1174
Chem. Res. Toxicol. 2002, 15, 1174-1178
Evaluation of Commercial Ginkgo and Echinacea Dietary Supplements for Colchicine Using Liquid Chromatography-Tandem Mass Spectrometry Wenkui Li, Yongkai Sun, John F. Fitzloff, and Richard B. van Breemen* Department of Medicinal Chemistry and Pharmacognosy, UIC/NIH Center for Botanical Dietary Supplements Research, College of Pharmacy, University of Illinois at Chicago, 833 South Wood Street, Chicago, Illinois 60612 Received May 16, 2002
In response to concerns that commercial dietary supplements containing Ginkgo biloba (ginkgo) and Echinacea purpurea, Echinacea angustifolia, or Echinacea pallida (echinacea) might be contaminated with colchicine, a highly selective and sensitive assay was developed for colchicine that is based on high-performance liquid chromatography-tandem mass spectrometry (LC-MS-MS). The method utilizes reversed-phase HPLC separation of compounds in a methanolic extract of the dietary supplement or botanical sample followed by positive ion electrospray ionization with collision-induced dissociation and multiple reaction monitoring of three characteristic fragmentation pathways of the protonated molecule of colchicine, m/z 400 f 358, 400 f 326, and 400 f 310. The minimal detectable concentration of colchicine using this assay was 10 pg on-column, which is equivalent to 20 ppb colchicine in a 0.5 g gingko leaf sample. The method was validated by analyzing 0.5 g samples spiked with colchicine and determining the recovery. A total of 26 commercial ginkgo and echinacea dietary supplements were purchased from pharmacies in Chicago, IL, and analyzed for colchicine. In contrast to a recent report, no colchicine was detected in any of the samples. In addition, authenticated gingko leaves were collected, assayed, and found to contain no colchicine, which is consistent with the botanical literature. On the basis of the results obtained using this new LC-MS-MS assay, which is more sensitive and more selective than previously published methods for colchicine, we find no cause for concern regarding colchicine contamination of gingko or echinacea dietary supplements.
Introduction Colchicine, (-)-N-5,6,7,9-tetrahydro-1,2,3,10-tetramethoxy-9-oxobenzo[a]heptalen-7-yl (S)-acetamide (Figure 1), is an alkaloid occurring in plants of the Liliaceae including Colchicum autumnale (autumn crocus) and Gloriosa superba (glory lily). Clinically, colchicine has been used for the treatment of several diseases including acute gouty arthritis, familial Mediterranean fever, and inflammatory diseases such as scleroderma (1, 2). Recently, colchicine has been demonstrated to be a potent inhibitor of cellular mitosis by interfering with microtubule polymerization (3). Colchicine poisonings are rare and begin with symptoms such as nausea, vomiting, and diarrhea, and are followed by multisystem organ failure. Doses of colchicine exceeding 0.8 mg/kg are usually fatal (4-12). Preparations of the leaves of Ginkgo biloba (gingko) have been used widely for the treatment of cerebrovascular insufficiency and peripheral circulatory problems and for slowing down the progression of cognitive deficits in Alzheimer’s disease. Echinacea purpurea, Echinacea angustifolia, and Echinacea pallida (echinacea) are used as immunostimulants and antioxidants for human health. Both ginkgo and echinacea are among the top 10 most popular dietary supplements in the United States (13). Recently, Petty et al. (14) reported detecting * To whom correspondence should be addressed. Telephone: (312) 996-9353. Fax: (312) 996-7107. E-mail:
[email protected].
