Journal of Chromatography B, 942–943 (2013) 63–69
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Simultaneous detection of 22 toxic plant alkaloids (aconitum alkaloids, solanaceous tropane alkaloids, sophora alkaloids, strychnos alkaloids and colchicine) in human urine and herbal samples using liquid chromatography–tandem mass spectrometry Sau Wah Ng, Chor Kwan Ching ∗ , Albert Yan Wo Chan, Tony Wing Lai Mak Hospital Authority Toxicology Reference Laboratory, Princess Margaret Hospital, Hong Kong
a r t i c l e
i n f o
Article history: Received 2 July 2013 Accepted 14 October 2013 Available online xxx Keywords: Alkaloids Liquid chromatography Tandem mass spectrometry Urine Herb
a b s t r a c t A liquid chromatography–tandem mass spectrometry method for simultaneous detection of 22 toxic plant alkaloids, including aconitum alkaloids and their hydrolyzed products (aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, benzoylaconine, benzoylmesaconine, benzoylhypaconine, deacetylyunaconitine, deacetylcrassicauline A), solanaceous tropane alkaloids (atropine, anisodamine, scopolamine, anisodine), sophora alkaloids (matrine, sophoridine, oxymatrine, cytisine, Nmethylcytisine), strychnos alkaloids (brucine, strychnine) and colchicine, in herbal and urine samples was developed and validated. Following sample preparation by liquid–liquid extraction, chromatographic separation was achieved on Eclipse XDB C8 column. Identification was based on two multiple reaction monitoring transitions and the relative ion intensity. Method selectivity was demonstrated. The limits of detection were 5 ng/mL for all analytes, except 50 ng/mL for cytisine. The herbal matrix effects ranged from 89% to 118%, whereas the urine matrix effects were between 91% and 109% for all analytes except cytisine (57%) and N-methylcytisine (67%). The urine extraction recovery ranged from 74% to 110% for all analytes, except cytisine (15%) and oxymatrine (30%). With the good extraction efficiency of the other major sophora alkaloids, the relatively low extraction recovery of the minor sophora alkaloids cytisine and oxymatrine did not affect identification of sophora alkaloids as a group. Carry-over was minimal at less than 0.1%. The method was successfully applied in analysis of 170 cases of suspected herbal poisoning, with aconitum alkaloids, sophora alkaloids, solanaceous tropane alkaloids, and strychnos alkaloids being detected in 53, 42, 18, and 6 cases, respectively. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Herbal medicine is commonly used worldwide, especially in Chinese communities. Though generally regarded as natural and harmless, some herbs are highly toxic with potentially fatal adverse effects, and there is no lack of herbal poisoning reports in the literature [1–3]. Herbal poisoning can occur as a result of overdose, incorrect processing and preparation, misidentification, contamination, and also in homicidal and suicidal cases. Aconitum alkaloids, solanaceous tropane alkaloids, strychno alkaloids, sophora alkaloids and colchicine, with various pharmacological and toxic effects, are five groups of frequently encountered toxic plant alkaloids in herbal poisoning cases.
∗ Corresponding author. Tel.: +852 29901881. E-mail address: [email protected] (C.K. Ching). 1570-0232/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jchromb.2013.10.020
The highly toxic aconitum alkaloids are present in Aconitum species, which have been widely used in herbal medicine for their analgesic, anti-inflammatory and cardiotonic effects. For example, the Chinese medicines “chuanwu” and “fuzi” correspond to the mother root and daughter root of Aconitum carmichaeli respectively, whereas “caowu” correspond to the main root of Aconitum kusnezoffii [4]. These herbs should be used after adequate processing and prolonged decoction to reduce the toxicity. Clinical features of acute aconite poisoning include weakness, perioral and limb numbness, arrhythmias, hypotension and gastrointestinal disturbances [1]. Tropane alkaloids can be found in many genera of the Solanaceae family like Datura, Atropa, Mandragora etc. They are competitive antagonist of muscarinic receptors and can lead to anticholinergic toxidrome. Chinese medicine-induced anticholinergic poisoning cases due to substitution or contamination of benign herbs with tropane alkaloid-containing herbs have been reported [2]. Strychnine and brucine, found in seeds of Strychnos nux-vomica as well as other Strychnos species, are highly toxic
64
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
alkaloids. Strychnine is mainly used as a pesticide in the western countries; however, seeds of S. nux-vomica, also known as “maqianzi”, are also used as a Chinese herbal medicine in the treatment arthritis and neurological diseases [4]. Severe strychnine poisoning can cause muscle twitching, repeated convulsions, rhabdomyolysis and death. The minimum lethal dose of strychnine in adults is in the range of 50–100 mg [5]. Sophora alkaloids like matrine and sophoridine are present in Sophora species. Some of these species, such as Sophora flavescens (“kushen”), Sophora alopecuroides (“kudougen”), Sophora tonkinensis (“shandougen”), are used in Chinese medicine. Overdose of “kushen” could lead to toxic effects including nausea, vomiting, dizziness, weakness and ataxia [6]. Colchicine, commonly used in the treatment of gout, is an alkaloid present in Colchicum species like Colchicum autumnale. Colchicine has a low therapeutic index, and excessive doses can cause toxic features such as vomiting, abdominal pain, diarrhea, shock, renal failure, multi-organ failure and bone marrow suppression. Fatal cases of colchicine poisoning have been reported [7]. Detection of toxic plant alkaloids in body fluids and herbal samples plays an important role in clinical and forensic toxicology investigation of suspected herbal poisoning cases. Analytical methods of some toxic alkaloids in biological samples using high performance liquid chromatography (HPLC), gas chromatography–mass spectrometry (GC–MS), capillary electrophoresis, liquid chromatography–mass spectrometry (LC–MS), and liquid chromatography–tandem mass spectrometry (LC–MS/MS) have been described [3,8–14]. However, most of these methods detect only one particular group of alkaloids, e.g. aconitum alkaloids, or at most several alkaloids of different classes. Besides, hydrolyzed derivatives of aconitum alkaloids, which can be formed during herb processing or human metabolism, are not covered in many reported methods. Simultaneous detection of a broad range of toxic plant alkaloids of different groups in herbal and biological samples is very useful in investigation of suspected herbal poisonings, especially in cases of which culprit herb is not clear from the history; cases of herb contamination or misidentification; or those with atypical or non-specific presentation. The objective of this study was to develop a sensitive and selective LC–MS/MS method for simultaneous detection of 22 toxic plant alkaloids, including aconitum alkaloids and their hydrolyzed products (aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, benzoylaconine, benzoylmesaconine, benzoylhypaconine, deacetylyunaconitine, deacetylcrassicauline A), solanaceous tropane alkaloids (atropine, anisodamine, scopolamine, anisodine), sophora alkaloids (matrine, sophoridine, oxymatrine, cytisine, N-methylcytisine), strychnos alkaloids (brucine, strychnine) and colchicine, in herbal and urine samples. The qualitative method was validated by assessing selectivity, limit of detection (LOD), matrix effect, extraction recovery, false positive rate and false negative rate at around LOD, carry-over, precision of retention time, precision of relative ion intensity, and stability. The validated method was successfully applied in analysis of real clinical samples of suspected herbal poisonings.
