPLC 2.4.1. HPLC instrumentation A Thermo Separation Products (TSP) HPLC device (Allschwil, Switzerland) was used with a P4000 pump and an AS 3000 sampler. For detection, a UV 6000 LP detector from TSP (with a linear dynamic range <5% deviation up to 2.0 AU) and a Corona Plus detector (with a dynamic range from 10 ng to 100 g [13]) from ESA was used. The evaluation-software used was Chromeleon (version 6.50) which was obtained from Dionex-Softron GmbH (Olten, Switzerland).
2.1. Chemicals and reagents Methanol, ethanol, acetonitrile and trifluoroacetic acid were of HPLC grade and purchased from Merck (Darmstadt, Germany). Water was purified on a Milli-Q system obtained from Millipore (Zug, Switzerland). For the preparation of liposomes, unsaturated soybean phosphatidylcholine (Lipoid S 100) was purchased from Lipoid (Ludwigshafen, Germany), N-carbonyl-methoxy(polyethyleneglycol 2000)-1,2-distearoyl-sn-glycero-3-phosphoethanolamine sodium salt (mPEG-2000-DSPE) was purchased from Genzyme Pharmaceuticals (Liestal, Switzerland), cholesterol was purchased from Fluka (Buchs, Switzerland) and d,l-␣-tocopherol was purchased from Novartis (Stein, Switzerland). For the hydration buffer potassium phosphate monobasic and disodium hydrogenphosphate dihydrate were obtained from Fluka (Buchs, Switzerland). 2.2. Preparation of PEG-liposomes The lipid composition used for all liposome preparations was as follows: unsaturated soybean phosphatidylcholine (Lipoid S 100), mPEG-2000-DSPE, cholesterol and ␣-tocopherol in a molar ratio of 100:7:20:0.9. All liposome batches were produced according to the thin film hydration method [16]. Briefly, the excipients were dissolved in ethanol and the solvent was evaporated using a rotavapor (Rotavap R 215, Büchi, Switzerland) for 1 h, slowly decreasing the pressure to 30 mbar and were further dried for 2 h to remove the solvent completely. The lipid film was then hydrated with a phosphate buffer pH 7.4 (potassium phosphate monobasic 20 mM, disodium hydrogenphosphate dihydrate 20 mM) under fast magnetic stirring in a 45 ◦ C hot water bath for 15 min to yield a liposome dispersion of approx. 80 mM. The liposome dispersion was extruded through a LIPEX extruder (Northern Lipids, Burnaby, Canada) first using a polycarbonate membrane with 100 nm pores three times (Millipore) then using a polycarbonate membrane with 50 nm pores six times (Millipore). 2.3. Particle size measurement using photon correlation spectroscopy (PCS) The average hydrodynamic diameter of the liposomes and the particle size distribution was analyzed by photon correlation spectroscopy using a Zetasizer 3000HS from Malvern, Worcestershire, UK.
2.4.2. HPLC conditions A XBridge C18 3.5 m column (150 mm × 3 mm) from Waters, Milford, MA was kept at 35 ◦ C with a column thermostat Jetstream 2 Plus from TSP (Allschwil, Switzerland) for all of the HPLC–CAD experiments performed, while the flow rate was 0.5 ml/min for the mobile phases (mobile phase A: acetonitrile/Millipore water (90/10, v/v) + 0.05% trifluoroacetic acid (TFA) and mobile phase B: methanol + 0.05% TFA). The injected volume was 10 l and the cooling temperature for the samples was 6 ◦ C, unless stated otherwise. 2.4.3. HPLC gradient HPLC was started at time point zero using 60% of mobile phase A and 40% of mobile phase B. After 25 min the mobile phase B was increased to 100%. Those conditions were maintained for another 10 min. Each HPLC run was stopped after 35 min. 2.4.4. Dilution of single lipid compounds For all excipients (phosphatidylcholine, mPEG-2000-DSPE, cholesterol and ␣-tocopherol) stock solutions were prepared with methanol/Millipore water (90/10, v/v) and then diluted with methanol/Millipore water (90/10, v/v) to reach the concentrations listed in Tables 1 and 2. 