Figure 1. Chemical structure and proposed fragmentation pattern of colchicine.
colchicine in commercial dietary supplements of ginkgo (26 ( 3 µg/tablet) and echinacea (2.0 ( 0.5 µg/tablet) and in the placental blood of pregnant women taking ginkgo supplements (49-763 µg/L). However, this report has been criticized for not disclosing the commercial products that were tested and for not confirming the results of the HPLC-UV assay by using an independent method (15). Since ginkgo and echinacea are not known to biosynthesize colchicine or any of its congeners (16), the paper by Petty raises the concern that colchicine contamination might be a general problem in commercial dietary supplements containing ginkgo and echinacea. To address this concern, we developed a highly sensitive and selective analytical method based on high-performance liquid chromatography-tandem mass spectrometry (LCMS-MS) and applied it to the analysis of 26 commercially
10.1021/tx025559m CCC: $22.00 © 2002 American Chemical Society Published on Web 08/29/2002
Ginkgo and Echinacea Dietary Supplements
Chem. Res. Toxicol., Vol. 15, No. 9, 2002 1175
Table 1. Ginkgo biloba Dietary Supplements Tested for Colchicine 1 2 3 4 5 6 7 8 9
product
lot no.
amount/dose
dosage
Ginkgo-Memo Ginkoba Finest Natural-Ginkgo biloba Nature’s Reward Ginkogin Natrol-Ginkgo biloba Sundown-Ginkgo biloba Sundown-Ginkgo biloba Sundown-Ginkgo biloba Nature’s Resource
JJ10559 FD0534A 6764-3 CG912-682B 941583 9915340204 2780240404 2432400404 LB11565N-01-04
40 mg extract 40 mg extract, 0.8 mg leaf 60 mg extract 60 mg extract 120 mg extract 40 mg extract, 400 mg leaf 60 mg extract 60 mg extract 60 mg extract
1 tablet 1 tablet 1 tablet 1 caplet 2 capsules 2 capsules 1 tablet 1 tablet 2 capsules
Table 2. Echinacea Dietary Supplements Tested for Colchicine product
lot no.
amount/dose
dosage
1 2 3 4 5 6 7 8
Sav-on-Osco by Alberson’s XTRA ResistEx Sav-on OSCO Vitasmart Nature’s Bounty Natrol Nature’s Valley
1FA1457 455697-12-02 OHN0152 ONA0492 OLN0181 45650-03 943595 9A00856
3 capsules 1 capsule 2 capsules 3 capsules 2 caplets 1 capsule 1 capsule 1 caplet
9 10 11 12 13 14 15 16 17
L’il Criters Sundown Finest Natural Nature’s Bounty L’il Critters Nature’s Bounty SYMTEC Essential solutions Nature’ sources
10267 983180204 6878/4 14119 N/A 14213 396M1 103236 LA12278N
Echinacea purpurea herb 750 mg, Goldenseal root 600 mg E. purpurea aerial parts extract 50 mg Echinacea purpurea root extract E. purpurea herb 1140 mg E. purpurea aerial parts extract 334 mg E. purpurea root 400 mg E. purpurea aerial parts extract 100 mg Echinacea angustifolia and E. purpurea root extract 75 mg, Goldenseal root extract 100 mg Extract blend 100 mg E. purpurea aerial parts 400 mg E. purpurea leaf extract 125 mg E. purpurea and E angustifolia 500 mg E. angustifolia and E. purpurea extract 120 mg Echinacea angustifolia, E. purpurea, Goldenseal 500 mg E. angustifolia root 120 mg E. purpurea extract 1000 mg Echinacea aerial parts 350 mg
available gingko and echinacea dietary supplements for the presence of colchicine.
