SHORT COMMUNICATION
Validation of a High-Performance Liquid Chromatography–Tandem Mass Spectrometry Method for the Determination of Perhexiline and Cis-Hydroxy-Perhexiline Plasma Concentrations Ian S. Westley, PhD,*† Giovanni Licari, PhD,‡ and Benedetta C. Sallustio, PhD*§
Background: The polymorphic nature of cytochrome P450 2D6 has made therapeutic drug monitoring of the anti-anginal agent perhexiline a compulsory step in reducing adverse events associated with plasma concentrations above the therapeutic range (0.15–0.60 mg/L). The aim of this study was to develop a highperformance liquid chromatography–mass spectrometry/mass spectrometry method for the determination of plasma perhexiline concentrations and its major metabolite cis-hydroxy-perhexiline to reduce sample extraction procedures and improve sample turnaround times.
Methods: The method was validated by determining the precision and accuracy of calibrators and quality control material, comparing quality assurance program samples and patient samples measured by a previously reported liquid–liquid extraction fluorescence (FL) detection high-performance liquid chromatography method and performing matrix effects investigations. Results: Replicates of calibrators at concentrations of 3.00 and 0.05 mg/L demonstrated imprecision of ,10.8% and inaccuracy of ,8.2% for perhexiline and ,10.1% and ,4.5% for cishydroxy-perhexiline, respectively. All samples measured by the 2 methods (n = 102) demonstrated Deming regression of perhexiline = 1.20 FL + 0.00 (Sy.x = 0.08, 1/slope = 0.67); cis-hydroxyperhexiline = 1.48 FL 2 0.20 (Sy.x = 0.40, 1/slope = 0.67). Conclusions: The assay performance was deemed acceptable and integrated into the routine therapeutic drug monitoring program of the department. Key Words: perhexiline, HPLC-MS/MS, validation (Ther Drug Monit 2015;37:821–826)
Received for publication June 9, 2014; accepted March 2, 2015. From the *Department of Clinical Pharmacology, Basil Hetzel Institute, The Queen Elizabeth Hospital, Woodville; †School of Pharmacy and Biomedical Science, University of South Australia; ‡Vascular Diseases and Therapeutics Research Group, Basil Hetzel Institute, The Queen Elizabeth Hospital; and §Discipline of Pharmacology, University of Adelaide, Australia. The authors declare no conflict of interest. Correspondence: Ian S. Westley, PhD, Department of Clinical Pharmacology, Basil Hetzel Institute, The Queen Elizabeth Hospital, 28 Woodville Rd, Woodville 5011, South Australia, Australia (e-mail: [email protected]). Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Ther Drug Monit Volume 37, Number 6, December 2015
INTRODUCTION
The anti-anginal agent perhexiline was first introduced in the 1970s for the treatment of angina1; however, its use declined because of the occurrence of adverse events such as neuropathy and hepatotoxicity.2,3 Perhexiline is primarily metabolized by cytochrome P450 2D6, which is subject to genetic polymorphisms that results in approximately 10% of the population having significantly reduced clearance (poor metabolisers).4 Therapeutic drug monitoring (TDM) has become beneficial to maintain perhexiline plasma concentrations in the range 0.15–0.60 mg/L to avoid adverse events.5,6 Using TDM and determining the concentration of the main metabolite, cis-hydroxy-perhexiline, patients can be phenotyped into poor, extensive, and ultra rapid metabolisers of perhexiline based on their individual metabolite to parent concentration ratio. Although poor metabolisers require doses between 50 and 100 mg once a week, doses for the other phenotypic groups range from 50 to 500 mg daily.7 It is therefore beneficial for patients to use resources like TDM to provide optimal therapy. Previously, our laboratory used a complex highperformance liquid chromatography (HPLC) method to determine perhexiline plasma concentrations that incorporated multiple liquid–liquid extractions and derivatization for fluorescent detection.8 There have been other fluorescent detection methods published for the separation of the perhexiline enationmers in human plasma9 and its metabolites in plasma, urine, and human liver microsomes,10 but using mass spectrometry detection can provide better specificity and allow simplified sample extraction procedures. The aim of this study was to validate a simple rapid analytical method for the determination of perhexiline and cis-hydroxyperhexiline to reduce sample extraction procedures and improve specimen turnaround times.
METHODS Materials Perhexiline maleate was purchased from Sigma Chemical Corporation (MO); cis-hydroxy-perhexiline and hexadiline hydrogen chloride were a gift from Marion Merrell Dow Inc (OH); ammonium acetate was purchased from VWR International (Leuven, Belgium), and formic acid was purchased from Merck KGaA (Darmstadt, Germany). Methanol, acetonitrile, and
821
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Westley et al
TABLE 1. Imprecision (n = 6) and Inaccuracy (n = 8) Data for Perhexiline and cis-Hydroxy-Perhexiline at Calibrator Concentrations of 3 and 0.05 mg/L Within Run (%) (n = 6)
Within Run (%) (n = 6)
Between Run (%) (n = 8)
Between Run (%) (n = 8)
3 mg/L
0.05 mg/L
3 mg/L
0.05 mg/L
6.3 3.0
8.8 28.2
4.1 24.5
10.8 26.9
4.3 2.7
8.2 4.5
6.6 3.0
10.1 21.6
Perhexiline-Spiked Concentration, mg/L Imprecision Inaccuracy cis-hydroxy-perhexiline–spiked concentration, mg/L Imprecision Inaccuracy
hydrochloric acid were purchased from Thermo Fisher Scientific Pty Ltd (Victoria, Australia). Water (.18.2 MV) was prepared using a Cascada AN-water system (Pall Life Sciences, NY).
Stock Solutions Perhexiline stock solution (100 mg/L) was prepared in 25 mL by dissolving 7.09 mg of perhexiline maleate in methanol (5 mL, 20% of final volume) and making up to volume with 0.1 mol/L hydrochloric acid. Cis-hydroxyperhexiline (100 mg/L) and hexadiline (14 mg/L) were prepared as above and stocks stored at 2–88C.