2.4.5. Dilution of liposome samples All liposomal samples were diluted 1:1000 with methanol/ Millipore water (90/10, v/v) which resulted in a final lipid concentration of approx. 80 M. This brought the lipid concentration into an appropriate range for CAD. 2.5. Identification of the phosphatidylcholine peaks by HPLC–MS Unsaturated soy phosphatidylcholine (PC) is a mixture of lipids with different fatty acid moieties. For Lipoid S 100 the chromatogram consists of five different peaks (Fig. 1). In addition an analysis with HPLC–mass spectrometry (MS) was performed with a Hewlett–Packard 1100 MSD device, to determine the main peaks corresponding to the different fatty acid compositions of phosphatidylcholine. The conditions used for HPLC were the same as those used for the HPLC–CAD method except for the sample temperature, which remained at room temperature. The detection conditions were the following: the gas temperature was set at 350 ◦ C, the volume of the drying gas was 12 l/min with a nebulizing pressure of 60 psig. The ionization mode was
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
783
Table 1 Correlation of the concentration of liposomal components to HPLC–CAD peak area. The correlation coefficient is taken as a parameter for linearity, the residual standard deviation (RSD) (≤3.5%, relative to the highest response) and the relative y-intercept (≤10.0%, relative to the highest response) were used as parameters for method precision (n = 2). Liposomal component
Concentration (g/ml)
Average area (mV s)
Residual standard deviation (RSD) (%)a
Correlation coefficient
y-Intercept (%)a
LODb (g/ml)
LOQc (g/ml)
Phosphatidylcholine (Main peak 1)
80 160 320 480 640
51.6 94.0 164.3 224.7 276.4
3.12
R2 = 0.9934
10.0
0.42
1.28
Phosphatidylcholine (Main peak 2)
80 160 320 480 640
48.5 91.3 163.6 229.1 289.5
2.15
R2 = 0.9970
7.1
0.28
0.86
mPEG-2000-DSPE
15 30 60 120 180
32.6 81.8 163. 298.1 418.5
3.15
R2 = 0.9948
2.7
0.33
0.99
Cholesterol
6 12 24 36 48 60
10.0 20.5 41.2 58.0 73.6 89.5
2.42
R2 = 0.9960
3.8
0.34
1.03
␣-Tocopherol
6 12 24 36 48 60
8.3 19.0 35.2 54.1 70.2 85.1
1.67
R2 = 0.9981
1.4
0.23
0.70
a b c
Calculated relative to the highest response. LOD = 3.3/S (slope S and as standard deviation of the y-intercept calculated from the calibration curve). LOQ = 10/S (slope S and as standard deviation of the y-intercept calculated from the calibration curve).
Table 2 The recovery rate of the liposomal lipid concentration is taken as a parameter for the accuracy of the HPLC–CAD detection method (n = 6). Liposomal component
Theoretical conc. (g/ml)
Recovered average conc. (g/ml)
Recovery (%)
Cholesterol ␣-Tocopherol Phosphatidylcholine (Main peak 1) Phosphatidylcholine (Main peak 2) mPEG-2000-DSPE
49.8 29.4 478.0 478.0 137.0
51.0 29.5 485.6 489.3 134.0
102.5 100.3 101.6 102.4 97.8
Fig. 1. HPLC–CAD chromatograms of every single liposomal compound (Toc: ␣-tocopherol; PC: phosphatidylcholine; Chol: cholesterol; PEG: mPEG-2000-DSPE) and a mixture of all excipients (Mix). The concentration of each compound is 50 g/ml.
784
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
atmospheric pressure ionization (API)–electrospray ionization (ESI) with positive polarity [17].
Table 3 Repeatability of the detection of single liposome compounds in solution, considering the mean retention time and peak area (n = 5).