Materials and Methods Materials. All organic solvents and chemicals were of HPLC grade (Fisher Scientific Co., Fair Lawn, NJ). A colchicine standard was obtained from the repository of natural product isolates at our own laboratories, and its identity and purity were verified using LC-MS-MS. Deionized water was generated using an in-house Nano-pure water system (Barnstead, Newton, MA). Authentic Gingko biloba leaves were collected at the University of Illinois at Chicago during September 2001, dried at room temperature, and powdered. Nine ginkgo and 17 echinacea dietary supplements were purchased from local pharmacies in Chicago, IL. The brand names, lot numbers, dosages, and product types (tablet, capsule, etc.) are listed in Tables 1 and 2 for gingko and echinacea, respectively. Sample Preparation. A stock solution of colchicine (0.333 mg/mL) was prepared in methanol and diluted to obtain working standard solutions of 0.1-100 ng/mL. All working solutions were stored at -20 °C and brought to room temperature immediately before to use. Note that the colchicine standard solutions and the dietary supplement extracts (see below) were prepared in separate laboratories to minimize the risk of contamination of the dietary supplements. As another precaution, only new glassware was used for the preparation and storage of each sample. One or more capsules, tablets, or caplets representing a single dose of a ginkgo or echinacea dietary supplement (as indicated on the product label) was weighed and placed into a 20-mL PTFE capped sample vial. Each sample was extracted twice with 15 mL of 80% methanol using a sonicator for 60 min at room temperature followed by filtration through Whatman no. 40 filter paper into a 250-mL round-bottom flask. While on the filter paper, the solid material was washed three times with 15 mL portions of 80% methanol. The methanol extracts were combined and evaporated to dryness in vacuo at 45-50 °C. The residue was redissolved in methanol (4 × 2 mL), transferred to a 10mL volumetric flask, and made up to volume with methanol.
2 tablets 1 capsule 1 tablet 1 mL 1 mL 1 mL 5 mL 1 ampule 1 capsule
Each extract was filtered through a disposable 0.2 µm nylon membrane prior to LC-MS-MS analysis. For liquid formulations of echinacea, 1 mL of the sample was diluted to 2-10 mL with water. Duplicate sample solutions were prepared and analyzed for each product. For method validation, 0.5 g aliquots of powdered ginkgo leaves were spiked in triplicate with 75, 100, 175, or 400 ng of the colchicine standard in methanol. After thorough mixing, each sample was air-dried overnight at room temperature. The next day, each spiked gingko leaf powder was extracted with methanol as described above. After the solvent was removed in vacuo, the extract was reconstituted in 10 mL of methanol and analyzed for colchicine using LC-MS-MS as described below. By comparing the measured and theoretical concentrations, the recovery rates (%) and the coefficients of variance were calculated for each concentration. Solvent blanks and method blanks were also prepared to test for the possibility of contamination by colchicines standards. Mass Spectrometry. Accurate mass measurements and high-resolution tandem mass spectra of colchicine were obtained using a Micromass (Manchester, U.K.) Q-Tof-2 hybrid mass spectrometer with an electrospray ion source operated at 80 °C in positive ionization mode. Colchicine samples were introduced using infusion. Following the selection of the protonated molecule of m/z 400 in the first quadrupole, collision-induced dissociation (CID) was carried out using argon as the collision gas in the second quadrupole collision chamber at a pressure of 4 × 10-7 mbar and collision energy of 25 eV. The highresolution product ion mass spectrum was obtained using the TOF in reflectron mode. LC-MS-MS experiments were carried out using a Micromass Quattro II triple quadrupole mass spectrometer equipped with a Waters (Milford, MA) Alliance 2690 HPLC system and an electrospray ion source. A Waters Xterra MS C18 column (2.1 × 100 mm, 3.5 µm) was used for HPLC separations with a mobile phase consisting of a 9 min linear gradient from 10 to 70% aqueous methanol at a flow rate of 0.2 mL/min. The injection volume was 10 µL. The entire column effluent was introduced directly into the electrospray ionization source without solvent splitting.
