Anal. Chem. 2003, 75, 2538-2542
Separation and Quantitation of the Stereoisomers of Ephedra Alkaloids in Natural Health Products Using Flow Injection-Electrospray Ionization-High Field Asymmetric Waveform Ion Mobility Spectrometry-Mass Spectrometry Margaret McCooeye, Luyi Ding, Graeme J. Gardner, Catharine A. Fraser, Joe Lam, Ralph E. Sturgeon, and Zolta´n Mester*
Institute for National Measurement Standards, National Research Council of Canada, Ottawa, ON, Canada, K1A 0R6
A method is described for the determination of ephedrine (E) and pseudoephedrine (PE) and their metabolites norephedrine (NE), norpseudoephedrine (NPE), methylephedrine (ME), and methylpseudoephedrine (MPE) alkaloids in natural health products by flow injectionelectrospray ionization-high field asymmetric waveform ion mobility spectrometry-mass spectrometry (FI-ESIFAIMS-MS). The determination of the six alkaloids requires the separation of diastereomic pairs of E-PE, NENPE, and ME-MPE. FAIMS was able to resolve/separate these isomeric pairs based on their gas-phase ion mobility differences. The FAIMS-based separation and detection approach has been tested on over-the-counter diet pills. Following the extraction of the tablets, either by pressurized fluid extraction developed in-house or with sonication, the ephedra alkaloids were quantified using a modified isotope dilution approach. Detection limits for the alkaloids ranged from 0.1 to 3 ng/mL, and a linear range of at least 2 orders of magnitude was observed for the six analytes. The throughput of the current configuration of the FI-ESI-FAIMS-MS system is 2 min/sample, which is significantly higher than conventional chromatographic approaches. The developed FI-ESI-FAIMS-MS method has been compared with a conventional LC-UV analysis, and good agreement has been found for the major alkaloids.
Traditional medicines, natural health products, dietary supplements, and over-the-counter remedies containing ephedra alkaloids are widely available in North America. The herb Ephedra sinica or Ma Huang is a common source of two physiologically active compounds, ephedrine (E) and pseudoephedrine (PE), as well as their metabolites, norephedrine (NE), norpseudoephedrine (NPE), methylephedrine (ME), and methylpseudoephedrine (MPE). Reports of adverse effects of these alkaloids, including stroke, heart attacks, heart rate irregularities, seizures, psychoses, and deaths, have resulted in a voluntary recall of some products containing ephedra/ephedrine in Canada.1 The U.S. Food and 2538 Analytical Chemistry, Vol. 75, No. 11, June 1, 2003
Drug Administration is considering regulations to reduce the potential negative effects of ephedrine-type alkaloids,2 and the International Olympic Committee has put ephedrine on their list of banned substances.3 Ephedra alkaloids, primarily pseudoephedrine, are also used as starting materials for clandestine laboratories producing methamphetamine (speed) for the illegal drug trade.4 Increasing concern on the part of both consumers and regulatory agencies as to the safety of ephedra alkaloids has led to the development of several methods to measure these analytes in a range of complex matrixes. Gas chromatographic (GC) methods have been described for determination of E and PE in plant material5 and for E, PE, NE, NPE, and ME in urine.6,7 Betz et al. have reported a method for chiral analysis of all six alkaloids in dietary supplements using GC with a flame ionization detector.8 Capillary electrophoresis has been used by by Iwata et al.9 to achieve the separation of nine ephedrine- and amphetamine-type stimulants; however, no matrix was considered in these studies. Six alkaloids of interest, E, PE, NE, NPE, ME, and MPE, have been separated in a urine matrix10 using capillary zone electrophoresis. High-performance liquid chromatography has been used with diode array detection to separate and quantify ephedrinetype alkaloids in plant material11 and herbal products12 and with (1) Health Canada Website: www.hc.sc.gc.ca/english/protection/warnings/ 2002. (2) Gay, M. L.; White, K. D.; Obermeyer, W. R.; Betz, J. M.; Musser, S. M. J. AOAC Int. 2001, 84, 761. (3) International Olympic Committee, World Anti-Doping Agency, website: www.wada-ama.org. (4) Andrews, K. M. J. Forensic Sci. 1995, 40, 551. (5) Li, H.