2. Materials and methods 2.1. Chemicals and solvents Standards of aconitine, race-anisodamine (R,S), and hypaconitine were obtained from LKT (St. Paul, MN, USA); anisodine hydrobromide, brucine, cytosine, scopolamine hydrobromide trihydrate and strychnine were purchased from Sigma–Aldrich (St. Louis, MO, USA); atropine was obtained from International Laboratory USA (San Francisco, USA); colchicine was obtained from Alexis Biochemicals (NY, USA); crassicauline A and yunaconitine
were purchased from Herbstandard Inc. (Davis, USA); matrine, oxymatrine and sophoridine were obtained from the National Institute for the Control of Pharmaceuticals and Biological Products (Beijing, China); mesaconitine was obtained from Wako (Richmond, VA, USA); N-methylcytisine was obtained from TRC (Toronto, Canada). Methyllycaconitine citrate, used as an internal standard, was obtained Tocris Bioscience (Bristol, UK). Acetonitrile (ACN) and methanol of HPLC grade were purchased from Merck (Darmstadt, Germany) and RCI-Labscan (Bangkok, Thailand) respectively. Ammonium acetate, ammonium formate, and ammonium hydroxide were obtained from Sigma–Aldrich (St. Louis, MO, USA), International Laboratory USA (San Francisco, USA), and Tedia (USA) respectively. Formic acid and glacial acetic acid were purchased from Fluka (Switzerland). Purified water was provided by Milli-Q water purification system from Millipore. 2.2. Preparation of standards and hydrolyzed derivatives of aconitum alkaloids Stock standard solutions of 1 mg/mL of each alkaloid (aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, atropine, anisodamine, scopolamine, anisodine, matrine, sophoridine, oxymatrine, cytisine, N-methylcytisine, brucine, strychnine, colchicine) were prepared in ACN and stored at −70 ◦ C. Stock internal standard solution of 1 mg/mL methyllycaconitine was prepared in methanol and stored at −70 ◦ C. The stock solutions of aconitine, hypaconitine, mesaconitine, yunaconitine, and crassicauline A were mixed and diluted to a concentration of 20 g/mL with 5 mM ammonium acetate (buffered at pH 7.0). The diluted solution was incubated at 90 ◦ C for three hours to produce deacetylated derivatives containing benzoylaconine, benzoylhypaconine, benzoylmesaconine, deacetylyunaconitine, and deacetylcrassicauline A, each at concentration equivalent of 20 g/mL of corresponding aconitum alkaloids. The prepared solution was checked for completeness of hydrolysis by LC–MS/MS analysis. Working standard mix containing 1 g/mL of each alkaloid (except cytisine) or equivalent for aconitum deacetylated derivatives was prepared by mixing and diluting the stock standards and aconitum alkaloids hydrolysate solution with methanol, and was stored at −70 ◦ C. Cytisine stock standard was used directly. Working internal standard of 200 ng/mL methyllycaconitine was prepared by dilution of stock solution with mobile phase. 2.3. Sample preparation Urine and herbal decoction samples (4 mL), with 100 L working internal standard added, were extracted by liquid–liquid extraction using Toxi-Tube A (Varian, Inc.) by mechanical shaking for 30 min. After centrifugation at 3500 rpm for five min, 100 L of the upper organic phase was aspirated, blow-dried, and then reconstituted with 100 L mobile phase solution (70% solvent A and 30% solvent B) and 75 L purified water. For herbal sample types other than herbal decoction (e.g. selected unused herbs, herbal powder, herbal pills), a two-step clean up procedure was employed because of the complex matrix. 70% ACN was added to grinded herbal samples, making the final concentration to 0.1 g/mL. After sonication for 30 min, 0.5 mL of the extract and 3.5 mL of purified water were then added to Toxi-Tube A (Varian, Inc.) for further extraction as previously described. 2.4. Instrumentation An Agilent 1100 LC system consisting of a binary pump, an autosampler, a degasser, and a column thermostat was used. Chromatographic separation was performed on an Agilent Zorbax
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
65
Table 1 MRM transitions and instrument parameters of the target toxic plant alkaloids. Plant alkaloid group
Analyte
Monoisotopic mass
MRM transition 1 (CEa , V)
MRM transition 2 (CEa , V)
DPb (V)
Aconitum alkaloid
Aconitine Benzoylaconine Benzoylhypaconine Benzoylmesaconine Crassicauline A Deacetylcrassicauline A Deacetylyunaconitine Hypaconitine Mesaconitine Yunaconitine
645.31 603.30 573.29 589.29 643.34 601.33 617.32 615.30 631.30 659.33
646/586 (49) 604/105 (89) 574/105 (87) 590/105 (97) 644/584 (47) 602/135 (71) 618/135 (69) 616/556 (47) 632.1/572 (51) 660/600 (49)
646/105 (89) 604/554 (49) 574/542 (47) 590/540 (51) 644/135 (83) 602/570 (49) 618/568 (51) 616/105 (87) 632.1/105 (97) 660/135 (87)
120 120 120 120 120 120 120 120 120 120
Solanaceous tropane alkaloid
Anisodamine Anisodine Atropine Scopolamine
305.16 319.14 289.17 303.15
306/140 (35) 320/156 (25) 290/124 (35) 304/138 (31)
306/77 (105) 320/138 (35) 290/93 (45) 304/156 (23)
86 66 91 66
Strychnos alkaloid
Brucine Strychnine
394.19 334.17
395/244 (53) 335/184 (55)
395/213 (65) 335/156 (67)
106 116
Colchicum alkaloid
Colchicine
399.17
400/358 (33)
400/310 (39)
101
Sophora alkaloid
Cytisine Matrine/sophoridine N-methylcytisine Oxymatrine
190.11 248.19 204.13 264.18
191/148 (31) 249/148 (47) 205/58 (47) 265/205 (41)
191/77 (77) 249/55 (75) 205/149 (21) 265/136 (47)
76 101 86 116
–
Methyllycaconitine (IS)
682.35
683/216 (67)
–
a b
67
CE, collision energy. DP, declustering potential.