3. Results and discussion
Liposomal component
3.1. Linearity The linearity of analysis using CAD was determined by analyzing different concentrations of each liposomal compound (Table 1). All solutions were injected twice for HPLC–CAD analysis. From MS results the main peak (Main peak 1) for phosphatidylcholine (758.6 g/mol [M+H]+ ) was identified having palmitic acid and linoleic acid as fatty acid chains (PC 16:0/18:2[9Z,12Z]) (retention time approx. 17.5 min). Another peak with a retention time of approx. 12 min, identified by MS as PC (18:1[9Z]/18:3[6Z, 9Z, 12Z]) (782.6 g/mol [M+H]+ ), was also examined in order to prove the linearity of the measurements (Main peak 2). The relation of peak area to concentration of excipients resulted in a good linearity with high correlation coefficients (>0.990) for all liposomal compounds (Table 1). 3.2. Precision As a criterion for determining the precision of the method, the residual standard deviation and y-intercept values were evaluated from linearity measurements taken relative to the highest response. It was requested that the residual standard deviation be lower than 3.5%, and that the y-intercept be lower than 10.0%. Therefore, the acceptance criteria were met (Table 1). 3.3. Accuracy In order to determine the accuracy of the measurements, a stock solution for each excipient was diluted with methanol/Millipore water (90/10, v/v) to obtain the theoretical concentration values listed in Table 2. Six diluted solutions were prepared equally and injected individually. The resulting average concentrations which were measured were then evaluated, in addition to an average recovery percentage which was used as quality parameters for determining the accuracy of the method. The average recovery was requested by the FDA’s guide for liposome drug products [7] according to the International Conference on Harmonisation (ICH) [18] to
Cholesterol ␣-Tocopherol Phosphatidylcholine (Main peak 1) Phosphatidylcholine (Main peak 2) mPEG-2000-DSPE
Retention time (min)
Area (mV min)
Mean (min)
RSD (%)
Mean (mV min)
RSD (%)
17.0 16.1 20.2
0.015 0.029 0.043
73.0 43.2 210.5
0.801 0.576 3.745
14.2
0.039
206.6
1.589
29.2
0.021
308.4
3.322
be in the range of 90.0–110.0%. This criterion was fulfilled for each of the excipients (Table 2). 3.4. Repeatability for each single excipient As evaluation-data for the repeatability of the method, both the retention time and the area under the peaks were evaluated (Table 3). The relative standard deviation (RSD) was evaluated as a criterion for the repeatability of the measurements. For the mean retention time all values of RSD were lower than 0.05%. Considering the peak area, the RSD was higher due to inexactness in the integration. 3.5. Repeatability for the measurement of liposome solution (mixture of lipids) The self-assembled liposomal structure only exists in an aqueous environment. After solution in an organic solvent, the liposomal configuration is destroyed. All lipids are separately dissolved in methanol/Millipore water (90/10, v/v). Hence, this lipid solution from liposomal dispersion (dilution 1:1000, refer to 2.3.4) was investigated by CAD analysis to ensure that there were no changes in retention time due to possible interactions between the single lipids. The diluted liposomal solution was injected six times. For retention time all of the measured values of RSD were lower than 0.08%, demonstrating high repeatability in the determination of lipid compounds by HPLC–CAD (Table 4). When peak area was examined the RSD was again higher due to inexactness in the integration.
Fig. 2. HPLC–CAD detection of PEG-liposomes. In total seven peaks from lipid compounds were detected: five peaks from phosphatidylcholine (conc. 50 g/ml, PC1 : 9.2 min; PC2 : 12.5 min; PC3 : 18.1 min; PC4 : 23.8 min; PC5 : 24.3 min) and one peak each for cholesterol (conc. 5 g/ml, Chol: 16.9 min) and mPEG-2000-DSPE (conc. 13 g/ml, PEG: 28.5 min). The peaks in the beginning of the chromatogram are from phosphate buffer (buffer). The concentration of the antioxidant ␣-tocopherol in the liposomal sample (conc. 0.25 g/ml) is at the limit of detection and therefore not visible.
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
785
Fig. 3. Comparison of CAD and UV detection for all liposomal compounds. (A) Cholesterol (conc. 30 g/ml), (B) ␣-tocopherol (conc. 18 g/ml), (C) phosphatidylcholine (Lipoid S 100) (conc. 160 g/ml), (D) mPEG-2000-DSPE (conc. 80 g/ml).