-X.; Ding, M.-Y.; Lv, K.; Yu, J.-Y. J. Chromatogr. Sci. 2001, 39, 370. (6) Eenoo, P. V.; Delbeke, F. T.; Roels, K.; Backer, P. J. Chromatogr., B 2001, 760, 255. (7) Spyridaki, M.-H. E.; Tsitsimpikou, C. J.; Siskos, P. A.; Georgakopoulos, C. G. J. Chromatogr., B 2001, 758, 311. (8) Betz, J. M.; Gay, M. L.; Mossoba, M. M.; Adams, S. J. AOAC Int. 1997, 80, 303. (9) Iwata, Y. T.; Garcia, A.; Ksnsmoti, T.; Inoue, H.; Kishi, T.; Ira, S. Electrophoresis 2002 2002, 23, 1328. (10) Chicharro, M.; Zapardiel, A.; Bermejo, E.; Perez-Lopez, J. A.; Hernandez, L. J. Liq. Chromatogr. 1995, 18, 1363. (11) Li, H.-X.; Ding, M.-Y.; Ly, K.; Yu, J.-Y. J. Liq. Chromatogr., Relat. Technol. 2002, 25, 313. 10.1021/ac0342020 CCC: $25.00 Published 2003 Am. Chem. Soc. Published on Web 05/03/2003
spectroscopy15 but has not been applied to stereoisomers in a natural matrix. A method based on high-field asymmetric waveform ion mobility spectrometry (FAIMS) has been applied to the determination of amphetamine analogues16 as well as morphine and codeine in urine.17 This report describes the application of an isotope dilution (ID) scheme combined with FAIMS separation, ESI, and MS detection to the separation and quantitation of three pairs of ephedrine-type diastereoisomers, E and PE, NE and NPE, and ME and MPE, in extracts of natural health products. The method is validated by comparison with a conventional LC-UV analysis of the same extracts.
Figure 1. (a) Ephedrine enantiomers and (b) pseudoephedrine enantiomers.
UV detection to determine these alkaloids in traditional medicines.13 A method employing liquid chromatography with mass spectrometric detection has also been developed by Gay et al.2 for the analysis of E, PE, NE, NPE, ME, and MPE in dietary supplements. Recently, the AOAC International has adopted an HPLC-UV method for the determination of ephedra alkaloids in dietary supplements, botanicals, and neutraceuticals.14 Ephedrine (2-methylamino-1-phenylpropanol) has two stereogenic carbons and can therefore exist as four stereoisomers. Two of these stereoisomers, ephedrine and pseudoephedrine, are optically active and must have enantiomers with identical physical properties and opposite specific rotation. The structures of these four stereoisomeric combinations, commonly referred to as diastereoisomers, are shown in Figure 1. Either of the ephedrine enantiomers is a diastereoisomer of either of the pseudoephedrine enantiomers. However, only one pair of these diastereoisomers, (-)-ephedrine and (+)-pseudoephedrine, as well as their metabolites, (-)-norephedrine, (+)-norpseudoephedrine, (-)-methylephedrine, and (+)-methylpseudoephedrine, are found in plant sources such as Ma Huang.2 Since each pair of diastereoisomers has the same mass-to-charge ratio (m/z) and fragmentation pattern, they cannot be individually determined by conventional electrospray ionization-mass spectrometry (ESI-MS) or ESI-MSMS unless preceded by a separation step, i.e., liquid chromatography (LC), GC, capillary zone electrophoresis, etc. Isomerresolved ion spectroscopy has been achieved by coupling ion mobility spectroscopy, mass spectrometry, and photoelectron (12) Hurlbut, J. A.; Carr, J. R. J. AOAC Int. 1998, 81, 1121. (13) Okamura, N.; et al. J. Pharm. Biomed. Anal 1999, 20, 363. (14) Castor, T. P. Association of Official Analytical Chemists website: www.aoac.org/vmeth/newsmth.htm.