1176
Chem. Res. Toxicol., Vol. 15, No. 9, 2002
Figure 2. Positive ion electrospray collision-induced dissociation (CID) tandem mass spectrum of the protonated molecule of colchicine. The electrospray ion source in the Quattro II triple quadrupole mass spectrometer was operated at 140 °C in positive ion mode. Nitrogen was used as both nebulizing and drying gas at a flow rate of 20 and 450 L/h, respectively. The protonated molecule of colchicine at m/z 400 was selected using the first quadrupole, collision-induced dissociation (CID) was carried out using argon as the collision gas at 1.0 × 10-3 mbar and 25 eV in the second quadrupole, and then specific product ions were selected using the third quadrupole. During these multiple reaction monitoring (MRM) measurements, the dwell time was 0.2 s for each of the precursor/product ion pairs. Both the first and last quadrupoles were operated at unit mass resolution.
Li et al.
Figure 3. LC-MS-MS positive ion electrospray CID MRM chromatograms of 100 pg colchicine. The value listed under each precursor/product ion pair refers to the peak height. In the text, the ratios for these three precursor/product ion pairs were calculated using peak areas and therefore differ slightly from these peak height ratios.
Results and Discussion During positive ion electrospray mass spectrometry, colchicine formed an abundant protonated molecule of m/z 400. Since no abundant fragment ions were observed, the protonated molecule was fragmented using CID to obtain the product ion tandem mass spectrum shown in Figure 2. To achieve maximum sensitivity, the most abundant fragment ions were selected for use with multiple reaction monitoring (MRM) during LC-MS-MS. The electrospray and CID parameters were optimized for the ion pairs of m/z 400 f 358, m/z 400 f 326, and m/z 400 f 310. Monitoring several precursor/product ion pairs during MRM provides higher selectivity than recording the signal for only one ion or one precursor/ product ion pair as in selected ion monitoring or selected reaction monitoring, respectively. In particular, identification of colchicine using this approach relies on the simultaneous chromatographic appearance of three fragment ions formed from the precursor ion of m/z 400, and these fragment ions should be detected in a ratio consistent with that of the colchicine standard. For example, the colchicine standard eluted at 7.4 min during the LCMS-MS analysis shown in Figure 3, and the ratio of the abundances of the three precursor/product ion pairs was 0.60:0.65:1.00. Furthermore, the product ions of m/z 358, 326, and 310 represent structurally significant side chains of colchicine corresponding to [MH - CH2CO]+, [MH CH3CONH2-CH3]+, and [MH - CH3CONH2-CH3-CH4]+, respectively (Figure 1). The elimination of COCH2 or CH3CONH2 is characteristic of the acetamide group of colchicine, and the loss of methane or methyl group is indicative of the presence of methoxy groups (Figure 1).
Figure 4. (A) LC-MS-MS calibration curve for colchicine using positive ion electrospray with CID and selected reaction monitoring of the precusor/product ion pair m/z 400 f 358. (B) LCMS-MS chromatogram of 10 pg colchicine at the limit of detection for the precursor/product ion pair m/z 400f358.
The elemental compositions of these fragment ions were confirmed using exact mass measurements. To investigate the potential of this LC-MS-MS method for quantitative analysis, a standard curve was generated based on the transition of m/z 400 f 358. Colchicine standards from 20 to 1000 pg were injected into the LCMS-MS system, and the peak area corresponding to the colchicine signal at 7.4 min was measured for each standard. The standard curve based on these peak areas was linear over the entire range with a correlation coefficient of 0.9994 (Figure 4A). The limit of detection (LOD), defined as a signal-to-noise of 5:1, for the LC-MSMS analysis of colchicine was determined to be 10 pg injected on-column. The limit of quantitation (LOQ),
Ginkgo and Echinacea Dietary Supplements
Chem. Res. Toxicol., Vol. 15, No. 9, 2002 1177
Figure 5. LC-MS-MS chromatogram of a methanolic extract of a commercial ginkgo product (Finest Natural-Ginkgo biloba, Lot no. 6764-3).
Figure 6. Typical HPLC chromatogram of the methanolic extract of a commercial echinacea product (Sav-on-Osco, Lot no. 1FA1457).
