Pharmaceutical Chemistry Journal

Vol. 43, No. 12, 2009

A RAPID CHROMATOGRAPHIC/MASS SPECTROMETRIC METHOD FOR DIGOXIN QUANTIFICATION IN HUMAN PLASMA T. Grabowski*, A. Œwierczewska*, B. Borucka*, R. Sawicka*, M. Sasinowska-Motyl†, S. W. Gumu³ka†, Y. Zahariev††, A. Mitova††, and K. Zhilkova†† The relatively simple liquid-liquid extraction LC-MS/MS method using a UPLC column for measurement of digoxin concentrations both in medium and human plasma is described. Digitoxin was used as internal standard. The developed method possesses satisfactory accuracy, precision, and repeatability, and is economical and not time-consuming. It is the first method for precise measurement of digoxin concentrations in human plasma using combined HPLC equipment and UPLC column with a low limit of quantitation equal to 0.1 ng ml – 1 as verified in more than 1500 samples analyzed in a GCP GLP bioequivalence study. The developed method was successfully used in digoxin bioequivalence studies in which 26 volunteers were enrolled. Key words: digoxin, UPLC, bioequivalence, pharmacokinetics

enterohepatic circulation. These factors are responsible for the high pharmacokinetics variability of digoxin and make it difficulty to demonstrate the bioequivalence between generic and original medicine [6 – 8]. Furthermore, the absorption kinetics of digoxin was reputed to have both large intra- and intersubject variability. The intersubject coefficient of variation of the pharmacokinetics parameters is higher than 50%, but intrasubject variation is lower than 30%. The intrasubject coefficient of variation is usually smaller than the intersubject, but this parameter is generally recommended for sample size calculations in bioequivalence crossover design studies [9 – 11]. In digoxin bioequivalence studies the lower number of participants created a higher risk of nonequivalence. Thus, in digoxin bioequivalence studies, the method of digoxin determination in plasma must be accurate, precise, and stable in large sets of analysis. Present strategies of digoxin analysis using mass spectrometry technique are still relatively limited [13]. The mass spectrometry method is most suitable for digoxin analysis because mass spectral data of DLIF factors are known [14, 15]. Critical for digoxin bioequivalence study is the LOQ value, which causes a long elimination phase of the drug. Low LOQ at the level 0.1 ng ml – 1 was reached in the method of digoxin determination in rat plasma [13]. In this study we report the development and validation of the method of determination of digoxin concentrations in human plasma as verified in more than 1500 samples. The presented method is not time- and material-consuming, and requires one-step liquid-liquid extraction only. The described method was validated using a new analytical strategy — ultraperformance chromatographic column (UPLC) and typical HPLC equipment with double mass spectrometer. Suitability of the method was verified in more than 1500 samples obtained in a bioequivalence study, after

INTRODUCTION Digoxin is a slightly hydrophilic cardiac glycoside used in medicine as a cardiotonic drug. Like other cardiac glycosides, the therapeutic range of digoxin is narrow; therefore its plasma concentrations are especially important. Determination of digoxin in human plasma is traditionally performed by immunoassay, radioimmunoassay, fluorescence polarization immunoassay, and enzyme-linked immunosorbent assay (RIA, FPIA, ELISA), whereas quantification is usually performed by high-pressure liquid chromatography (HPLC) with UV or fluorescence detection. However, lower limits of quantitation (LLOQ) are often not suitable for bioequivalence study of this drug. Immunoassay, UV, and fluorescence detection are usually not specific techniques or are not sensitive enough, whereas derivatization and solid-phase extraction is highly time-consuming [1, 2]. Unfortunately, immunoassay methods are not specific, and cross reactions with digoxin metabolites or numerous endogenous digoxin-like factors (DLIF) has been described [3 – 5]. It has been shown that digoxin is a substrate of P-glycoprotein (P-gp) and multidrug-resistant protein (MDR-1), and it possesses a long halflife elimination phase and enters the *