EXPERIMENTAL SECTION Instrumentation and Methods. A mixture of diet pills available in natural health food outlets and containing ephedra was used in this study. Samples were subjected to either a pressurized fluid extraction (PFE) developed in-house or solvent extraction with sonication.13 For PFE, a Dionex ASE 200 (Oakville, ON, Canada) instrument was used. Approximately 2.5 g of sample was weighed dispersed on Ottawa sand and packed in a 22-mL PFE extraction cell with two cellulose filters on the bottom. The extraction solvent was high-purity deionized water (DDW, 18 MΩ cm), obtained from a NanoPure system (Barnstead/Thermolyne, Boston, MA) fed with a reverse osmosis supply line, to which 3% methanol was added. Three static cycles were implemented at 90 °C and 1500 psi. Heating time was 5 min, and flush volume was 150%. The eluent was brought up to 100 mL with extracting solvent and weighed. For extraction using sonication, ∼2.5 g of sample was weighed and 100 mL of extracting solvent, 3% methanol in DDW, was added to the flask and weighed. The flask was shaken for 15 min and then placed in an ultrasonic bath for 15 min at room temperature. Extracts were centrifuged to remove suspended particles and aliquots were taken for ID-ESI-FAIMS-MS analysis. For LC analysis, standard addition spikes were added for quantitation as appropriate and samples were cleaned up on 500-mg Strata SCX SPE cartridges (Phenomenex, Torrance, CA). All samples were analyzed using both ID-ESI-FAIMS-MS and LCUV methods. The FAIMS device used in this work is a prototype Ionalytics Selectra (Ionalytics, Ottawa, Canada). FAIMS has been described in detail previously.18 Briefly, the FAIMS consists of two axially symmetric steel cylinders mounted inside a PEEK holder and fastened to the orifice plate of a PE Sciex API300 mass spectrometer. A high-voltage asymmetric waveform supplies 4000 V, referred to as the dispersion voltage, to the inner cylinder of the FAIMS device. The effect of using positive versus negative dispersion voltages to achieve the separation of the ephedra alkaloids was examined. A dc voltage, referred to as the compensation voltage (CV), was also supplied to the inner cylinder of the device to overcome the tendency of an ion to drift toward one electrode under the influence of the alternating high and low (15) Fromherz, R.; Gantefor, G.; Shvartsburg, A. Phys. Rev. Lett. 2002, 89, 083001 1. (16) McCooeye, M.; Mester, Z.; Ells, B.; Barnett, D. A.; Randy W, P.; Guevremont, R. Anal. Chem. 2002, 74, 3071. (17) McCooeye, M. A.; Ells, B.; Barnett, D. A.; Purves, R. W.; Guevremont, R. J. Anal. Toxicol. 2001, 25, 81. (18) Guevremont, R.; Barnett, D. A.; Purves, R. W.; Vandermey, J. Anal. Chem. 2000, 72, 4577.
Analytical Chemistry, Vol. 75, No. 11, June 1, 2003
2539
electric fields supplied by the asymmetric waveform. The CV reflects the difference in ion mobility at high and low electric fields for a particular ion and could either be set to a specific voltage or scanned over a voltage range. The magnitude of the CV depends on properties of both the ion and the curtain/carrier gas.19 A dc bias voltage of 20 V was applied to both the outer cylinder of the FAIMS analyzer and the orifice plate of the mass spectrometer. The ESI needle was placed ∼1 cm from the orifice of the FAIMS device at an angle of ∼45°. A bias voltage of +4700 V was applied to the ESI needle, producing a current of 55-65 nA. For optimization and evaluation of experimental conditions, a sample solution was delivered to the ESI needle by a Harvard Apparatus model 22 syringe pump (South Natick, MA) from a 250-µL syringe (Hamilton, Reno, NE) at a flow rate of 1 µL/min. Analyses were performed using an Agilent 1100 pump and autosampler (Agilent Technologies Inc.,Palo Alto, CA). The pump was operated at 0.5 mL/min, and 70-µL samples were injected and split ∼100/1 before delivery to the electrospray needle. The FAIMS apparatus was operated at atmospheric pressure and room temperature. Nitrogen served