Calibrators
The method was validated using “in-house” calibrators and controls. Calibrators and quality control (QC) samples were prepared from separate stock solutions and spiked into drug-free human plasma. Calibrators (n = 6) were prepared at concentrations of 0.05, 0.25, 0.50, 1.00, 2.00, and 3.00 mg/L for perhexiline and cis-hydroxy-perhexiline and QCs spiked with weighed-in concentrations of 0.15, 0.75, and 1.50 mg/L for both analytes. Aliquots of calibrators and QCs were stored at 2208C until use.
Patient and Quality Assurance Program Samples Patient samples (n = 45) were analyzed for perhexiline and cis-hydroxy-perhexiline as part of routine TDM by the
previously described liquid–liquid extraction method and assayed on the same day by the HPLC-MS/MS method described in this article. Quality assurance program (QAP) samples (n = 57) were collated and assayed as part of the international external QAP (managed by our laboratory) by the previously described liquid–liquid extraction method and by the HPLCMS/MS method described in this article. QAP samples were a mixture of pooled patient samples and weighed in perhexiline/cis-hydroxy-perhexiline concentrations.
Sample Preparation
Calibrator, QC, or patient samples (50 mL) were aliquoted into clean 1.5 mL centrifuge tubes. Acetonitrile (150 mL) containing the internal standard (IS) hexadiline (0.2 mg/L) was added and tubes mixed and then centrifuged at 16,000g for 5 minutes. Supernatant (50 mL) was transferred to a clean 1.5 mL centrifuge containing 0.05% formic acid (100 mL) and mixed. The contents were transferred into HPLC vials and analytes separated using the chromatographic conditions described below.
Chromatographic Conditions The HPLC system consisted of an Agilent 1100 series (Agilent Technologies, CA) pump, degasser, injector, and column oven. Twenty-five microliters of preparation was injected, and analytes were separated using a Luna 2.1 · 50 mm (5 mm) Phenyl-hexyl column (Phenomonex Inc,
TABLE 2. Imprecision (n = 6) and Inaccuracy (n = 8) Data for Perhexiline and cis-Hydroxy-Perhexiline at QC Samples at Concentrations of 0.15, 0.75, and 1.50 mg/L
Perhexiline-Spiked QC Concentration, mg/L Imprecision Inaccuracy cis-hydroxy-perhexiline– spiked QC concentration, mg/L Imprecision Inaccuracy
822
Within Run (%) (n = 6)
Within Run (%) (n = 6)
Within Run (%) (n = 6)
Between Run (%) (n = 8)
Between Run (%) (n = 8)
Between Run (%) (n = 8)
0.15 mg/L
0.75 mg/L
1.50 mg/L
0.15 mg/L
0.75 mg/L
1.50 mg/L
3.0 11.7
3.3 22.0
7.0 24.8
13.4 23.6
8.7 27.4
9.1 214.8
5.1 24.6
3.0 28.3
6.6 24.8
14.2 27.78.7
14.3 28.2
15.0 215.0
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Validation of HPLC-MS/MS Assay for Perhexiline and Metabolite
CA) maintained at 508C. The mobile phases consisted of A: 100% methanol, 2 mmol/L ammonium acetate, 0.1% formic acid, B: 100% water, 2 mmol/L ammonium acetate, 0.1% formic acid. The initial mobile phase consisted of methanol: water (50:50 vol/vol A:B) pumped at a flow rate of 300 mL/ min and increased to 80% methanol (A) over 0.7 minutes. Analytes were eluted using 80% methanol (A) between 0.7 and 2.8 minutes, then returning to methanol:water (50:50 vol/ vol A:B) with a total run time of 4 minutes.
0.05 mg/L, respectively, a patient sample with a perhexiline concentration of 0.40 mg/L and cis-hydroxy-perhexiline concentration of 1.50 mg/L, and a zero sample. Analytes eluted with times of 2.0, 2.4, and 2.4 minutes for cis-hydroxyperhexiline, perhexiline, and hexadiline, respectively. On instrument, sample stability was assessed by extracting a calibration curve and QC samples as per protocol and analyzed. The extracts were kept on the instrument for a 24-hour period and reinjected, data analyzed, and calibration curve established. QC sample concentrations calculated from the reinjected samples remained in the accepted ranges of 62 SDs of the mean perhexiline and cis-hydroxyperhexiline concentrations determined during imprecision and inaccuracy experiments.
Mass Spectrometry Conditions Analytes were detected using multiple reaction monitoring in positive ionization mode on an API2000 mass spectrometer (AB SCIEX Pty Ltd, MA). Perhexiline m/z 278.3 . 81.1, cis-hydroxy-perhexiline m/z 294.8 . 81.1, and hexadiline m/z 277.7 . 83.9 had mass spectrometer parameters set; dwell times of 180, 420, and 180 milliseconds, declustering potentials of 40, 25, and 10 V, entrance potentials of 9, 20, and 11 V, collision entrance potentials of 15, 16, and 25 V, collision energies of 40, 47, and 35 V, and collision exit potentials of 5, 0, and 5 V, respectively. The valco valve was position to send the first 0–0.7 minutes of eluent to waste, then 1.2–2.8 minutes to the mass spectrometer, and 2.8–4.0 minutes to waste.