Centre of Pharmacokinetics Research FILAB, Ravimed Sp. z o.o., Poland † Department of Pharmacodynamics, Medical University, Warsaw, Poland †† Laboratory of Pharmacokinetics and Clinical Trials, Medical and Regulatory Affairs Department, Sopharma PLC, Sofia, Bulgaria Correspondence: Dr. Tomasz Grabowski Centre of Pharmacokinetics Research FILAB Ravimed Sp. z o.o. Polna 54, 05-119 £ajski, Poland Tel/fax: (+ 48) 22 78 22 167 e-mail: [email protected]

710 0091-150X/09/4312-0710 © 2009 Springer Science+Business Media, Inc.

A Rapid Chromatographic/Mass Spectrometric Method

per os administration of a 0.25 mg digoxin tablet in 26 healthy volunteers. EXPERIMENTAL Chemicals and Reagents. Reference materials digoxin — DIG, CAS [20830-75-5], were purchased from LGC Promochem, and digitoxin — IS, CAS [71-63-6], was purchased from Fluka. Methanol, acetonitrile, and water (HPLC-MS grade) were purchased from J. T. Baker. Ammonium hydroxide 30%, 2-propanol, hexane, and chloroform were purchased from J. T. Baker. Ammonium trifluoroacetate was purchased from Fluka. Instrumentation. All liquid transfers, including plasma sample transfers from 2 ml Eppendorf microfuge tubes into 13/100 mm glass tubes (Pyrex®, USA) and IS addition, were performed using a pipette controller Pipetus (Hirschmann® Laborgeräte GmbH & Co. KG, Eberstadt, Germany) and automatic volumetric pipettes 10 – 100 ml and 100 – 1000 ml (Socorex Isba S. A. Switzerland). Centrifuge MPW 350 R (MPW Med. instruments, Spó³dzielnia Pracy, Warsaw, Poland) was utilized during the sample preparation. Analytical balance WAX-40/160 (RADWAG® Wagi Elektroniczne, Radom, Poland) was used for standard preparations. Vortex TTS-2 (IKA® Werke GmbH & Co. KG, Staufen, Germany) and magnetic stirrers Big Squid (IKA® Werke GmbH & Co. KG, Staufen, Germany) were used. Evaporation was performed into a Zymark TurboVap LV® (Caliper Life Sciences, Hopkinton, USA) by nitrogen in agreement with PN-C-84920 – 97, 99.995% (Anco Gaz Bia³ystok). All analytical samples were filtered by Syringe filter Titan PTFE, 0.45 mm. Analysis was performed using Total Recovery Vials (Waters Corporation, Milford, USA). A Quattro micro™ API Micromass® system with an electrospray ion source, nitrogen generator NM30LA (Peak Scientific Instruments Ltd.), and separation module Alliance Waters 2695 (Waters Corporation, Milford, USA), equipped and operating under MassLynx™ v. 4.0 (Waters Corporation, Milford, USA) software, was used. Validation of the bioanalytical method was confirmed for all concentration levels obtain during 3 subsequent days. All calculations were processed by Validation Manager* version: 2.18I © VWR INTERNATIONAL (VWR International S. A. S., France) software. Pharmacokinetics analysis was performed using WinNonLin 5.0.1 Professional, software (Pharsight Corporation, Mountain View, CA, USA). Chromatographic conditions. The gradient HPLC elution mobile phase A was composed of 70% acetonitrile and 30% (v/v) 0.001 M ammonium trifluoroacetate, mobile phase B — 100% water. The gradient flow rate was (time — min;%A;%B; flow rate, ml min–1): (0.00;5.0;95.0;0.05), (0.20;5.0;95.0;0.10), (3.00;50.0; 50.0;0.10), (6.00;90.0;10.0;0.10), (6.10;99.0; 1.0;0.10), (9.50;99.0;1.0;0.10), (10.00;5.0;95.0;0.10). A UPLC® AQUITY® column was used for sample analysis (1.7 mm,