Eclipse XDB C8 column (150 mm × 4.6 mm i.d., 5 m particle size) at 25 ◦ C. The mobile phase consisted of 1 mM ammonium formate and 0.1% formic acid in purified water (solvent A) and 1 mM ammonium formate and 0.1% formic acid in ACN (solvent B). The gradient program started with 12% solvent B and changed linearly to 60% solvent B in 20 min, and then further changed to 100% solvent B at 21 min. The total run time was 24 min. The injection volume was 10 L. MS/MS was performed with an Applied Biosystems 4000 QTrap triple-quadruple mass spectrometer equipped with turbo ion spray source in positive ionization mode. The MS instrumental settings were as follows: curtain gas of 25 units; collisionally activated dissociation with nitrogen gas of medium level; ionspray voltage of 5500 V; ionspray temperature of 600 ◦ C; GS1 and GS2 both of 50 units; collision cell exit potential of 15 V. The dwell time was set to 10 ms.
2.5. Compound optimization and mass spectrometry parameters Multiple reaction monitoring (MRM) mode with two transition pairs per compound was used in MS acquisition. Compound optimization was done by direct syringe infusion of standard solutions into the MS system. The two transition pairs with the highest response and minimum interference were selected for the method set-up. The MRM parameters of the analytes were shown in Table 1.
2.6. Identification criteria The toxic plant alkaloids were identified based on the corresponding retention time (within ±0.5 min of those of the reference standards), presence of two MRM transitions, and relative ion intensity within allowable limits of those of the references. The maximum permitted tolerances were ±20%, ±25%, ±30% and ±50% for relative ion intensities of >50%, >20–50%, >10–20%, and ≤10% respectively [15]. The internal standard was used to guard the whole analytical process of individual samples. Samples with internal standard peak areas less than 50,000 counts were rejected.
2.7. Method validation Blank urine samples collected from healthy volunteers and hospital in-patients who were not taking herbs and colchicine, and Liquichek Urine Toxicology Negative Control from Bio-Rad, were used in the method validation by spiking with the standards as appropriate. Ten blank herbal matrices which did not contain the covered toxic plant alkaloids, spiked with the analytes as appropriate, were used in the method validation of herbal samples. The validation protocol of this qualitative method is based on published guidelines [15–17]. Selectivity was assessed in two ways. Firstly, 20 blank urine samples and 10 blank herbal matrices were tested for any interfering compounds. Secondly, individual toxic plant alkaloids were spiked at a high concentration (1 g/mL) for evaluation of any cross-reactivity with other analytes. Limit of detection (LOD) was determined by spiking 10 urine blanks and 10 herbal blanks with each analyte at a range of concentrations (0.5, 2.5, 5, 50, and 100 ng/mL). LOD was defined as the lowest concentration with 100% positive results based on the pre-defined identification criteria. Matrix effects in urine and herbal samples, and extraction recovery in urine, were evaluated. Matrix-free solvent was spiked with the analytes at the LOD levels (Set A). Two sets of blank matrices from 10 different sources were spiked with the analytes at the LOD levels before extraction (Set B) and after extraction (Set C) respectively. Matrix effect (%) was calculated as (mean peak area of Set C/mean peak area of Set A) × 100. Extraction recovery (%) was calculated as (mean peak area of Set B/mean peak area of Set C) × 100. Carry-over effect was tested by injecting two blank samples before and after a sample containing high concentration of the target analytes (at level of 100 times of the LOD). Carry-over rate (%) was calculated as [(peak area of blank sample after high concentration sample − peak area of blank sample before high concentration sample)/peak area of high concentration sample] × 100. For qualitative analysis, precision at around the cut-off levels was determined by false positive rate and false negative rate. Blank urine and herbal samples were analyzed, and false positive rate (%)
66
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
was calculated as (false positives/total known negatives) × 100. 15 blank urine samples and 15 blank herbal samples, each spiked with the target analytes at levels of 20% above the LOD, were analyzed. False negative rate (%) at around the cut-off level was calculated as (false negatives/total known positives) × 100. The retention time precision and the relative ion intensity precision of all the target analytes, both within-day and between-day, were determined and expressed in relative standard deviation. The relative ion intensity of analytes in 10 different matrices was also evaluated. Post-preparative stability was evaluated by analysis of positive quality control sample in five replicates on day 1, 2, 4 and 7 respectively. The samples were stored at 4 ◦ C until analysis. 2.8. Quality control samples Liquichek Urine Toxicology Negative Control from Bio-Rad (Hercules, CA, USA) was used as negative urine control samples. Positive urine control samples were prepared by spiking the standards to the Liquichek Urine Toxicology Negative Control to yield final concentrations of 240 ng/mL for cytisine; 24 ng/mL for aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, atropine, anisodamine, scopolamine, anisodine, matrine, sophoridine, oxymatrine, N-methylcytisine, brucine, strychnine, and colchicine; and concentrations equivalent of 24 ng/mL of corresponding aconitum alkaloids for aconitum hydrolyzed products. Negative herbal quality control samples were prepared from Rhizoma Fagopyri Dibotrydis, a herb that does not contain the target toxic plant alkaloids. 50 g of Rhizoma Fagopyri Dibotrydis, obtained from Guangzhou Zhixin (Guangzhou, China), was added to 500 mL of water at 90 ◦ C, which was then sonicated for 20 min to obtain the herbal decoction used as negative herbal quality controls. Positive herbal quality control samples were prepared by spiking the toxic plant alkaloid standards to the Rhizoma Fagopyri Dibotrydis herbal decoction to achieve the same final concentrations as the positive urine controls. The quality control samples were stored at −70 ◦ C, and were included in each run of patient samples. 2.9. Patient samples The validated method was applied to analysis of real clinical samples. From 2011 to 2012, 172 cases of suspected herbal poisoning, involving 154 urine samples and 115 herbal samples, were referred to the authors’ laboratory, the only tertiary referral center for clinical toxicology analysis in Hong Kong, for toxic plant alkaloid analysis. Clinical information, including implicated Chinese herbal formulae, were also collected and reviewed if available. Urine samples were stored at 4 ◦ C and herbal samples were stored at room temperature before analysis. 3. Results and discussion We developed a LC–MS/MS method for simultaneous detection of 22 toxic plant alkaloids, including aconitum alkaloids and their hydrolyzed products, solanaceous tropane alkaloids, sophora alkaloids, strychnos alkaloids and colchicine, in herbal and urine samples. Urine was the biological specimen type of choice because of the longer detection window period compared with blood sample. With the complex matrices of herbs and urine, sample extraction was required for sample clean up and also for analyte concentration. The Toxi-Tube A (Varian, Inc.) used for liquid–liquid extraction in this method contained sodium carbonate and bicarbonate in a mixture of heptane, dichloromethane and dichloroethane. The urine samples were normalized to around pH of 9 under the condition. The plant alkaloids were converted to free base under high pH and extracted to the organic layer. The use
Table 2 Selectivity of target toxic plant alkaloids. Analyte tested at 1 g/mLa (alkaloid group)
Interfering peaks present (alkaloid group)
Area of interfering peak to analyte peak (%)
Anisodine (solanaceous tropane alkaloid)
Scopolamine (solanaceous tropane alkaloid) Matrine/sophoridine (sophora alkaloid) Mesaconitine (aconitum alkaloid)
0.6
Oxymatrine (sophora alkaloid) Aconitine (aconitum alkaloid) Yunaconitine (aconitum alkaloid)
Mesaconitine (aconitum alkaloid) Crassicauline A (aconitum alkaloid)
0.4 0.3 0.02 0.05
a No interfering peaks were produced by hypaconitine, mesaconitine, crassicauline A, benzoylaconine, benzoylmesaconine, benzoylhypaconine, deacetylyunaconitine, deacetylcrassicauline A, atropine, anisodamine, scopolamine, matrine, sophoridine, cytisine, N-methylcytisine, brucine, strychnine and colchicine at 1 g/mL.