Assay Validation
The assay was validated using “in-house” calibrator and QC material spanning the concentration range 0.05–3.00 mg/L for both analytes. Calibration curves were determined using peak area ratio (analyte peak area/IS peak area) versus analyte concentration with a regression weighting of 1/x (1/analyte concentration). Imprecision and inaccuracy were determined by analyzing each calibrator and QC within run (n = 6) and between run (n = 8) (Table 1, Table 2). To evaluate assay recovery and matrix effects, experiments were performed based on Matuszewski et al11; drug-free plasma (n = 6) was (1) spiked with each individual analyte (perhexiline and cis-hydroxy-perhexiline) at concentrations of 3.00 and 0.05 mg/L and extracted, (2) drug-free plasma was extracted and the supernatant spiked at the same concentrations, and (3) a stock solution of each of the analytes was prepared at the concentrations being tested. The analyte peak areas were compared, and the assay parameters are shown in Table 3. Assay parameters for perhexiline and cis-hydroxy-perhexiline were tested in the presence of hexadiline calculating a peak area ratio. An example of the chromatography is demonstrated in Figure 1 showing calibrators at concentrations of 3.00 and
RESULTS The assay was linear for the concentration range 0.05–3.00 mg/L for both analytes, and calibration curves demonstrated a mean correlation coefficient of r = 0.997 for perhexiline and r = 0.997 for cis-hydroxy-perhexiline for 8 consecutive assays. The lower limit of detection of the assay was determined, and both analytes showed that there was no calculated perhexiline or cis-hydroxyperhexiline when perhexiline-free plasma (n = 6) was extracted as per protocol. The overall efficiency, extraction efficiency, and matrix effects were 92.3%, 91.1%, and +2.3%, and 118.5%, 113.8%, and +5.9% for perhexiline at concentrations of 3.0 and 0.05 mg/L. Similarly, cis-hydroxy-perhexiline demonstrated parameters of 120.8%, 107.7%, and +12.4% and 195.2%, 141.9%, and +39.6% at concentrations of 3.0 and 0.05 mg/L, respectively. Cardiac patient samples (n = 8) undergoing alternative drug therapies (not administered perhexiline) were tested for matrix effects and assay recovery as described above. Medication that the cardiac patients were being administered included digoxin, metoprolol, spironolactone, warfarin, aspirin, isosorbide dinitrite, frusemide, salbutamol, atorvastatin, paracetamol, benzylpenicillin, metformin, esomeprazole, and estradiol valerate. All calibrators and QCs demonstrated acceptable within- and between-run imprecision and inaccuracy as shown in Table 1 for calibrators and Table 2 for QCs. The limit of quantitation of the assay was 0.03 mg/L for perhexiline demonstrating a percentage of coefficient of variation (CV%) of 20.2% and bias of 20.6% during between-run anaylsis and 0.02 mg/L for cis-hydroxy-perhexiline demonstrating a CV% of 7.9% and bias of 20.3% during within-run
TABLE 3. Assay Parameters for Perhexiline, Cis-Hydroxy-Perhexiline, and Hexadiline (IS) at Concentrations of 3 and 0.05 mg/L in Drug-Free Human Plasma (n = 6) and Plasma From Patients (n = 8) Taking Commonly Coadministered Medications
Concentration, mg/L Overall efficiency, % Extraction efficiency, % Matrix effect, %
Perhexiline DrugFree Plasma (n = 6)
Perhexiline Patient Plasma (n = 8)
3.0 73.6 68.1 24.7
3.0 70.7 85.9 217.9
0.05 118.5 64.1 24.7
0.05 57.9 85.3 228.2
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
cis-Hydroxy-Perhexiline Drug-Free Plasma (n = 6) 3.0 77.5 65.8 18.7
0.05 102.2 83.3 26.6
cis-Hydroxy-Perhexiline Patient Plasma (n = 8) 3.0 68.9 100.6 230.8
0.05 55.9 89.6 236.6
Hexadiline Patient Plasma (n = 8) 0.2 86.0 99.9 212.9
823
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Westley et al
Ther Drug Monit Volume 37, Number 6, December 2015
FIGURE 1. Chromatography of calibrator samples with perhexiline (A: 3 mg/L, B: 0.05 mg/L, and C: 0 mg/L), cis-hydroxyperhexiline (D: 3 mg/L, E: 0.05 mg/L, and F: 0 mg/L), and the internal standard hexadiline (G, H, and I: 0.34 mg/L) with retention times of 2.4, 2.0, and 2.4 minutes, respectively.
anaylsis. However, for patient sample reporting the assay was established to span the range 0.05 to 3.00 mg/L for both analytes, which equates to the lowest calibrator being 3-fold lower than the lower limit of the therapeutic range. Assay performance was investigated using samples received as part of a QAP enabling a comparison with method means and previously reported QAP samples, assayed by a liquid–liquid extraction using fluorescence (FL) detection,8 to the newly developed method. Deming regression analysis of the new mass spectrometry method to previously reported QAP samples (n = 57) using the routine FL detection method8 (our result) and QAP method means demonstrated perhexiline = 1.28 QAP + 0.01 (Sy.x = 0.08, 1/slope = 0.78), perhexiline = 1.07 FL + 0.02 (Sy.x = 0.08, 1/slope = 0.93); cis-hydroxy-perhexiline = 1.12 QAP 2 0.03 (Sy.x = 0.21, 1/slope = 0.89), hydroxy-perhexiline = 1.06 Fl 2 0.01 (Sy.x = 0.16, 1/slope = 0.94). When comparing routine TDM samples (n = 45) by FL and mass spectrometry methods,
Deming regression analysis demonstrated perhexiline = 1.11 FL 2 0.00 (Sy.x = 0.05, 1/slope = 0.91); cis-hydroxyperhexiline = 1.78 FL 2 0.22 (Sy.x = 0.35, 1/slope = 0.56). Combining all samples measured by the 2 methods (n = 102), the comparison demonstrated perhexiline = 1.20 FL + 0.00 (Sy.x = 0.08, 1/slope = 0.67); cis-hydroxy-perhexiline = 1.48 FL 2 0.20 (Sy.x = 0.40, 1/slope = 0.67) (Fig. 2). The measure of uncertainty was calculated for a 2-month period during the evaluation of the assay. QC samples at concentrations of 0.15, 0.75, and 1.50 mg/L demonstrated an uncertainty of 0.02, 0.07, and 0.18 mg/L, respectively, for perhexiline and 0.02, 0.10, and 0.20 mg/L, respectively, for cis-hydroxy-perhexiline.