711

100 mm ´ 2.1 mm; Waters Corporation, Milford, USA). The column was maintained at a temperature of 35°C, while the autosampler temperature was set at 10°C. The pressure of the system during the analysis was about 3000 psi. The injection volume was 10 ml, and the total run time was set for 10 min. Mass spectrometric conditions. The mass spectrometer was operated in the negative ion mode, using the electrospray source and interface (ESI–). The tuning parameters were optimized for DIG and IS by infusing a solution containing 100.00 ng ml-1 via a syringe directly connected to the mass spectrometer. Function 1 — source cleaning from 0 – 4 min of each analysis: source ionization (ESI+), cone gas flow 200 l h– 1. desolvation gas flow 600 l h – 1. Other parameters as in function 2. Function 2 — main analysis 4 – 10 min: The source temperature was 80°C, desolvation temperature 285°C, desolvation gas flow 800 l h – 1, and cone gas flow 100 l h – 1. The capillary voltage was set at 3.9 kV, while the optimized cone voltage values for DIG and IS were 18 V. The multiplier was set at 650 V, and 0 flow argon was used as the collision gas. The analytes were detected by monitoring the maternal ion using a single ion recording (SIR) scan mode with 100 ms dwell time for each ion. The selected transitions m/z were 780.94 ® 893.5 for DIG and 764.94 ® 877.0 for IS corresponding to adduct of molecules [M+CF3COO–]. All the data were acquired using the MassLynx™ v. 4.0 (Waters Corporation, Milford, USA) software. Preparation of the standard quality control and method validation samples. Stock solutions of DIG [1.0 mg ml – 1 (DIG1) and the intermediate standard solution 1.0 mg ml – 1 (DIG2) and 0.05 mg ml – 1 (DIG3)] and IS [1.00 mg ml – 1 (IS1) and the intermediate standard solution 0.50 mg ml – 1 (IS2)] were prepared by dissolving each of the accurately weighed reference compounds in MeOH. Working solutions of DIG were 3.00 ng ml – 1, 2.50 ng ml – 1, 2.00 ng ml – 1, 1.50 ng ml – 1, 1.00 ng ml – 1, 0.80 ng ml – 1, 0.50 ng ml – 1, 0.20 ng ml – 1, and 0.10 ng ml – 1. Working solutions of IS were 15.00 ng ml-1. Working solutions of DIG and IS were prepared by diluting DIG2, DIG3, and DGX2 with MeOH. Working stock solutions were used to prepare three levels of QC working solutions at the lowest concentrations — the lowest quality control LQC, the medium quality control MQC, and the higher quality control HQC, for DIG and IS respectively (0.20 ng ml – 1, 1.50 ng ml – 1, 2.50 ng ml – 1, 15.00 ng ml – 1). All working solutions were prepared in a 25 ml volumetric flask and were stored along with the stock solutions at 4°C. Calibration standards, LQC, MQC, HQC were prepared in the same matrix (human plasma). The calibration curve consisted of a one blank sample and ten standards covering the expected range of concentrations to be quantified. Sample extraction and preparation. All plasma samples were thawed at room temperature and vortexed. After de-capping the Eppendorf tubes, 1.0 ml of plasma was transferred into a glass tube containing 30 ml of IS working stan-

712

T. Grabowski et al.

DIG

reconstitued in 50 ml of a mixture of 0.001 M trifluoroacetic acid and acetonitrile 7:3 (v/v), then filtered by a syringe filter Titan PTFE 0.45 mm into an autosampler vial. RESULTS The extracts provided by the fast liquid-liquid extraction were very clean, and after analysis of > 1500 samples, the retention times for DIG and IS were stable and equaled 5.26 and 7.33 min respectively. The applied UPLC column gives 3 – 5 times higher DIG and IS signals than the typical MS-LC columns (unpublished data). As the particle size in the UPLC column was 1.7 mm (high pressure), the mobile phase flow was lower than 0.10 ml min – 1. Furthermore, the column backpressure measured during the analysis of the last samples increased circa 5 – 10% compared with the initial values. The effect of samples on the analysis was of minor significance. The effect of plasma on the ion source and source capillary was not observed. A range of more than 1500 analysis using the same column without source cleaning and source capillary replacement was performed.