as a combination curtain/carrier gas and was passed through a charcoal/molecular sieve filter before entering the FAIMS apparatus at a flow rate of 2.5 L/min. A stock standard solution of 10 µg/mL was prepared for each analyte in methanol and diluted as required in a buffer solution (i.e., 0.25 mM ammonium acetate in methanol/DDW (90/10)). To determine the best conditions for transmission through the FAIMS device, standard solutions of each analyte were used. For quantitative measurements with the ESI-FAIMS-MS instrumentation, a modified isotope dilution scheme was devised. Two solutions of deuterated ephedrine hydrochloride in methanol were prepared for use as isotope-labeled spiking solutions. The more concentrated solution, 361 µg/g deuterated ephedrine in methanol, was used to determine ephedrine and pseudoephedrine in sample extracts. A more dilute spiking solution, 7 µg/g deuterated ephedrine in methanol, was used for the determination of norephedrine, norpseudoephedrine, methylephedrine, and methylpseudoephedrine in sample extracts. Aliquots of standard solutions and sample extracts were weighed, and a weighed aliquot of the appropriate isotope-labeled spiking solution was added to each. The mixtures of standard solutions and labeled ephedrine were diluted with methanol/DDW (90/ 10) containing 0.2 mM ammonium acetate, and the mixtures of sample extracts and labeled ephedrine were diluted either 200fold with methanol or 1000-fold with methanol/DDW (90/10) containing 0.1 mM ammonium acetate. Dilution factors of 1000fold were used for samples analyzed for ephedrine and pseudoephedrine and 200-fold for the nor and methyl metabolites. An Agilent Technologies 1100 autosampler (Agilent Technologies Inc.) was used to inject 70 µL of each sample into a flowing stream of buffer delivered by the LC pump. The stream was split ∼100/1 to provide a flow of ∼5 µL/min to the electrospray needle. The software of the mass spectrometer allowed the optimum CV for each analyte to be scanned with its m/z. For each injection of sample or standard solution, the protonated molecular ion (M + H) of at least two of the analytes of interest and the deuterated ephedrine ion (m/z 169) was monitored at the CV (19) Barnett, D. A.; Ells, B.; Guevremont, R.; Purves, R. W.; Viehland, L. A. J. Am. Soc. Mass Spectrom. 2000, 11, 1125.
2540
Analytical Chemistry, Vol. 75, No. 11, June 1, 2003
for optimum transmission. Transient peaks were integrated using IGOR Pro software (Wavemetrics, Lake Oswego, OR). Concentrations were calculated as follows:
concentration ) RsmMcalibCcalibMspike(sample)/(RcmMsampleMspike(calib))
where Rsm is the ratio of signal from native compound/signal from labeled compound in the sample solution, Rcm is the ratio of signal from native compound/signal from labeled compound in the calibration solution, Mcalib is the weight of calibration solution taken for analysis (g), Ccalib is the concentration of native compound in the calibration solution (mg/g), Mspike(sample) is the weight of isotopically labeled spiking solution added to sample solution (g), Mspike(calib) is the weight of isotopically labeled spiking solution added to calibration solution (g), and Msample is the weight of sample material taken for analysis (g). LC analyses were performed on an Agilent Technologies 1100 LC system equipped with a diode array detector and autosampler. The column used was a 4.6 mm × 150 mm Phenomenex SynergiPolarRP column with 4-µm particle size. Analytes were eluted with methanol/DDW, 100 mM potassium phosphate buffer, 3/97, v/v, running isocratically for 16 min. The column was then rinsed with acetonitrile/DDW/phosphoric acid, 50/50/0.1, for 10 min and reconditioned with the mobile phase for 10 min before injection of the next sample. This elution program was designed to remove from the column any caffeine that may be present in the samples, thus preventing interference from the late-eluting peak. The diode array detector was set to monitor 194 (bandwidth 