824
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DISCUSSION The method described in this article has been validated over the perhexiline and cis-hydroxy-perhexiline concentration
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Validation of HPLC-MS/MS Assay for Perhexiline and Metabolite
at a range of 0.05–3.00 mg/L. The range is comparable with the previously published FL detection method used in our laboratory,8 however, using a reduced sample volume (50 versus 500 mL), reduced sample run time (4 versus 15 minutes), and a protein precipitation extraction, enabled samples to be extracted, analyzed, and reported within the same day. This is a vast improvement on the labor intensive liquid–liquid extraction derivatization method previously reported8 that typically involved a 24-hour turnaround time. The method shows acceptable precision and accuracy over the concentration range, acceptable assay extraction efficiencies, and excellent comparison with routine patient samples and QAP samples analyzed by the FL detection method. The FL method did trend higher when analyzing the cis-hydroxy-perhexiline concentrations, which may have been a result of chromatographic separation. The peak of interest eluted near the solvent front and may have resulted in an overestimation of the cis-hydroxyperhexiline concentrations because of interference in patient samples. The mass spectrometry detection allowed for greater specificity and faster run times. Assay parameters were considered to be satisfactory when measuring perhexiline/hexadiline peak area ratios at concentrations of 3.00 and 0.05 mg/L; however, when using peak area ratio for cis-hydroxy-perhexiline, there appeared to be ion enhancement when extracted together. This suggests that hexadiline may not be the best internal standard for calculating cis-hydroxy-perhexiline concentrations. As the metabolite is primarily determined for the identification of poor metabolizers, ion enhancement was not a concern in this case as there was an excellent correlation
between patient and QAP samples. Nordoxepin has been used in another extraction procedure12 as an internal standard and may be considered if ion enhancement was a cause of concern. In blank samples, a peak was seen that may be the result of IS contamination with the analyte or cross talk between the analytes of interest. It was not considered to be significant as it was less than 5% of the peak area of the lowest calibrator. Reducing the internal standard concentration used in the assay may lower or remove this peak if it was a concern to assay performance. Samples from cardiac patients being treated with other medications showed that the assay parameters did not appear to be affected by their presence as the method demonstrated an excellent correlation between MS/MS and FL methods in cardiac patients and QAP samples. Comparing the method described in this article with the other HPLC-MS13 and HPLC-MS/MS12 methods available, it demonstrated a reduced sample volume required (50 versus 200 mL13) and reduced sample analysis time (4 versus 20 minutes13) and was comparable in specificity and sensitivity to Zhang et al.12 Longitudinal assay performance of QC samples has been monitored and is consisted with the measure of uncertainty values described above during assay validation that equates to approximately a potential 10% error during routine analysis at the QC concentrations described. This should be taken into consideration when validating the assay as multiple users of the method introduce another variable for potential error. The assay has therefore shown a robustness that is acceptable for routine analysis and performance.
FIGURE 2. Deming regression analysis of perhexiline: (A) QAP sample comparison (n = 57), (B) QAP method mean comparison (n = 57), (C) patient sample comparison (n = 45), and (D) all samples (n = 102) measured by MS and FL; cis-hydroxy-perhexiline MS versus FL, (E) QAP sample comparison (n = 57), (F) QAP method mean comparison (n = 57), (G) patient sample comparison (n = 45), and (H) all samples (n = 102) measured by HPLC-MS/MS, and FL. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
825
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Westley et al
The validated method has been successfully integrated into the TDM service available at our institution providing an accurate and specific method for the quantitation of perhexiline and cis-hydroxy-perhexiline concentrations in human plasma. REFERENCES 1. Armstrong ML. A comparative study of perhexiline, beta-adrenergic blocking agents and placebos in the management of angina pectoris. Postgrad Med J. 1973;49(suppl 3):108–111. 2. Shah RR, Oates NS, Idle JR, et al. Impaired oxidation of debrisoquine in patients with perhexiline neuropathy. Br Med J. 1982;284:295–299. 3. Morgan MY, Reshef R, Shah RR, et al. Impaired oxidation of debrisoquine in patients with perhexiline liver injury. Gut. 1984;25:1057–1064. 4. Sallustio BC, Westley IS, Morris RG. Pharmacokinetics of the antianginal agent perhexiline: relationship between metabolic ratio and steady state dose. Br J Clin Pharmacol. 2002;54:107–114. 5. Horowitz JD, Sia STB, Macdonald PS, et al. Perhexiline maleate for severe angina pectoris—correlations with pharmacokinetics. Int J Cardiol. 1986;13:219–229. 6. Cole PL, Beamer AD, McGowan N, et al. Efficacy and safety of perhexiline maleate in refractory angina. A double-blind placebo controlled clinical trial of a novel antianginal agent. Circulation. 1980;81:1260–1270. 7. Davies BJ, Coller JK, James HM, et al. The influence of CYP2D6 genotype on trough plasma perhexiline and cis-OH-perhexiline concentrations
826
8.
9.
10.
11.
12.
13.
following a standard loading regimen in patients with myocardial ischaemia. Br J Clin Pharmacol. 2006;61:321–325. Morris RG, Sallustio BC, Saccoia NC, et al. Application of an improved HPLC perhexiline assay to human plasma specimens. J Liq Chromatogr. 1992;15:3219–3232. Davies BJ, Herbert MK, Culbert JA, et al. Enantioselective assay for the determination of perhexiline enantiomers in human plasma by liquid chromatography. J Chromatogr B Analyt Technol Biomed Life Sci. 2006;832:114–120. Davies BJ, Herbert MK, Coller JK, et al. Determination of the 4-monohydroxy metabolites of perhexiline in human plasma, urine and liver microsomes by liquid chromatography. J Chromatogr B Analyt Technol Biomed Life Sci. 2006;843:302–309. Matuszewski BK, Constanzer ML, Chavez-Eng CM. Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS. Anal Chem. 2003;75:3019–3030. Zhang M, Moor GA, Barclay ML, et al. Determination of perhexiline and its metabolite hydroxyperhexiline in human plasma by liquid chromatography/tandem mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci. 2009;877:3025–3030. Beck O, Stephanson N, Morris RG, et al. Determination of perhexiline and hydroxyperhexiline in plasma by liquid chromatography-mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci. 2004;805:87–91.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Validation of a High-Performance Liquid Chromatography–Tandem Mass Spectrometry Method for the Determination of Perhexiline and Cis-Hydroxy-Perhexiline Plasma Concentrations Ian S. Westley, PhD,*† Giovanni Licari, PhD,‡ and Benedetta C. Sallustio, PhD*§
Background: The polymorphic nature of cytochrome P450 2D6 has made therapeutic drug monitoring of the anti-anginal agent perhexiline a compulsory step in reducing adverse events associated with plasma concentrations above the therapeutic range (0.15–0.60 mg/L). The aim of this study was to develop a highperformance liquid chromatography–mass spectrometry/mass spectrometry method for the determination of plasma perhexiline concentrations and its major metabolite cis-hydroxy-perhexiline to reduce sample extraction procedures and improve sample turnaround times.