Table 3. Colchicine Recovery from Spiked Gingko Leaf Powder (n ) 3)
supplement products or in the solvent blanks. Since these results were negative, no method blanks were analyzed for this set of samples. The following three criteria were used to screen the 26 commercial products for the presence of colchicine. (1) All three MRM channels, m/z 400 f 358, m/z 400 f 326, and m/z 400 f 310, must show peaks at 7.4 min. (2) Each MRM channel must produce a signal-to-noise ratio of at least 5:1. (3) The ratios of these three MRM peaks must match those obtained using the colchicine standard (0.60: 0.65:1.00). Since none of these criteria were met for the LC-MS-MS analysis of any of the ginkgo or echinacea dietary supplements, there was no evidence for the presence of colchicine in any of these samples.
spiked amount (ng)
found (ng)
average recovery (%)
RSD %
75 100 175 400
65.55 101.64 171.38 389.61
87.40 101.64 97.93 97.40
2.28 6.14 1.15 1.14
defined as a signal-to-noise ratio of 10:1, was 20 pg oncolumn. For example, the LC-MS-MS chromatogram showing the precursor/product ion pair of m/z 400 f 358 for a 10 pg colchicine standard injected on-column is shown in Figure 4B. The recoveries of colchicine from gingko leaf powder samples that had been spiked with four different concentrations of colchicine were determined using LC-MSMS. The lowest recovery (87.4%) was obtained for the least concentrated samples (150 ppb), and the recoveries exceeded 97% for samples spiked with g200 ppb of colchicine (Table 3). The relative standard deviation (RSD %) of each set of recovery experiments was calculated and ranged from 1 to 6%. These recovery values and their standard deviations indicated that the extraction method was suitable for the quantitative analysis of botanical samples for colchicine contamination. Although the precision of the assay could have been improved by including an internal standard, the linearity of the standard curve and the RSD values for the reproducibility of the spiked samples were more than adequate for the purpose of this investigation. Next, 26 ginkgo and echinacea dietary supplements purchased from local pharmacies (Tables 1 and 2) were extracted and analyzed using LC-MS-MS. As described in the Materials and Methods, great care was used to prevent contamination of these products by the colchicine standard or standard solutions. The LC-MS-MS MRM chromatograms of all of the ginkgo preparations were similar to each other, and all of the echinacea samples produced MRM chromatograms that were similar to each other. Representative LC-MS-MS MRM chromatograms for one gingko sample and one echinacea sample are shown in Figures 5 and 6, respectively. No colchicine was detected in any of the 9 ginkgo or 17 echinacea dietary
Conclusions Unlike most synthetic drugs, botanical extracts and dietary supplements usually contain hundreds of constituents. For example, ginkgo contains ginkgolides, phenolic compounds, more than 30 flavonoids, tannins, anacardic acids, chlorophylls, and lipids (16). Since many of these compounds contain UV-vis chromophores, HPLC coupled with UV-vis detection alone is inadequate for the identification of a contaminant in a dietary supplement. The complexity of these extracts also suggests that any one assay, such as LC-MS (17), GC with flame ionization detection, or even GC-MS (8), would be insufficient for identification. However, any one of these approaches may be used as an initial screening step, and any positive results would need to be confirmed independently using a second method of identification. Preferably, the second assay would be even more selective than the initial screening method such as the LC-MSMS assay described here. However, a second method might be any of those described above. In addition to using at least two independent assays to identify and confirm the presence of an impurity in a dietary supplement, we recommend that great care be used to avoid contamination by reference standards of the substances being analyzed. For example, the glassware being used to store and prepare the samples must be new or thoroughly cleaned such as through acid washing to eliminate the possibility of contamination.
1178
Chem. Res. Toxicol., Vol. 15, No. 9, 2002
Also, the standards must be weighed on separate analytical balances and in laboratory space physically separated from the samples. Finally, solvent blanks and method (or laboratory) blanks must be analyzed along with each batch of samples as controls for sample carry over or contamination. Without negative results for the analyses of these controls, any positive results for the samples under investigation would be inconclusive. After applying the analytical methods described here, we find no evidence for colchicine contamination of ginko or echinacea dietary supplements in retail stores in the Chicago area.