of mixed mode cation exchange and reverse phase sorbent solid phase extraction was also evaluated but the extraction recovery was lower (data not shown). The chromatographic conditions in the present method provided good separation of the target toxic plant alkaloids, apart from matrine and sophoridine which are diastereomers and could not be separated by our system. The relative ion intensity of MRM transitions was included as identification criteria for all analytes except N-methylcytisine. The ion ratio of N-methylcytisine was not reproducible because even the second abundant ion was less than 5% of the base peak, hence identification criteria of this analyte included only the retention time and the two MRM transitions. The extracted ion chromatograms of all the analytes were shown in Fig. 1. The method was validated for selectivity, LOD, matrix effect, extraction recovery, carry-over effect, false positive rate and false negative rate at around the cut-off level, precision of retention time, precision of relative ion intensity, and post-preparative stability. The method selectivity was demonstrated by the absence of interfering peaks in the tested 20 blank urine and 10 blank herbal matrices. Besides, analysis of individual toxic plant alkaloid standards at high concentration (1 g/mL) showed no or very low level of interference with other analytes. Among the 22 toxic plant alkaloids, only anisodine, oxymatrine, aconitine, and yunaconitine produced small interfering peaks in other analytes, of which all were of the same plant alkaloid group as the standard tested. The areas of these non-specific peaks were 0.02–0.6% of those of the target analytes. As a result, there was no significant interference and identification of the class of toxic plant alkaloids was not affected. The details were shown in Table 2. The LOD of all target analytes except cytisine in both urine and herbal matrices were estimated to be 5 ng/mL (or levels equivalent of 5 ng/mL corresponding aconitum alkaloids for aconitum deacetylated derivatives), and the LOD of cytisine was 50 ng/mL. The matrix effect and extraction recovery results at LOD levels were summarized in Table 3. The herbal matrix effects for all analytes ranged from 89% to 118%, whereas the urine matrix effects for all analytes, except cytisine and N-methylcytisine, were from 91% to 109%. The urine matrix effects of cytisine and N-methylcytisine were 57% and 67% respectively, which were still fit for the purpose for this qualitative method. The urine extraction recovery ranged from 74% to 110% for all analytes except cytisine (15%) and oxymatrine (30%). The low extraction recovery of cytisine and oxymatrine could be related to their low molecular weights, being 190.24 and 264.36 respectively, rendering them easily lost during the drying step. The drying time has been optimized to 2 min to minimize
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
67
Fig. 1. Extracted ion chromatograms of blank urine spiked with 6 ng/mL of N-methylcytisine, cytisine, matrine, sophoridine, oxymatrine, anisodine, anisodamine, scopolamine, brucine, strychnine, atropine, benzoylmesaconine, benzoylaconine, colchicine, benzoylhypaconine, methyllyaconitine (internal standard), benzoylaconine, mesaconitine, deactylcrassicauline A, yunaconitine, aconitine, hypaconitine and crassicauline A.
68
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
Table 3 Matrix effect and extraction recovery of target analytes at LOD levels. Plant alkaloid group
Analyte
Matrix effect in herbs (%)
Matrix effect in urine (%)
Extraction recovery in urine (%)
Aconitum alkaloid
Aconitine Benzoylaconine Benzoylhypaconine Benzoylmesaconine Crassicauline A Deacetyl-crassicauline A Deacetyl-yunaconitine Hypaconitine Mesaconitine Yunaconitine
102 97 95 102 93 103 97 103 103 100
96 96 100 106 101 102 103 102 103 97
97 106 105 103 98 104 101 95 94 98
Solanaceous tropane alkaloid
Anisodamine Anisodine Atropine Scopolamine
97 118 98 110
96 92 103 93
76 96 93 96
Strychnos alkaloid
Brucine Strychnine
89 91
103 96
90 89
106 114 103 107 89
105 57 101 67 97
95 15 98 110 30
Colchicum alkaloid Sophora alkaloid
Colchicine Cytisine Matrine/sophoridine N-methylcytisine Oxymatrine
the loss. Despite the relatively low extraction recovery of cytisine and oxymatrine, which are minor components of sophora alkaloids, identification of sophora alkaloids as a group was not affected with the good extraction efficiency of the other major sophora alkaloids such as matrine and sophoridine. The carry-over effects were found to be less than 0.1% for all analytes. The precision at around the cut-off levels was assessed by the false positive rate and false negative rate. The false positive rates in blank urine and blank herbal samples were 0% for all analytes. The false negative rates at around LOD in urine matrix were 6.7% (one out of 15 samples tested) for both anisodine and oxymatrine and were 0% for the remaining 20 analytes. For herbal matrix,
the false negative rates at around LOD were 13.3% (two out of 15 samples tested) for cytisine; 6.7% (one out of 15 samples tested) for atropine, scopolamine, anisodamine, anisodine, brucine, matrine, and sophoridine; and 0% for the remaining 14 analytes. The withinday and between-day precisions of retention time were determined to be below 1.5% for all analytes. Regarding the relative ion intensity of MRM transitions, the within-day precisions ranged from 3.1 to 12.7% for various analytes, and between-day precisions were from 5.6% to 17.5%. The variability of relative ion intensities in 10 different matrices were tested, and the relative standard deviations ranged from 0.9% to 13.2% for different analytes. N-methylcytisine was not included in the testing of relative ion intensity precision, as
Fig. 2. Extracted ion chromatogram of a representative patient urine sample.