4) and 210 nm (bandwidth 8). Peak areas were integrated manually, and the areas were exported to an Excel spreadsheet for further processing. Reagents. Ephedrine and pseudoephedrine were obtained from Sigma-Aldrich Canada Ltd. (Mississauga, ON, Canada). Norephedrine, methylephedrine, and methylpseudoephedrine were obtained from Chromadex (Santa Ana, CA). L-Norpseudoephedrine was purchased as a solution, having a concentration of 1 mg/mL in methanol, from Alltech Associates (State College, PA). Glass-distilled HPLC grade methanol and acetonitrile (Anachemia, Montreal, PQ, Canada) were used as received. Phosphoric acid was also supplied by Anachemia. Industrial-grade nitrogen and helium were supplied by Air Products Canada Ltd. (Ottawa, ON, Canada). Strata SCX SPE cartridges were obtained from Phenomenex, and Fisher Scientific (Ottawa, ON, Canada) supplied potassium phosphate. RESULTS AND DISCUSSION CVs for optimum transmission of each ion using ESI-FAIMSMS were determined by infusing standard solutions of each analyte and scanning CV over a 10-V range. Figure 2a shows ionselected CV (IS-CV) spectra of a standard solution of 1 ppm each of E, PE, NE, and NPE and 3 ppm ME and MPE in 9/1 methanol/ DDW containing 0.2 mM ammonium acetate. The curtain/carrier gas was nitrogen at 2.5 L/min, and dispersion voltage was set to +4000 V. For each analyte, the protonated molecular ion (M + H) was monitored. Table 1 lists the target analytes, m/z monitored, and CV for optimal transmission of each analyte. Figure 2b shows the IS-CV spectra for the same solution of
Figure 3. Flow injection peaks showing five injections of spiked standard solution followed by five injections of sample B diluted 200fold. The upper trace is E-d3, m/z 169, monitored at CV ) -4.4, and the lower trace is NE, m/z 152, monitored at CV ) -4.8. The traces are offset for clarity.
Figure 2. IS-CV spectra for a solution of 1 µg/mL E, PE, NE, and NPE and 3 µg/mL ME and MPE in methanol/DDW (90/10) with 0.2 mM ammonium acetate added. (a) P1 mode, DV ) 4000 V, and carrier gas is 2.5 L/min nitrogen. (b) P2 mode, DV ) -4000 V, and carrier gas is 1.25 L/min nitrogen + 1.4 L/min helium. Table 1. Characteristics of the FI-ESI-FAIMS-MS Method for Ephedrine Analysis analyte
m/z (M + H)
CV (V)
est LOD (ng/mL)
RSD (%)
ephedrine pseudoephedrine norephedrine norpseudoephedrine methylephedrine methylpseudoephedrine
166 166 152 152 180 180
-4.47 -5.76 -4.84 -6.25 -2.46 -2.89
0.1 0.1 0.2 2 1 3
3.0, n ) 5 2.2, n ) 5 2.0, n ) 5 7.2, n ) 4 12.5, n ) 5 2.1, n ) 4
analytes with dispersion voltage set to -4000 V and 52% helium added to the curtain/carrier gas. The enhanced sensitivity achieved using negative dispersion voltage is obvious, especially for ME and MPE. However, baseline separation of ME and MPE is not achieved under these conditions and NE and NPE are not separated. Increasing the helium level in the carrier gas caused the CV peaks to broaden further. The IS-CV spectra obtained using positive dispersion voltage and shown in Figure 2a indicate baseline separation of the three stereoisomeric pairs, E-PE, NENPE, and ME-MPE, as well as adequate sensitivity for the determination of these compounds in natural health products. All subsequent quantitative analyses were conducted using +4000-V dispersion voltage with the Ionalytics Selectra instrument. Estimates of linear range and limits of detection for the ESI-FAIMSMS analysis of ephedrine-type alkaloids were made by analyzing standard solutions of the six target analytes with concentrations