IS

Standard curve

Fig. 1. Representative SIR chromatogram of digoxin — DIG (lowest point of the curve — lower limit of quantitation LLOQ) and digitoxin — IS from spiked plasma sample.

dard solution (15.00 ng ml – 1), and 100 ml 10% ammonium hydroxide was added. During a 5-sec period of mixing on a vortex at 1400 rpm, 3.5 ml of chloroform was added and the whole shaken for 30 min at 250 rpm on a vortex. The next step was centrifugation of the glass tubes for 10 min at 4.075 ´ 103 g at a temperature of 20°C. After centrifugation, the organic layer was transferred into a clean glass tube. The organic layer was evaporated for 30 min at a temperature of 40°C under a stream of nitrogen. The dry residue was

A calibration curve, containing nine nonzero standards ranging between 3.00 ng ml – 1 and 0.20 ng ml – 1 DIG for each analytical run, was prepared. This range of linearity was optimal for a bioequivalence study after per os administration of a 0.25 mg tablet in 26 healthy volunteers. Typical chromatograms of plasma samples spiking DIG at values equaling the lowest calibration curve point and IS are presented in Fig. 1. Linearity in the validation procedure was assessed for nine calibrations for DIG 3.00 ng ml – 1, 2.50 ng ml – 1, 2.00 ng ml – 1, 1.50 ng ml – 1, 1.00 ng ml – 1, 0.80 ng ml – 1, 0.50 ng ml – 1, 0.20 ng ml – 1, and 0.10 ng ml – 1. The experimental limit of detection, LOD, was 0.025 ng ml – 1. A study of linearity was conducted during the subsequent 3 days (Table 1). For each calibration curve the following parameters were calculated: compatibility of the y-intercept with the value 0 using the Student’s test to confirm a lack of systematic error in the method, analysis of the variance of the regression (test F1), and correlation coefficient r. Linearity of the method was accepted when the coefficient of correlation is r £ 0.990, the test F1 (F calculated > F table) is confirmed, and the test compatibility of the y-intercept with the value 0

TABLE 1. Results of calibration curves from linearity analysis Calibration curve parameters

Correlation coefficient

F1 test

Mandel test

Day of study

I II III

slope

Y intercept

r calc.

r tab.

F calc.

F tab.

F tab.

F calc.

3.62208e+01 4.00106e+01 3.40199e+01

7.33364e+00 6.93951e+00 1.28893e+01

0.9954 0.9962 0.9969

0.990 0.990 0.990

2726.90 3309.40 4059.30

4.240 4.240 4.240

5.720 5.720 5.720

2.17 0.39 3.98

A Rapid Chromatographic/Mass Spectrometric Method

(tcalculated > ttable) is confirmed. A zero and a blank plasma sample were also prepared for each analytical run during the study to confirm the absence of interference. The determination coefficients (R2) for the 13th batch during the study were equal to 0.9978 ± 0.001, and the average linear slope was 0.3298 ± 1.0286 for DIG. Accuracy and precision Accuracy was established over the range of linearity based on the data from a linearity study performed in nine runs during 3 days. Back-calculated concentrations of DIG were obtained from calibration standards with the equation of the relevant calibration curve (for 3 days). The fitting line for the calculated concentrations of DIG versus the known concentrations should be validated by the Student’s test of the compatibility of the calculated slope with the value 1 and compatibility of the y-intercept with the value 0. Assessment of the mean recovery associated with its confidence interval is done through calculation regression analysis. In all cases the values were within the acceptable range (Table 2). The accuracy acceptance criterion, accuracy of the method, is accepted when back-calculated concentrations of DIG are within the calibration standards using the equation of the relevant calibration curve (for 3 days). Fitting lines to Student’s tests of compatibility are confirmed for a calculated slope with the value 1, and the y-intercept with the value 0. The confidence interval of the mean recovery over the range of linearity should include a 100% value. The precision was estimated for all calibration standards of a bioanalytical method, based on linearity data obtain during 3 days. Repeatability and intermediate precision were estimated by calculating the intra- and inter-group variance. Precision acceptance criterion: the repeatability coefficient of variation in percentage (RSD%) should be £ 10 expected at the lowest limit of quantitation, LLOQ £ 15. The intermediate precision coefficient of the variation in percentage (RSD%) should be £ 15 expected at LOQ £ 20 (Table 3).