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
the ion ratio was not used for compound identification. The postpreparative stability of the analytes was demonstrated by the stable signals within ±20% in five replicates for seven days. This validated method was successfully applied to the analysis of 154 urine samples and 115 herbal samples from a total of 170 patients with suspected herbal poisoning referred to our center in 2011 and 2012. A representative chromatogram of a patient urine sample was shown in Fig. 2. Among the 170 cases, aconitum alkaloids, sophora alkaloids, solanaceous tropane alkaloids, and strychnos alkaloids were detected in 53 cases (involving 47 urine and 42 herbal samples), 42 cases (involving 33 urine and 27 herbal samples), 18 cases (involving 17 urine and 15 herbal samples), and 6 cases (involving 1 urine and 5 herbal samples) respectively. Mild to potentially fatal poisoning features were seen in these patients. Laboratory identification of toxic plant alkaloids in urine and/or herbal samples, together with other clinical information such as clinical presentation and herbal formula ingredients, played an indispensable role in confirmation of herbal poisoning cases. Unequivocal detection of these toxic plant alkaloids was particularly important for cases of which culprit herb is not clear from the history, or in “hidden” poisoning cases due to herb contamination or misidentification in our experience [1,2]. Though our established method only provided identification, a urine quantitative method was deemed not necessary for confirmation of toxic plant alkaloid poisonings by the authors, taking into account that the urine alkaloid concentration could be affected by multiple factors including time of collection and degree of urine dilution, and that there was little evidence about correlation of urine level with poisoning severity. 4. Conclusion A sensitive and specific method for simultaneous detection of 22 toxic plant alkaloids (aconitum alkaloids, solanaceous tropane
69
alkaloids, sophora alkaloids, strychnos alkaloids and colchicine) in human urine and herbal samples using LC–MS/MS was developed and validated. The method was successfully applied in real clinical samples for identification of toxic plant alkaloids and facilitated investigation of suspected herbal poisoning cases. References [1] S.P.L. Chen, S.W. Ng, W.T. Poon, C.K. Lai, T.M.S. Ngan, M.L. Tse, T.Y.K. Chan, A.Y.W. Chan, T.W.L. Mak, Drug Saf. 35 (2012) 575. [2] K.L. Cheng, Y.C. Chan, T.W.L. Mak, M.L. Tse, F.L. Lau, Hong Kong Med. J. 19 (2013) 38. [3] Z. Wang, J. Zhao, J. Xing, Y. He, D. Guo, J. Anal. Toxicol. 28 (2004) 141. [4] The State Pharmacopoeia Commission of the People’s Republic of China, Pharmacopoeia of the People’s Republic of China, Beijing, 2010. [5] TOXBASE, Strychnine, http://toxbase.org/Poisons-Index-A-Z/S-Products/ Strychnine/ (accessed 16.04.13). [6] A.K. Drew, A. Bensoussan, I.M. Whyte, A.H. Dawson, X. Zhu, S.P. Myers, J. Toxicol. Clin. Toxicol. 40 (2002) 173. [7] Y. Finkelstein, S.E. Aks, J.R. Hutson, D.N. Juurlink, P. Nguyen, G. DubnovRaz, U. Pollak, G. Koren, Y. Bentur, Clin. Toxicol. (Phila.) 48 (2010) 407. [8] K. Ito, Y. Ohyama, Y. Konishi, S. Tanaka, M. Mizugaki, Planta Med. 63 (1997) 75. [9] Z. Yu, Z. Wu, F. Gong, R. Wong, C. Liang, Y. Zhang, Y. Yu, J. Sep. Sci. 35 (2012) 2773. [10] E. Aehle, B. Drager, J. Chromatogr. B 878 (2010) 1391. [11] C.K. Lai, W.T. Poon, Y.W. Chan, J. Anal. Toxicol. 30 (2006) 426. [12] K. Usui, Y. Hayashizaki, M. Hashiyada, A. Nakano, M. Funayama, Leg. Med. (Tokyo) 14 (2012) 126. [13] P. Qiu, X. Chen, X. Chen, L. Lin, C. Ai, J. Chromatogr. B 875 (2008) 471. [14] X. Wu, W. Huang, L. Lu, L. Lin, X. Yang, Anal. Bioanal. Chem. 398 (2010) 1319. [15] Clinical and Laboratory Standards Institute, C50-A, Mass Spectrometry in the Clinical Laboratories: General Principles and Guidance; Approved Guideline, Wayne, Pennsylvania, 2007. [16] AOAC Peer Verified Methods Program, Manual on Policies and Procedures, Arlington, VA, 1993. [17] EURACHEM, The Fitness for Purpose of Analytical Methods, a Laboratory Guide to Method Validation and Related Topics, Teddington, 1998.