Methods: The method was validated by determining the precision and accuracy of calibrators and quality control material, comparing quality assurance program samples and patient samples measured by a previously reported liquid–liquid extraction fluorescence (FL) detection high-performance liquid chromatography method and performing matrix effects investigations. Results: Replicates of calibrators at concentrations of 3.00 and 0.05 mg/L demonstrated imprecision of ,10.8% and inaccuracy of ,8.2% for perhexiline and ,10.1% and ,4.5% for cishydroxy-perhexiline, respectively. All samples measured by the 2 methods (n = 102) demonstrated Deming regression of perhexiline = 1.20 FL + 0.00 (Sy.x = 0.08, 1/slope = 0.67); cis-hydroxyperhexiline = 1.48 FL 2 0.20 (Sy.x = 0.40, 1/slope = 0.67). Conclusions: The assay performance was deemed acceptable and integrated into the routine therapeutic drug monitoring program of the department. Key Words: perhexiline, HPLC-MS/MS, validation (Ther Drug Monit 2015;37:821–826)
Received for publication June 9, 2014; accepted March 2, 2015. From the *Department of Clinical Pharmacology, Basil Hetzel Institute, The Queen Elizabeth Hospital, Woodville; †School of Pharmacy and Biomedical Science, University of South Australia; ‡Vascular Diseases and Therapeutics Research Group, Basil Hetzel Institute, The Queen Elizabeth Hospital; and §Discipline of Pharmacology, University of Adelaide, Australia. The authors declare no conflict of interest. Correspondence: Ian S. Westley, PhD, Department of Clinical Pharmacology, Basil Hetzel Institute, The Queen Elizabeth Hospital, 28 Woodville Rd, Woodville 5011, South Australia, Australia (e-mail: [email protected]). Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Ther Drug Monit Volume 37, Number 6, December 2015
INTRODUCTION
The anti-anginal agent perhexiline was first introduced in the 1970s for the treatment of angina1; however, its use declined because of the occurrence of adverse events such as neuropathy and hepatotoxicity.2,3 Perhexiline is primarily metabolized by cytochrome P450 2D6, which is subject to genetic polymorphisms that results in approximately 10% of the population having significantly reduced clearance (poor metabolisers).4 Therapeutic drug monitoring (TDM) has become beneficial to maintain perhexiline plasma concentrations in the range 0.15–0.60 mg/L to avoid adverse events.5,6 Using TDM and determining the concentration of the main metabolite, cis-hydroxy-perhexiline, patients can be phenotyped into poor, extensive, and ultra rapid metabolisers of perhexiline based on their individual metabolite to parent concentration ratio. Although poor metabolisers require doses between 50 and 100 mg once a week, doses for the other phenotypic groups range from 50 to 500 mg daily.7 It is therefore beneficial for patients to use resources like TDM to provide optimal therapy. Previously, our laboratory used a complex highperformance liquid chromatography (HPLC) method to determine perhexiline plasma concentrations that incorporated multiple liquid–liquid extractions and derivatization for fluorescent detection.8 There have been other fluorescent detection methods published for the separation of the perhexiline enationmers in human plasma9 and its metabolites in plasma, urine, and human liver microsomes,10 but using mass spectrometry detection can provide better specificity and allow simplified sample extraction procedures. The aim of this study was to validate a simple rapid analytical method for the determination of perhexiline and cis-hydroxyperhexiline to reduce sample extraction procedures and improve specimen turnaround times.
METHODS Materials Perhexiline maleate was purchased from Sigma Chemical Corporation (MO); cis-hydroxy-perhexiline and hexadiline hydrogen chloride were a gift from Marion Merrell Dow Inc (OH); ammonium acetate was purchased from VWR International (Leuven, Belgium), and formic acid was purchased from Merck KGaA (Darmstadt, Germany). Methanol, acetonitrile, and
821
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Westley et al
TABLE 1. Imprecision (n = 6) and Inaccuracy (n = 8) Data for Perhexiline and cis-Hydroxy-Perhexiline at Calibrator Concentrations of 3 and 0.05 mg/L Within Run (%) (n = 6)
Within Run (%) (n = 6)
Between Run (%) (n = 8)
Between Run (%) (n = 8)
3 mg/L
0.05 mg/L
3 mg/L
0.05 mg/L
6.3 3.0
8.8 28.2
4.1 24.5
10.8 26.9
4.3 2.7
8.2 4.5
6.6 3.0
10.1 21.6
Perhexiline-Spiked Concentration, mg/L Imprecision Inaccuracy cis-hydroxy-perhexiline–spiked concentration, mg/L Imprecision Inaccuracy
hydrochloric acid were purchased from Thermo Fisher Scientific Pty Ltd (Victoria, Australia). Water (.18.2 MV) was prepared using a Cascada AN-water system (Pall Life Sciences, NY).
Stock Solutions Perhexiline stock solution (100 mg/L) was prepared in 25 mL by dissolving 7.09 mg of perhexiline maleate in methanol (5 mL, 20% of final volume) and making up to volume with 0.1 mol/L hydrochloric acid. Cis-hydroxyperhexiline (100 mg/L) and hexadiline (14 mg/L) were prepared as above and stocks stored at 2–88C.