Data from nine runs were used in order to assess the extraction efficiency of the analytical method. Samples from all standard curve points, extracted according to the developed

TABLE 2. Results of determination of the accuracy validation method

I II III

F1 test F calc.

2724.1 3309.9 4057.6

method, were compared with standard curve samples for each concentration obtained by diluting working solutions directly in the mobile phase (unextracted samples). The recovery was evaluated by calculating the mean of the area or the response of each concentration and dividing the extracted sample mean by the unextracted sample mean of the corresponding concentration. The total recovery (efficiency of extraction) of DIG was equal to 102.00%, with RSD% for all concentrations equal to 6.55%. Stability studies Stability experiments were performed as part of the method validation protocol in order to evaluate the autosampler stability, freeze and thaw stability, and working standard solution stability. In the stability tests that used the freeze and thaw steps, samples were frozen at –75°C. To evaluate all the stabilities, plasma samples containing LQC, MQC, and HQC concentrations of DIG were used. The acceptable variation was < 15%. Three subsequent cycles of freeze and thaw were performed. Stability was estimated based on the equation analyte area/IS area x IS concentration after each freeze and thaw cycle. Results were acceptable for all cycles. RSD% after the third freeze and thaw cycle for LQC, MQC, and HQC concentrations were equal to –10.48, –0.70, –4.01 respectively. The autosampler stability was established for the time period from samples placed in the autosampler to reanalysis after 48 h. Samples of analytes and internal standard in the biological matrix were stable for 72 h during storage in the autosampler at a temperature of 10°C. RSD% after the autosampler stability test for LQC, MQC, and HQC concentrations were equal to –0.12, –0.83, and –1.12 respectively. The working standard solution of DIG and the internal standard are stabile for storage condition 0 – 8°C for a period of 14 days. RSD% working standard solutions for LQC, MQC, and HQC concentrations were equal

TABLE 3. Results of the evaluation of precision

Extraction recovery

Day of study

713

F tab.

4.24 4.24 4.24

Mean recovery, Mean recovery 80 – 120% interval, %

101.13 99.36 100.21

2.55 2.89 2.40

No.

Acceptance Acceptance Digoxin criteria Repeatability Intermediate criteria concentrafor the interprecision, precision, for the tion, mediate RSD % RSD % repeatability, ng ml – 1 precision, % %

1.

0.100

7.94

7.94

£ 15

£ 20

2.

0.200

5.70

7.00

£ 10

£ 15

3.

0.500

5.21

5.55

£ 10

£ 15

4.

0.800

3.80

8.34

£ 10

£ 15

5.

1.00

6.61

6.61

£ 10

£ 15

6.

1.50

4.36

4.36

£ 10

£ 15

7.

2.00

4.81

4.81

£ 10

£ 15

8.

2.50

2.04

3.96

£ 10

£ 15

9.