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Simultaneous detection of 22 toxic plant alkaloids (aconitum alkaloids, solanaceous tropane alkaloids, sophora alkaloids, strychnos alkaloids and colchicine) in human urine and herbal samples using liquid chromatography–tandem mass spectrometry Sau Wah Ng, Chor Kwan Ching ∗ , Albert Yan Wo Chan, Tony Wing Lai Mak Hospital Authority Toxicology Reference Laboratory, Princess Margaret Hospital, Hong Kong
a r t i c l e
i n f o
Article history: Received 2 July 2013 Accepted 14 October 2013 Available online xxx Keywords: Alkaloids Liquid chromatography Tandem mass spectrometry Urine Herb
a b s t r a c t A liquid chromatography–tandem mass spectrometry method for simultaneous detection of 22 toxic plant alkaloids, including aconitum alkaloids and their hydrolyzed products (aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, benzoylaconine, benzoylmesaconine, benzoylhypaconine, deacetylyunaconitine, deacetylcrassicauline A), solanaceous tropane alkaloids (atropine, anisodamine, scopolamine, anisodine), sophora alkaloids (matrine, sophoridine, oxymatrine, cytisine, Nmethylcytisine), strychnos alkaloids (brucine, strychnine) and colchicine, in herbal and urine samples was developed and validated. Following sample preparation by liquid–liquid extraction, chromatographic separation was achieved on Eclipse XDB C8 column. Identification was based on two multiple reaction monitoring transitions and the relative ion intensity. Method selectivity was demonstrated. The limits of detection were 5 ng/mL for all analytes, except 50 ng/mL for cytisine. The herbal matrix effects ranged from 89% to 118%, whereas the urine matrix effects were between 91% and 109% for all analytes except cytisine (57%) and N-methylcytisine (67%). The urine extraction recovery ranged from 74% to 110% for all analytes, except cytisine (15%) and oxymatrine (30%). With the good extraction efficiency of the other major sophora alkaloids, the relatively low extraction recovery of the minor sophora alkaloids cytisine and oxymatrine did not affect identification of sophora alkaloids as a group. Carry-over was minimal at less than 0.1%. The method was successfully applied in analysis of 170 cases of suspected herbal poisoning, with aconitum alkaloids, sophora alkaloids, solanaceous tropane alkaloids, and strychnos alkaloids being detected in 53, 42, 18, and 6 cases, respectively. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Herbal medicine is commonly used worldwide, especially in Chinese communities. Though generally regarded as natural and harmless, some herbs are highly toxic with potentially fatal adverse effects, and there is no lack of herbal poisoning reports in the literature [1–3]. Herbal poisoning can occur as a result of overdose, incorrect processing and preparation, misidentification, contamination, and also in homicidal and suicidal cases. Aconitum alkaloids, solanaceous tropane alkaloids, strychno alkaloids, sophora alkaloids and colchicine, with various pharmacological and toxic effects, are five groups of frequently encountered toxic plant alkaloids in herbal poisoning cases.
∗ Corresponding author. Tel.: +852 29901881. E-mail address: [email protected] (C.K. Ching). 1570-0232/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jchromb.2013.10.020
The highly toxic aconitum alkaloids are present in Aconitum species, which have been widely used in herbal medicine for their analgesic, anti-inflammatory and cardiotonic effects. For example, the Chinese medicines “chuanwu” and “fuzi” correspond to the mother root and daughter root of Aconitum carmichaeli respectively, whereas “caowu” correspond to the main root of Aconitum kusnezoffii [4]. These herbs should be used after adequate processing and prolonged decoction to reduce the toxicity. Clinical features of acute aconite poisoning include weakness, perioral and limb numbness, arrhythmias, hypotension and gastrointestinal disturbances [1]. Tropane alkaloids can be found in many genera of the Solanaceae family like Datura, Atropa, Mandragora etc. They are competitive antagonist of muscarinic receptors and can lead to anticholinergic toxidrome. Chinese medicine-induced anticholinergic poisoning cases due to substitution or contamination of benign herbs with tropane alkaloid-containing herbs have been reported [2]. Strychnine and brucine, found in seeds of Strychnos nux-vomica as well as other Strychnos species, are highly toxic
64
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
alkaloids. Strychnine is mainly used as a pesticide in the western countries; however, seeds of S. nux-vomica, also known as “maqianzi”, are also used as a Chinese herbal medicine in the treatment arthritis and neurological diseases [4]. Severe strychnine poisoning can cause muscle twitching, repeated convulsions, rhabdomyolysis and death. The minimum lethal dose of strychnine in adults is in the range of 50–100 mg [5]. Sophora alkaloids like matrine and sophoridine are present in Sophora species. Some of these species, such as Sophora flavescens (“kushen”), Sophora alopecuroides (“kudougen”), Sophora tonkinensis (“shandougen”), are used in Chinese medicine. Overdose of “kushen” could lead to toxic effects including nausea, vomiting, dizziness, weakness and ataxia [6]. Colchicine, commonly used in the treatment of gout, is an alkaloid present in Colchicum species like Colchicum autumnale. Colchicine has a low therapeutic index, and excessive doses can cause toxic features such as vomiting, abdominal pain, diarrhea, shock, renal failure, multi-organ failure and bone marrow suppression. Fatal cases of colchicine poisoning have been reported [7]. Detection of toxic plant alkaloids in body fluids and herbal samples plays an important role in clinical and forensic toxicology investigation of suspected herbal poisoning cases. Analytical methods of some toxic alkaloids in biological samples using high performance liquid chromatography (HPLC), gas chromatography–mass spectrometry (GC–MS), capillary electrophoresis, liquid chromatography–mass spectrometry (LC–MS), and liquid chromatography–tandem mass spectrometry (LC–MS/MS) have been described [3,8–14]. However, most of these methods detect only one particular group of alkaloids, e.g. aconitum alkaloids, or at most several alkaloids of different classes. Besides, hydrolyzed derivatives of aconitum alkaloids, which can be formed during herb processing or human metabolism, are not covered in many reported methods. Simultaneous detection of a broad range of toxic plant alkaloids of different groups in herbal and biological samples is very useful in investigation of suspected herbal poisonings, especially in cases of which culprit herb is not clear from the history; cases of herb contamination or misidentification; or those with atypical or non-specific presentation. The objective of this study was to develop a sensitive and selective LC–MS/MS method for simultaneous detection of 22 toxic plant alkaloids, including aconitum alkaloids and their hydrolyzed products (aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, benzoylaconine, benzoylmesaconine, benzoylhypaconine, deacetylyunaconitine, deacetylcrassicauline A), solanaceous tropane alkaloids (atropine, anisodamine, scopolamine, anisodine), sophora alkaloids (matrine, sophoridine, oxymatrine, cytisine, N-methylcytisine), strychnos alkaloids (brucine, strychnine) and colchicine, in herbal and urine samples. The qualitative method was validated by assessing selectivity, limit of detection (LOD), matrix effect, extraction recovery, false positive rate and false negative rate at around LOD, carry-over, precision of retention time, precision of relative ion intensity, and stability. The validated method was successfully applied in analysis of real clinical samples of suspected herbal poisonings.