Calibrators
The method was validated using “in-house” calibrators and controls. Calibrators and quality control (QC) samples were prepared from separate stock solutions and spiked into drug-free human plasma. Calibrators (n = 6) were prepared at concentrations of 0.05, 0.25, 0.50, 1.00, 2.00, and 3.00 mg/L for perhexiline and cis-hydroxy-perhexiline and QCs spiked with weighed-in concentrations of 0.15, 0.75, and 1.50 mg/L for both analytes. Aliquots of calibrators and QCs were stored at 2208C until use.
Patient and Quality Assurance Program Samples Patient samples (n = 45) were analyzed for perhexiline and cis-hydroxy-perhexiline as part of routine TDM by the
previously described liquid–liquid extraction method and assayed on the same day by the HPLC-MS/MS method described in this article. Quality assurance program (QAP) samples (n = 57) were collated and assayed as part of the international external QAP (managed by our laboratory) by the previously described liquid–liquid extraction method and by the HPLCMS/MS method described in this article. QAP samples were a mixture of pooled patient samples and weighed in perhexiline/cis-hydroxy-perhexiline concentrations.
Sample Preparation
Calibrator, QC, or patient samples (50 mL) were aliquoted into clean 1.5 mL centrifuge tubes. Acetonitrile (150 mL) containing the internal standard (IS) hexadiline (0.2 mg/L) was added and tubes mixed and then centrifuged at 16,000g for 5 minutes. Supernatant (50 mL) was transferred to a clean 1.5 mL centrifuge containing 0.05% formic acid (100 mL) and mixed. The contents were transferred into HPLC vials and analytes separated using the chromatographic conditions described below.
Chromatographic Conditions The HPLC system consisted of an Agilent 1100 series (Agilent Technologies, CA) pump, degasser, injector, and column oven. Twenty-five microliters of preparation was injected, and analytes were separated using a Luna 2.1 · 50 mm (5 mm) Phenyl-hexyl column (Phenomonex Inc,
TABLE 2. Imprecision (n = 6) and Inaccuracy (n = 8) Data for Perhexiline and cis-Hydroxy-Perhexiline at QC Samples at Concentrations of 0.15, 0.75, and 1.50 mg/L
Perhexiline-Spiked QC Concentration, mg/L Imprecision Inaccuracy cis-hydroxy-perhexiline– spiked QC concentration, mg/L Imprecision Inaccuracy
822
Within Run (%) (n = 6)
Within Run (%) (n = 6)
Within Run (%) (n = 6)
Between Run (%) (n = 8)
Between Run (%) (n = 8)
Between Run (%) (n = 8)
0.15 mg/L
0.75 mg/L
1.50 mg/L
0.15 mg/L
0.75 mg/L
1.50 mg/L
3.0 11.7
3.3 22.0
7.0 24.8
13.4 23.6
8.7 27.4
9.1 214.8
5.1 24.6
3.0 28.3
6.6 24.8
14.2 27.78.7
14.3 28.2
15.0 215.0
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Validation of HPLC-MS/MS Assay for Perhexiline and Metabolite
CA) maintained at 508C. The mobile phases consisted of A: 100% methanol, 2 mmol/L ammonium acetate, 0.1% formic acid, B: 100% water, 2 mmol/L ammonium acetate, 0.1% formic acid. The initial mobile phase consisted of methanol: water (50:50 vol/vol A:B) pumped at a flow rate of 300 mL/ min and increased to 80% methanol (A) over 0.7 minutes. Analytes were eluted using 80% methanol (A) between 0.7 and 2.8 minutes, then returning to methanol:water (50:50 vol/ vol A:B) with a total run time of 4 minutes.
0.05 mg/L, respectively, a patient sample with a perhexiline concentration of 0.40 mg/L and cis-hydroxy-perhexiline concentration of 1.50 mg/L, and a zero sample. Analytes eluted with times of 2.0, 2.4, and 2.4 minutes for cis-hydroxyperhexiline, perhexiline, and hexadiline, respectively. On instrument, sample stability was assessed by extracting a calibration curve and QC samples as per protocol and analyzed. The extracts were kept on the instrument for a 24-hour period and reinjected, data analyzed, and calibration curve established. QC sample concentrations calculated from the reinjected samples remained in the accepted ranges of 62 SDs of the mean perhexiline and cis-hydroxyperhexiline concentrations determined during imprecision and inaccuracy experiments.
Mass Spectrometry Conditions Analytes were detected using multiple reaction monitoring in positive ionization mode on an API2000 mass spectrometer (AB SCIEX Pty Ltd, MA). Perhexiline m/z 278.3 . 81.1, cis-hydroxy-perhexiline m/z 294.8 . 81.1, and hexadiline m/z 277.7 . 83.9 had mass spectrometer parameters set; dwell times of 180, 420, and 180 milliseconds, declustering potentials of 40, 25, and 10 V, entrance potentials of 9, 20, and 11 V, collision entrance potentials of 15, 16, and 25 V, collision energies of 40, 47, and 35 V, and collision exit potentials of 5, 0, and 5 V, respectively. The valco valve was position to send the first 0–0.7 minutes of eluent to waste, then 1.2–2.8 minutes to the mass spectrometer, and 2.8–4.0 minutes to waste.