of each analyte ranging from 1 ng/mL to 1 µg/mL in 9/1 methanol/water with 0.2 mM ammonium acetate added. Detection limits based on three times the standard deviation of the background signal are given in Table 1, and a linear range of at least 2 orders of magnitude was observed for the six analytes. Figure 3 shows a set of flow injection peaks showing five injections of a 5.5 µg/g NE standard solution in methanol/DDW, 90/10, containing 0.2 mM ammonium acetate spiked with deuterated ephedrine, followed by five injections of a sample extract spiked with deuterated ephedrine. The standard solution was diluted 200-fold with buffer and the sample with methanol prior to injection. The FAIMS device was operated at CV ) -4.8 V monitoring m/z 152 for norephedrine and CV ) -4.4 monitoring m/z 169 for deuterated ephedrine with 2.5 L/min nitrogen curtain/ carrier gas. The lower trace is the norephedrine signal, and the upper trace is deuterated ephedrine signal. Note that 10 analyses are accomplished in ∼25 min. Concentrations of E, PE, NE, NPE, ME, and MPE determined by ESI-FAIMS-MS and LC-UV in two samples are given in Table 2. For ESI-FAIMS-MS analysis, the M + H ion for each analyte was monitored at the CV appropriate for optimum transmission given in Table 1. Sample A was extracted using the PFE method, and sample B was solvent extracted as described previously. For LC-UV determinations, 20 µL of cleaned up sample was injected onto the column and eluted as described previously. Signals were monitored at 194 and 210 nm. An LC chromatogram resulting from the injection of unspiked sample B is shown in Figure 4. The inset provides an expansion of the axes. The two analytical methods are in excellent agreement when applied to the major analytes (E and PE) in the extracts, and in good agreement for NPE and ME. However, they differ significantly when applied to MPE and NE. The FI-ESI-FAIMS-MS results are 4-5-fold higher than the LC-UV results for MPE and 15-30% lower for NE. This is probably explained by the very poor resolution of the MPE and NE peaks in the LC chromatogram, as is evident in Figure 4. Figure 5 illustrates the clear advantage of using FAIMS separation technology with mass spectrometric detection. Panels a and b of Figure 5 show mass spectra collected with the FAIMS device installed and CV set at (a) -2.89 for the determination of MPE Analytical Chemistry, Vol. 75, No. 11, June 1, 2003
2541
Table 2. Comparison of FI-ESI-FAIMS-MS and LC-UV Determinations of Ephedra Alkaloids (mg/g) sample A
ephedrine pseudoephedrine norephedrine norpseudoephedrine methylephedrine methylpseudoephedrine
sample B
ESI-FAIMS-MS
LC-UV
ESI-FAIMS-MS
LC-UV
5.01 ( 0.14 1.21 ( 0.04 0.062 ( 0.003 0.0821 ( 0.0059 0.133 ( 0.019 0.0318 ( 0.0007
4.94 ( 0.03 1.24 ( 0.02 0.0816 ( 0.0019 0.0930 ( 0.0027 0.124 ( 0.003 0.0086 ( 0.0008
10.6, 11.3 2.42, 2.44 0.14 ( 0.01 0.188, 0.189 0.30 ( 0.03 0.0687, 0.0669
10.97 ( 0.02 2.44 ( 0.01 0.168 ( 0.0002 0.198 ( 0.001 0.298 ( 0.002 0.0144 ( 0.004
Figure 4. LC chromatogram of sample B. The inset is an expansion of the baseline.
and (b) -2.46 for the determination of ME in sample B diluted 1/200 with methanol. Note the improvement in signal-to-noise ratio from 4 in Figure 5c to 28 for ME in Figure 5a and 47 for MPE in Figure 5b. To produce Figure 5c, the FAIMS device was removed and the ESI-MS system was operated using the same parameters and with the same sample as in Figure 5a and b. Note the difference in scale. In Figure 5c, it is very difficult to differentiate m/z 180, which represents the sum of ME and MPE in this case, from the intense background signal.
CONCLUSION This is the first reported application of FAIMS for the separation of diastereoisomers. Results obtained using FI-ESIFAIMS-MS to separate and measure ephedra alkaloids in weightloss products agree well with those acquired using the more conventional LC-UV method. The FI-ESI-FAIMS-MS approach has the advantage of requiring no cleanup step prior to analysis, and elimination of the column separation dramatically reduces the time required for analysis. The FI-ESI-FAIMS-MS has been
2542 Analytical Chemistry, Vol. 75, No. 11, June 1, 2003
Figure 5. Mass spectra of sample A diluted 200-fold: (a) ESIFAIMS-MS at DV ) 4000 V, CV ) -2.89; (b) ESI-FAIMS-MS at DV ) 4000 V, CV ) -2.46; (c) ESI-MS.
shown to be rugged, reliable, and amenable to use with an autosampler. Received for review February 27, 2003. Accepted April 15, 2003. AC0342020