2. Materials and methods 2.1. Chemicals and solvents Standards of aconitine, race-anisodamine (R,S), and hypaconitine were obtained from LKT (St. Paul, MN, USA); anisodine hydrobromide, brucine, cytosine, scopolamine hydrobromide trihydrate and strychnine were purchased from Sigma–Aldrich (St. Louis, MO, USA); atropine was obtained from International Laboratory USA (San Francisco, USA); colchicine was obtained from Alexis Biochemicals (NY, USA); crassicauline A and yunaconitine
were purchased from Herbstandard Inc. (Davis, USA); matrine, oxymatrine and sophoridine were obtained from the National Institute for the Control of Pharmaceuticals and Biological Products (Beijing, China); mesaconitine was obtained from Wako (Richmond, VA, USA); N-methylcytisine was obtained from TRC (Toronto, Canada). Methyllycaconitine citrate, used as an internal standard, was obtained Tocris Bioscience (Bristol, UK). Acetonitrile (ACN) and methanol of HPLC grade were purchased from Merck (Darmstadt, Germany) and RCI-Labscan (Bangkok, Thailand) respectively. Ammonium acetate, ammonium formate, and ammonium hydroxide were obtained from Sigma–Aldrich (St. Louis, MO, USA), International Laboratory USA (San Francisco, USA), and Tedia (USA) respectively. Formic acid and glacial acetic acid were purchased from Fluka (Switzerland). Purified water was provided by Milli-Q water purification system from Millipore. 2.2. Preparation of standards and hydrolyzed derivatives of aconitum alkaloids Stock standard solutions of 1 mg/mL of each alkaloid (aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, atropine, anisodamine, scopolamine, anisodine, matrine, sophoridine, oxymatrine, cytisine, N-methylcytisine, brucine, strychnine, colchicine) were prepared in ACN and stored at −70 ◦ C. Stock internal standard solution of 1 mg/mL methyllycaconitine was prepared in methanol and stored at −70 ◦ C. The stock solutions of aconitine, hypaconitine, mesaconitine, yunaconitine, and crassicauline A were mixed and diluted to a concentration of 20 g/mL with 5 mM ammonium acetate (buffered at pH 7.0). The diluted solution was incubated at 90 ◦ C for three hours to produce deacetylated derivatives containing benzoylaconine, benzoylhypaconine, benzoylmesaconine, deacetylyunaconitine, and deacetylcrassicauline A, each at concentration equivalent of 20 g/mL of corresponding aconitum alkaloids. The prepared solution was checked for completeness of hydrolysis by LC–MS/MS analysis. Working standard mix containing 1 g/mL of each alkaloid (except cytisine) or equivalent for aconitum deacetylated derivatives was prepared by mixing and diluting the stock standards and aconitum alkaloids hydrolysate solution with methanol, and was stored at −70 ◦ C. Cytisine stock standard was used directly. Working internal standard of 200 ng/mL methyllycaconitine was prepared by dilution of stock solution with mobile phase. 2.3. Sample preparation Urine and herbal decoction samples (4 mL), with 100 L working internal standard added, were extracted by liquid–liquid extraction using Toxi-Tube A (Varian, Inc.) by mechanical shaking for 30 min. After centrifugation at 3500 rpm for five min, 100 L of the upper organic phase was aspirated, blow-dried, and then reconstituted with 100 L mobile phase solution (70% solvent A and 30% solvent B) and 75 L purified water. For herbal sample types other than herbal decoction (e.g. selected unused herbs, herbal powder, herbal pills), a two-step clean up procedure was employed because of the complex matrix. 70% ACN was added to grinded herbal samples, making the final concentration to 0.1 g/mL. After sonication for 30 min, 0.5 mL of the extract and 3.5 mL of purified water were then added to Toxi-Tube A (Varian, Inc.) for further extraction as previously described. 2.4. Instrumentation An Agilent 1100 LC system consisting of a binary pump, an autosampler, a degasser, and a column thermostat was used. Chromatographic separation was performed on an Agilent Zorbax
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
65
Table 1 MRM transitions and instrument parameters of the target toxic plant alkaloids. Plant alkaloid group
Analyte
Monoisotopic mass
MRM transition 1 (CEa , V)
MRM transition 2 (CEa , V)
DPb (V)
Aconitum alkaloid
Aconitine Benzoylaconine Benzoylhypaconine Benzoylmesaconine Crassicauline A Deacetylcrassicauline A Deacetylyunaconitine Hypaconitine Mesaconitine Yunaconitine
645.31 603.30 573.29 589.29 643.34 601.33 617.32 615.30 631.30 659.33
646/586 (49) 604/105 (89) 574/105 (87) 590/105 (97) 644/584 (47) 602/135 (71) 618/135 (69) 616/556 (47) 632.1/572 (51) 660/600 (49)
646/105 (89) 604/554 (49) 574/542 (47) 590/540 (51) 644/135 (83) 602/570 (49) 618/568 (51) 616/105 (87) 632.1/105 (97) 660/135 (87)
120 120 120 120 120 120 120 120 120 120
Solanaceous tropane alkaloid
Anisodamine Anisodine Atropine Scopolamine
305.16 319.14 289.17 303.15
306/140 (35) 320/156 (25) 290/124 (35) 304/138 (31)
306/77 (105) 320/138 (35) 290/93 (45) 304/156 (23)
86 66 91 66
Strychnos alkaloid
Brucine Strychnine
394.19 334.17
395/244 (53) 335/184 (55)
395/213 (65) 335/156 (67)
106 116
Colchicum alkaloid
Colchicine
399.17
400/358 (33)
400/310 (39)
101
Sophora alkaloid
Cytisine Matrine/sophoridine N-methylcytisine Oxymatrine
190.11 248.19 204.13 264.18
191/148 (31) 249/148 (47) 205/58 (47) 265/205 (41)
191/77 (77) 249/55 (75) 205/149 (21) 265/136 (47)
76 101 86 116
–
Methyllycaconitine (IS)
682.35
683/216 (67)
–
a b
67
CE, collision energy. DP, declustering potential.
Eclipse XDB C8 column (150 mm × 4.6 mm i.d., 5 m particle size) at 25 ◦ C. The mobile phase consisted of 1 mM ammonium formate and 0.1% formic acid in purified water (solvent A) and 1 mM ammonium formate and 0.1% formic acid in ACN (solvent B). The gradient program started with 12% solvent B and changed linearly to 60% solvent B in 20 min, and then further changed to 100% solvent B at 21 min. The total run time was 24 min. The injection volume was 10 L. MS/MS was performed with an Applied Biosystems 4000 QTrap triple-quadruple mass spectrometer equipped with turbo ion spray source in positive ionization mode. The MS instrumental settings were as follows: curtain gas of 25 units; collisionally activated dissociation with nitrogen gas of medium level; ionspray voltage of 5500 V; ionspray temperature of 600 ◦ C; GS1 and GS2 both of 50 units; collision cell exit potential of 15 V. The dwell time was set to 10 ms.