Assay Validation
The assay was validated using “in-house” calibrator and QC material spanning the concentration range 0.05–3.00 mg/L for both analytes. Calibration curves were determined using peak area ratio (analyte peak area/IS peak area) versus analyte concentration with a regression weighting of 1/x (1/analyte concentration). Imprecision and inaccuracy were determined by analyzing each calibrator and QC within run (n = 6) and between run (n = 8) (Table 1, Table 2). To evaluate assay recovery and matrix effects, experiments were performed based on Matuszewski et al11; drug-free plasma (n = 6) was (1) spiked with each individual analyte (perhexiline and cis-hydroxy-perhexiline) at concentrations of 3.00 and 0.05 mg/L and extracted, (2) drug-free plasma was extracted and the supernatant spiked at the same concentrations, and (3) a stock solution of each of the analytes was prepared at the concentrations being tested. The analyte peak areas were compared, and the assay parameters are shown in Table 3. Assay parameters for perhexiline and cis-hydroxy-perhexiline were tested in the presence of hexadiline calculating a peak area ratio. An example of the chromatography is demonstrated in Figure 1 showing calibrators at concentrations of 3.00 and
RESULTS The assay was linear for the concentration range 0.05–3.00 mg/L for both analytes, and calibration curves demonstrated a mean correlation coefficient of r = 0.997 for perhexiline and r = 0.997 for cis-hydroxy-perhexiline for 8 consecutive assays. The lower limit of detection of the assay was determined, and both analytes showed that there was no calculated perhexiline or cis-hydroxyperhexiline when perhexiline-free plasma (n = 6) was extracted as per protocol. The overall efficiency, extraction efficiency, and matrix effects were 92.3%, 91.1%, and +2.3%, and 118.5%, 113.8%, and +5.9% for perhexiline at concentrations of 3.0 and 0.05 mg/L. Similarly, cis-hydroxy-perhexiline demonstrated parameters of 120.8%, 107.7%, and +12.4% and 195.2%, 141.9%, and +39.6% at concentrations of 3.0 and 0.05 mg/L, respectively. Cardiac patient samples (n = 8) undergoing alternative drug therapies (not administered perhexiline) were tested for matrix effects and assay recovery as described above. Medication that the cardiac patients were being administered included digoxin, metoprolol, spironolactone, warfarin, aspirin, isosorbide dinitrite, frusemide, salbutamol, atorvastatin, paracetamol, benzylpenicillin, metformin, esomeprazole, and estradiol valerate. All calibrators and QCs demonstrated acceptable within- and between-run imprecision and inaccuracy as shown in Table 1 for calibrators and Table 2 for QCs. The limit of quantitation of the assay was 0.03 mg/L for perhexiline demonstrating a percentage of coefficient of variation (CV%) of 20.2% and bias of 20.6% during between-run anaylsis and 0.02 mg/L for cis-hydroxy-perhexiline demonstrating a CV% of 7.9% and bias of 20.3% during within-run
TABLE 3. Assay Parameters for Perhexiline, Cis-Hydroxy-Perhexiline, and Hexadiline (IS) at Concentrations of 3 and 0.05 mg/L in Drug-Free Human Plasma (n = 6) and Plasma From Patients (n = 8) Taking Commonly Coadministered Medications
Concentration, mg/L Overall efficiency, % Extraction efficiency, % Matrix effect, %
Perhexiline DrugFree Plasma (n = 6)
Perhexiline Patient Plasma (n = 8)
3.0 73.6 68.1 24.7
3.0 70.7 85.9 217.9
0.05 118.5 64.1 24.7
0.05 57.9 85.3 228.2
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
cis-Hydroxy-Perhexiline Drug-Free Plasma (n = 6) 3.0 77.5 65.8 18.7
0.05 102.2 83.3 26.6
cis-Hydroxy-Perhexiline Patient Plasma (n = 8) 3.0 68.9 100.6 230.8
0.05 55.9 89.6 236.6
Hexadiline Patient Plasma (n = 8) 0.2 86.0 99.9 212.9
823
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Westley et al
Ther Drug Monit Volume 37, Number 6, December 2015
FIGURE 1. Chromatography of calibrator samples with perhexiline (A: 3 mg/L, B: 0.05 mg/L, and C: 0 mg/L), cis-hydroxyperhexiline (D: 3 mg/L, E: 0.05 mg/L, and F: 0 mg/L), and the internal standard hexadiline (G, H, and I: 0.34 mg/L) with retention times of 2.4, 2.0, and 2.4 minutes, respectively.
anaylsis. However, for patient sample reporting the assay was established to span the range 0.05 to 3.00 mg/L for both analytes, which equates to the lowest calibrator being 3-fold lower than the lower limit of the therapeutic range. Assay performance was investigated using samples received as part of a QAP enabling a comparison with method means and previously reported QAP samples, assayed by a liquid–liquid extraction using fluorescence (FL) detection,8 to the newly developed method. Deming regression analysis of the new mass spectrometry method to previously reported QAP samples (n = 57) using the routine FL detection method8 (our result) and QAP method means demonstrated perhexiline = 1.28 QAP + 0.01 (Sy.x = 0.08, 1/slope = 0.78), perhexiline = 1.07 FL + 0.02 (Sy.x = 0.08, 1/slope = 0.93); cis-hydroxy-perhexiline = 1.12 QAP 2 0.03 (Sy.x = 0.21, 1/slope = 0.89), hydroxy-perhexiline = 1.06 Fl 2 0.01 (Sy.x = 0.16, 1/slope = 0.94). When comparing routine TDM samples (n = 45) by FL and mass spectrometry methods,
Deming regression analysis demonstrated perhexiline = 1.11 FL 2 0.00 (Sy.x = 0.05, 1/slope = 0.91); cis-hydroxyperhexiline = 1.78 FL 2 0.22 (Sy.x = 0.35, 1/slope = 0.56). Combining all samples measured by the 2 methods (n = 102), the comparison demonstrated perhexiline = 1.20 FL + 0.00 (Sy.x = 0.08, 1/slope = 0.67); cis-hydroxy-perhexiline = 1.48 FL 2 0.20 (Sy.x = 0.40, 1/slope = 0.67) (Fig. 2). The measure of uncertainty was calculated for a 2-month period during the evaluation of the assay. QC samples at concentrations of 0.15, 0.75, and 1.50 mg/L demonstrated an uncertainty of 0.02, 0.07, and 0.18 mg/L, respectively, for perhexiline and 0.02, 0.10, and 0.20 mg/L, respectively, for cis-hydroxy-perhexiline.