It has to be mentioned that the amount of the antioxidant ␣tocopherol in the liposomal formulation was below the limit of detection (Fig. 2). 3.6. Comparison CAD to UV detection A direct comparison between UV detection and the CAD for the same HPLC analysis was done to emphasize the need for a detection method for lipids which is more precise and sensitive than UV. Higher and sharper peaks were obtained using CAD compared to UV detection (Fig. 3). Additionally, the baseline increased with time for UV detection using the HPLC gradient elution. This was due to the increasing amount of methanol in the mobile phase during gradient elution. The absorption of the organic solvent overlapped with the lipid signals Therefore, some essential lipid peaks could not be precisely detected by UV. 3.7. Helpful observations made during method development During method development we found that the addition of TFA to the mobile phase was necessary to optimize the peak sharpness for mPEG-2000-DSPE. Another parameter that was found to
Table 4 Repeatability for the excipient mixture used in liposome dispersion, considering the mean retention time and peak area (n = 6). The amount of ␣-tocopherol is below the limit of detection. Liposomal component
Cholesterol ␣-Tocopherol Phosphatidylcholine (Main peak 1) Phosphatidylcholine (Main peak 2) mPEG-2000-DSPE
Retention time (min)
Area (mV min)
Mean (min)
RSD (%)
Mean (mV min)
RSD (%)
16.9 n.a. 18.3
0.029 n.a. 0.065
6.1 n.a. 19.6
0.575 n.a. 0.462
12.5
0.026
24.9
1.568
28.4
0.075
19.4
0.785
influence the sharpness of the peak for mPEG-2000-DSPE was the amount of methanol in the mobile phase. The use of an organic solvent in the mobile phase caused an increasing baseline following the composition of the gradient when using UV detection, but had no influence on the baseline flatness when using CAD detection. 4. Conclusion During this study a suitable HPLC–CAD method for the detection of lipids from liposomal preparations was established. It could be shown that this method is reproducible and significant for the most important evaluation parameters such as linearity, precision and accuracy. CAD was compared to measurements using a conventional UV detection and an improvement in the sharpness and height of the lipid peaks was seen with CAD. In any case, the charged aerosol detector appears to function as a universal detector which has many advantages in the field of liposome analysis where a precise and fast method of characterizing lipid excipients in detail is strongly needed. In future studies other lipids which are typically used in the process of preparing liposomes should be investigated for their detection behavior using HPLC–CAD based on the current data. Furthermore, a liposomal preparation bearing a drug substance should be analyzed by CAD in order to see if there are any difficulties in the parallel detection of lipids and a drug molecule. Acknowledgements We would like to thank our laboratory technicians Evelyne Scossa and Karin Mettenberger for the excellent assistance in the analyses and Marie Follo for proof-reading the manuscript. References [1] A.D. Bangham, M.M. Standish, J.C. Watkins, J. Mol. Biol. 13 (1965) 238. [2] A. Gabizon, H. Shmeeda, Y. Barenholz, Clin. Pharmacokinet. 42 (2003) 419.
786 [3] [4] [5] [6] [7]
C. Schönherr et al. / J. Chromatogr. A 1216 (2009) 781–786
P. Stano, S. Bufali, C. Pisano, J. Lipos. Res. 14 (1) (2004) 87. A. Gabizon, A. Meshorer, Y. Barenholz, J. Natl. Cancer Inst. 77 (1986) 459. T.M. Allen, P.R. Cullis, Science 303 (2004) 1818. K.J. Harrington, Expert Opin. Invest. Drugs 10 (6) (2001) 1045. L. Zhou, Draft Guidance for Industry on Liposome Drug Products, Food and Drug Administration, Center for Drug Evaluation and Research, 2002, http://www.fda.gov/cder/guidance/index.htm. [8] S.H. Lee, I.A. Blair, Methods Enzymol. 433 (2007) 159. [9] R.A. Moreau, P.T. Asmann, H.A. Norman, Phytochemistry 29 (1990) 2461. [10] R.A. Moreau, in: J.T. Lin, T.A. McKeon (Eds.), The Evaporative Light Scattering Detector, in HPLC of Acryl Lipids, HNB Publ., New York, p. 93.
[11] [12] [13] [14] [15] [16] [17] [18]
J. Loughlin, H. Phan, M. Wan, Am. Lab. 39 (2007) 24. T. Gorecki, F. Lynen, R. Szucs, Anal. Chem. 78 (2006) 3186. Q. Lu, J.A. Koropchak, Anal. Chem. 76 (2004) 5539. P.H. Gamache, R.S. McCarthy, S.M. Freeto, LC-GC Eur. 18 (2005) 345. P.H. Gamache, R.S. McCarthy, S.M. Freeto, LC-GC N. Am. Suppl. S (2005) 101. V.P. Torchilin, V. Weissig, Liposomes, Oxford Univ. Press, p. 4. M.C. Thomas, T.W. Mitchell, J. Am. Chem. Soc. 128 (2006) 58. International Conference on Harmonisation (ICH), Guidance for Industry: Q6A Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products: Chemical Substances, Federal Register 65 (251) (2000) Notices, pp. 83041–83063.