3.00

4.93

5.18

£ 10

£ 15

714

T. Grabowski et al.

2.5

concentration [ng ml -1]

2

1.5

1

0.5

0 0

10

20

30

40

50

60

70

time [h]

Fig. 2. Mean plasma concentration — time profiles of digoxin in 26 healthy volunteers after per os administration of a 0.25 mg tablet.

to 9.36, –5.33, –1.27 respectively. Long-term stability was performed between the first and last day of the study (130 days). RSD% for plasma samples with LQC, MQC, and HQC concentrations were equal to 7.50, –5.23, and –3.35 respectively. Quality assurance during the bioequivalence study SST — System Suitability Test. Chromatogram quality was estimated with chosen parameters of SST and MassLynx™ v. 4.0 software. Whole chromatograms (column condition), analytes, and internal standard peaks were estimated. The parameters taken for this estimation were the signal/noise ratio (S/N), peak height/peak area ratio (H/A), and peak skew. Separation was correct (accepted) when S/N > 10, NH < 5, and peak skew ± 1.5 for all peaks. Therefore, no outliers could be detected in a range of more than 1500 analyses. More than 5% LQC, MQC, and HQC samples were prepared for every batch in the study. A total of 125 LQC, MQC, and HQC was prepared. LQC samples have a mean value of (RSD%) 9.84, MQC 7.19, and HQC 7.82. Application to a bioequivalence study The present method was applied to the analysis of plasma samples obtained from 26 healthy volunteers at a bioequivalence study. The study followed ethical principles of the Declaration of Helsinki. The clinical part of the study was performed according to Good Clinical Practice (GCP) rules. The bioanalytical part of the study was performed in accordance with Good Laboratory Practice (GLP) requirements. The mean remaining area of the curve ([1 – (AUC0–Clast)/(AUC0–¥)] ´ 100%) was not applicable for this study because of the very long elimination phase of

digoxin (>24 h), making it possible to stop sampling after 72 h [16 – 18]. The pharmacokinetic parameters of DIG were described as follows. The arithmetic mean and standard deviation (± SD) value of the area under the curve, AUC (ng h ml – 1), from time 0 to 72 h (AUC0–72 h) was 18.18 ± 3.58 for the test and 18.60 ± 3.43 for the reference drug. The value of the peak concentration, Cmax (ng ml – 1), was 1.88 ± 0.51 for the test and 2.24 ± 0.71 for the reference drug. The value of the time of peak concentration, tmax (h), was 0.90 ± 0.22 for the test and 0.85 ± 0.26 for the reference drug. The value of the halflife of the drug in the elimination phase, t1/2 (h), was 37.76 ± 11.01 for the test and 38.36 ± 10.01 for the reference drug. The concentration time profile of DIG is presented in Fig. 2. DISCUSSION The current analytical method was fully validated in our laboratory according to the US Food and Drug Administration (FDA) and the European Medicines Agency (EMEA) bioanalytical method validation requirements [19 – 25]. Liquid-liquid extraction and LC-MS/MS using a technologically advanced UPLC column for the quantification of digoxin in human plasma are presented. It is the first method for cardiac glycoside analysis in human plasma and uses the combined HPLC/UPLC equipment strategy. Application of a high-resolution UPLC column provides great sensitivity and does not require a longer analytical run time than a simple HPLC column. Finally, a short run time of analysis and a small number of steps in sample preparation with a simple mobile phase was applied. The presented method has excellent linearity not requiring any weighting in quantitation. Applying a UPLC column and a simple extraction method without SPE ensures excellent values of LOQ in human plasma