824
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
DISCUSSION The method described in this article has been validated over the perhexiline and cis-hydroxy-perhexiline concentration
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Validation of HPLC-MS/MS Assay for Perhexiline and Metabolite
at a range of 0.05–3.00 mg/L. The range is comparable with the previously published FL detection method used in our laboratory,8 however, using a reduced sample volume (50 versus 500 mL), reduced sample run time (4 versus 15 minutes), and a protein precipitation extraction, enabled samples to be extracted, analyzed, and reported within the same day. This is a vast improvement on the labor intensive liquid–liquid extraction derivatization method previously reported8 that typically involved a 24-hour turnaround time. The method shows acceptable precision and accuracy over the concentration range, acceptable assay extraction efficiencies, and excellent comparison with routine patient samples and QAP samples analyzed by the FL detection method. The FL method did trend higher when analyzing the cis-hydroxy-perhexiline concentrations, which may have been a result of chromatographic separation. The peak of interest eluted near the solvent front and may have resulted in an overestimation of the cis-hydroxyperhexiline concentrations because of interference in patient samples. The mass spectrometry detection allowed for greater specificity and faster run times. Assay parameters were considered to be satisfactory when measuring perhexiline/hexadiline peak area ratios at concentrations of 3.00 and 0.05 mg/L; however, when using peak area ratio for cis-hydroxy-perhexiline, there appeared to be ion enhancement when extracted together. This suggests that hexadiline may not be the best internal standard for calculating cis-hydroxy-perhexiline concentrations. As the metabolite is primarily determined for the identification of poor metabolizers, ion enhancement was not a concern in this case as there was an excellent correlation
between patient and QAP samples. Nordoxepin has been used in another extraction procedure12 as an internal standard and may be considered if ion enhancement was a cause of concern. In blank samples, a peak was seen that may be the result of IS contamination with the analyte or cross talk between the analytes of interest. It was not considered to be significant as it was less than 5% of the peak area of the lowest calibrator. Reducing the internal standard concentration used in the assay may lower or remove this peak if it was a concern to assay performance. Samples from cardiac patients being treated with other medications showed that the assay parameters did not appear to be affected by their presence as the method demonstrated an excellent correlation between MS/MS and FL methods in cardiac patients and QAP samples. Comparing the method described in this article with the other HPLC-MS13 and HPLC-MS/MS12 methods available, it demonstrated a reduced sample volume required (50 versus 200 mL13) and reduced sample analysis time (4 versus 20 minutes13) and was comparable in specificity and sensitivity to Zhang et al.12 Longitudinal assay performance of QC samples has been monitored and is consisted with the measure of uncertainty values described above during assay validation that equates to approximately a potential 10% error during routine analysis at the QC concentrations described. This should be taken into consideration when validating the assay as multiple users of the method introduce another variable for potential error. The assay has therefore shown a robustness that is acceptable for routine analysis and performance.
FIGURE 2. Deming regression analysis of perhexiline: (A) QAP sample comparison (n = 57), (B) QAP method mean comparison (n = 57), (C) patient sample comparison (n = 45), and (D) all samples (n = 102) measured by MS and FL; cis-hydroxy-perhexiline MS versus FL, (E) QAP sample comparison (n = 57), (F) QAP method mean comparison (n = 57), (G) patient sample comparison (n = 45), and (H) all samples (n = 102) measured by HPLC-MS/MS, and FL. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
825
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Ther Drug Monit Volume 37, Number 6, December 2015
Westley et al
The validated method has been successfully integrated into the TDM service available at our institution providing an accurate and specific method for the quantitation of perhexiline and cis-hydroxy-perhexiline concentrations in human plasma. REFERENCES 1. Armstrong ML. A comparative study of perhexiline, beta-adrenergic blocking agents and placebos in the management of angina pectoris. Postgrad Med J. 1973;49(suppl 3):108–111. 2. Shah RR, Oates NS, Idle JR, et al. Impaired oxidation of debrisoquine in patients with perhexiline neuropathy. Br Med J. 1982;284:295–299. 3. Morgan MY, Reshef R, Shah RR, et al. Impaired oxidation of debrisoquine in patients with perhexiline liver injury. Gut. 1984;25:1057–1064. 4. Sallustio BC, Westley IS, Morris RG. Pharmacokinetics of the antianginal agent perhexiline: relationship between metabolic ratio and steady state dose. Br J Clin Pharmacol. 2002;54:107–114. 5. Horowitz JD, Sia STB, Macdonald PS, et al. Perhexiline maleate for severe angina pectoris—correlations with pharmacokinetics. Int J Cardiol. 1986;13:219–229. 6. Cole PL, Beamer AD, McGowan N, et al. Efficacy and safety of perhexiline maleate in refractory angina. A double-blind placebo controlled clinical trial of a novel antianginal agent. Circulation. 1980;81:1260–1270. 7. Davies BJ, Coller JK, James HM, et al. The influence of CYP2D6 genotype on trough plasma perhexiline and cis-OH-perhexiline concentrations
826
8.
9.
10.
11.
12.
13.
following a standard loading regimen in patients with myocardial ischaemia. Br J Clin Pharmacol. 2006;61:321–325. Morris RG, Sallustio BC, Saccoia NC, et al. Application of an improved HPLC perhexiline assay to human plasma specimens. J Liq Chromatogr. 1992;15:3219–3232. Davies BJ, Herbert MK, Culbert JA, et al. Enantioselective assay for the determination of perhexiline enantiomers in human plasma by liquid chromatography. J Chromatogr B Analyt Technol Biomed Life Sci. 2006;832:114–120. Davies BJ, Herbert MK, Coller JK, et al. Determination of the 4-monohydroxy metabolites of perhexiline in human plasma, urine and liver microsomes by liquid chromatography. J Chromatogr B Analyt Technol Biomed Life Sci. 2006;843:302–309. Matuszewski BK, Constanzer ML, Chavez-Eng CM. Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS. Anal Chem. 2003;75:3019–3030. Zhang M, Moor GA, Barclay ML, et al. Determination of perhexiline and its metabolite hydroxyperhexiline in human plasma by liquid chromatography/tandem mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci. 2009;877:3025–3030. Beck O, Stephanson N, Morris RG, et al. Determination of perhexiline and hydroxyperhexiline in plasma by liquid chromatography-mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci. 2004;805:87–91.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