Acknowledgment. This research was supported by the University of Illinois Functional Foods for Health Program and Grant P50 AT00155 provided jointly by the National Center for Complementary and Alternative Medicine (NCCAM), the Office of Dietary Supplements (ODS), the Office for Research on Women’s Health (ORWH), and the National Institute of General Medicine (NIGMS) of the National Institutes of Health (NIH). The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the NCCAM, ODS, ORWH, NIGMS, or NIH.
References (1) Ellenhorn, M. J., and Barceloux, D. G. (1988) Medical ToxicologyDiagnosis and Treatment of Human Poisoning, Elsevier, Amsterdam. (2) Naidus, R. M., Rodvien, R., and Mielke, C. H. (1977) Colchicine toxicology. Arch. Inern. Med. 137, 394-396. (3) Kirshner, M. W. (1978) Microtubule assembly and nucleation. Int. Rev. Cytol. 54, 1-65. (4) Marray, S., Kramlinger, K., McMichan, J., and Mohr, D. (1987) Acute toxicology after ingestion of colchhicine. Mayo Clin. Proc. 58, 528-532.
Li et al. (5) Hobson, C. H., and Rankin, A. P. (1986) A fatal colchicine overdose. Anaesth. Intensive Care 14, 453-455. (6) Allender, W. J. (1982) Colchicine poisning as a mode of suicide. J. Forensic Sci. 27, 944-947. (7) Luciani, I. (1989) Case review: Fatal IV colchicine injection in a 60-year-old woman. J. Emerg. Nursing 15, 80-82. (8) Clevenger, C. V., August, T. F., and Shaw, L. M. (1991) Colchicine poisoning: report of fatal case with body fluid analysis by GCMS and histopathologic examination of postmortem tissue. J. Anal. Toxicol. 15, 151-154. (9) Sauder, P., Kopferschmitt, J., Jaeger, A., and Mantz, J. M. (1983) Haemodynamic studies in eight cases of acute colchicine poisoning. Hum. Toxicol. 2, 169-173. (10) McIntyre, I. M., Ruszkiewicz, A. R., Crump, K., and Drummer, O. H. (1994) Death following colchicine poisoning. J. Forensic Sci. 39, 280-286. (11) Stapczynski, J. S., and Rothstein, R. J. (1981) Clchicine overdose: report of tow cases and review of the literature. Ann. Emerg. Med. 10, 364-369. (12) Folpini, A., and Furfoni, P. (1995) Colchicine toxicology-clinical features and treatment, massive overdose case report. Clin. Toxicol. 33, 71-77. (13) Mar, C., and Bent, S. (1999) An evidence-based review of the 10 most commonly used herbs. West J. Med. 171, 168-171. (14) Petty, H. R., Fernando, M., Kindzelskii, A. L., Zarewych, B. N., Ksebatti, M. B., Hryhorczuk, L. M., and Mobashery, S. (2001) Identification of colchicine in placental blood from patients using herbal medicine. Chem. Res. Toxicol. 14, 1254-1258. (15) Borman, S. (2001) Toxin reported in supplements. Chem. Eng. News 79, 33-34. (16) Farnsworth, N. R. NAPRALERT database. University of Illinois at Chicago, IL (an on-line database available directly through the University of Illinois at Chicago or through the Scientific and Technical Network (STN) of Chemical Abstracts Services) (last accessed May 1, 2002). (17) Tracqui, A., Kintz, P., Ludes, B., Rouge´, C., Douibi, H., and Mangin, P. (1996) High-performance liquid chromatography coupled to ion spray mass spectrometry for the determination of colchicine at ppb levels in human biofluids. J. Chromatrogr. B 675, 235-242.
TX025559M