A Rapid Chromatographic/Mass Spectrometric Method

[26 – 29]. The method was successfully applied and verified in a bioequivalence study of digoxin based on more than 1500 samples analyzed. The presented data describe the first bioanalytical method for digoxin determination using HPLC/UPLC instrumentation based on more than 1500 samples verified in a bioequivalence study with GCP and GLP rules. The UPLC column gives increased peak concentrations with reduced chromatographic dispersion at lower flow rates than HPLC. That effect was shown as a new analytical procedure in pharmacokinetics study of digoxin in human plasma. ACKNOWLEDGEMENTS We thank dr hab. Wieslaw Raszewski for critical reading and financial support. REFERENCES 1. J. Hui, D. R. Geraets, A. Chandrasekaran, et al., J. Clin. Pharmacol., 7, 734 – 741 (1994). 2. M. C. Tzou, R. A. Sams, R. H. Reuning, J. Pharm. Biomem. Anal., 13, 1531 – 1540 (1995). 3. A. Dasgupta, K. T. Yeo, S. Malik, et al., Biochem. Biophys. Res. Commun., 2, 623 – 628 (1987). 4. A. Sophocleous, I. Elmatzoglou, A. Souvatzoglou, J. Endocrinol. Invest., 7, 668 – 674 (2003). 5. M. Yoshika, Y. Komiyama, M. Konishi, et al., Hypertension, 49, 209 – 214 (2007). 6. J. E. Doherty, W. J. Flanigan, M. L. Murphy, et al., J. Sherwood, Circulation, 42, 867 – 873 (1970). 7. I. A. de Lannoy, M. Silverman, Biochem. Biophys. Res. Commun., 189, 551 – 555 (1992). 8. C. Verstuyft, M. Schwab, E. Schaeffeler, et al., Eur. J. Clin. Pharmacol., 58, 809 – 812 (2003). 9. J. H. Zar, Biostatistical Analysis, 2nd Ed., Prentice-Hall, Inc., Englewood Cliffs, New Jersey (1984). 10. P. Armitage, Statistical Methods in Medical Research, Wiley and Sons, New York (1973).

715

11. S. Bolton and C. Bon, Pharmaceutical Statistics: Practical and Clinical Applications, 4th Ed., Revised and Expanded Marcel Dekker ed. (2004), pp. 151 – 172. 12. Y. Ming, H. Zhang, S. Chong, et al., J. Pharm. Biomed. Anal., 32, 1189 – 1197 (2003). 13. M. Yao, H. Zhang, S. Chong, et al., J. Pharm. Biomed. Anal., 32, 1189 – 1197 (2003). 14. W. R. Mathews, D. W. DuCharme, J. M. Hamlyn, et al., Hypertension, 6, 930 – 935 (1991). 15. H. M. A. M. Qazzaz, R. Valdez, Arch. Biochem. Biophys., 328, 193 – 200 (1996). 16. Anonymous, Guidance for Industry. Bioavailability and Bioequivalence Studies for Orally Administered Drug Products – General Considerations, 1 – 21 (2002). 17. Anonymous, Notice for Industry. Bioequivalence Requirements for Long Half-life Drugs. Health Canada, 1 – 2 (2005). 18. CDER, Reviewer Guidance Validation of Chromatographic Methods. Food and Drug Administration, Center for Drug Evaluation and Research, Rockville MD (1994). 19. V. P. Shah, K. K. Midha, S. Dighe, et al., Eur. J. Drug. Metab. Pharmacokinet., 16, 249 – 312 (1991). 20. ICH, Text on Validation of Analytical Procedures Q2A (1994). 21. WHO, Use of anticoagulants in diagnostic laboratory investigations, Rev.2 (1999). 22. CDER, Guidance for Industry Analytical Procedures and Methods Validation. Food and Drug Administration, Center for Drug Evaluation and Research, Rockville MD (2000). 23. CDER, Guidance for Industry: Bioanalytical Method Validation. US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research, Rockville MD (2001). 24. EMEA, Note for Guidance on the investigation of bioavailability and bioequivalence, CPMP/EWP/QWP/ 1401/98 (2001). 25. European Parliament and The Council, Directive 2004/10/EC. Off., J. Europ. Un., 44 – 58 (2004). 26. P. Kaiser, U. Kramer, D. Meissner, et al., Clin Lab., 49(7 – 8), 329 – 43 (2003). 27. K. Mitamura, A. Horikawa, Y. Yamane, et al., Biol. Pharm. Bull., 9, 1653 – 1656 (2007). 28. Y. Hashimoto, K. Shibakawa, S. Nakade, Y. Miyata, J. Chromatogr. B Anal. Technol. Biomed. Life Sci., 869(1 – 2), 126 – 32 (2008). 29. S. Shi, Z. Li, H. Chen, F. Zeng, J. Chromatogr. B Anal. Technol. Biomed. Life. Sci., 875(2), 405 – 410 (2008).