2.5. Compound optimization and mass spectrometry parameters Multiple reaction monitoring (MRM) mode with two transition pairs per compound was used in MS acquisition. Compound optimization was done by direct syringe infusion of standard solutions into the MS system. The two transition pairs with the highest response and minimum interference were selected for the method set-up. The MRM parameters of the analytes were shown in Table 1.
2.6. Identification criteria The toxic plant alkaloids were identified based on the corresponding retention time (within ±0.5 min of those of the reference standards), presence of two MRM transitions, and relative ion intensity within allowable limits of those of the references. The maximum permitted tolerances were ±20%, ±25%, ±30% and ±50% for relative ion intensities of >50%, >20–50%, >10–20%, and ≤10% respectively [15]. The internal standard was used to guard the whole analytical process of individual samples. Samples with internal standard peak areas less than 50,000 counts were rejected.
2.7. Method validation Blank urine samples collected from healthy volunteers and hospital in-patients who were not taking herbs and colchicine, and Liquichek Urine Toxicology Negative Control from Bio-Rad, were used in the method validation by spiking with the standards as appropriate. Ten blank herbal matrices which did not contain the covered toxic plant alkaloids, spiked with the analytes as appropriate, were used in the method validation of herbal samples. The validation protocol of this qualitative method is based on published guidelines [15–17]. Selectivity was assessed in two ways. Firstly, 20 blank urine samples and 10 blank herbal matrices were tested for any interfering compounds. Secondly, individual toxic plant alkaloids were spiked at a high concentration (1 g/mL) for evaluation of any cross-reactivity with other analytes. Limit of detection (LOD) was determined by spiking 10 urine blanks and 10 herbal blanks with each analyte at a range of concentrations (0.5, 2.5, 5, 50, and 100 ng/mL). LOD was defined as the lowest concentration with 100% positive results based on the pre-defined identification criteria. Matrix effects in urine and herbal samples, and extraction recovery in urine, were evaluated. Matrix-free solvent was spiked with the analytes at the LOD levels (Set A). Two sets of blank matrices from 10 different sources were spiked with the analytes at the LOD levels before extraction (Set B) and after extraction (Set C) respectively. Matrix effect (%) was calculated as (mean peak area of Set C/mean peak area of Set A) × 100. Extraction recovery (%) was calculated as (mean peak area of Set B/mean peak area of Set C) × 100. Carry-over effect was tested by injecting two blank samples before and after a sample containing high concentration of the target analytes (at level of 100 times of the LOD). Carry-over rate (%) was calculated as [(peak area of blank sample after high concentration sample − peak area of blank sample before high concentration sample)/peak area of high concentration sample] × 100. For qualitative analysis, precision at around the cut-off levels was determined by false positive rate and false negative rate. Blank urine and herbal samples were analyzed, and false positive rate (%)
66
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
was calculated as (false positives/total known negatives) × 100. 15 blank urine samples and 15 blank herbal samples, each spiked with the target analytes at levels of 20% above the LOD, were analyzed. False negative rate (%) at around the cut-off level was calculated as (false negatives/total known positives) × 100. The retention time precision and the relative ion intensity precision of all the target analytes, both within-day and between-day, were determined and expressed in relative standard deviation. The relative ion intensity of analytes in 10 different matrices was also evaluated. Post-preparative stability was evaluated by analysis of positive quality control sample in five replicates on day 1, 2, 4 and 7 respectively. The samples were stored at 4 ◦ C until analysis. 2.8. Quality control samples Liquichek Urine Toxicology Negative Control from Bio-Rad (Hercules, CA, USA) was used as negative urine control samples. Positive urine control samples were prepared by spiking the standards to the Liquichek Urine Toxicology Negative Control to yield final concentrations of 240 ng/mL for cytisine; 24 ng/mL for aconitine, hypaconitine, mesaconitine, yunaconitine, crassicauline A, atropine, anisodamine, scopolamine, anisodine, matrine, sophoridine, oxymatrine, N-methylcytisine, brucine, strychnine, and colchicine; and concentrations equivalent of 24 ng/mL of corresponding aconitum alkaloids for aconitum hydrolyzed products. Negative herbal quality control samples were prepared from Rhizoma Fagopyri Dibotrydis, a herb that does not contain the target toxic plant alkaloids. 50 g of Rhizoma Fagopyri Dibotrydis, obtained from Guangzhou Zhixin (Guangzhou, China), was added to 500 mL of water at 90 ◦ C, which was then sonicated for 20 min to obtain the herbal decoction used as negative herbal quality controls. Positive herbal quality control samples were prepared by spiking the toxic plant alkaloid standards to the Rhizoma Fagopyri Dibotrydis herbal decoction to achieve the same final concentrations as the positive urine controls. The quality control samples were stored at −70 ◦ C, and were included in each run of patient samples. 2.9. Patient samples The validated method was applied to analysis of real clinical samples. From 2011 to 2012, 172 cases of suspected herbal poisoning, involving 154 urine samples and 115 herbal samples, were referred to the authors’ laboratory, the only tertiary referral center for clinical toxicology analysis in Hong Kong, for toxic plant alkaloid analysis. Clinical information, including implicated Chinese herbal formulae, were also collected and reviewed if available. Urine samples were stored at 4 ◦ C and herbal samples were stored at room temperature before analysis. 3. Results and discussion We developed a LC–MS/MS method for simultaneous detection of 22 toxic plant alkaloids, including aconitum alkaloids and their hydrolyzed products, solanaceous tropane alkaloids, sophora alkaloids, strychnos alkaloids and colchicine, in herbal and urine samples. Urine was the biological specimen type of choice because of the longer detection window period compared with blood sample. With the complex matrices of herbs and urine, sample extraction was required for sample clean up and also for analyte concentration. The Toxi-Tube A (Varian, Inc.) used for liquid–liquid extraction in this method contained sodium carbonate and bicarbonate in a mixture of heptane, dichloromethane and dichloroethane. The urine samples were normalized to around pH of 9 under the condition. The plant alkaloids were converted to free base under high pH and extracted to the organic layer. The use
Table 2 Selectivity of target toxic plant alkaloids. Analyte tested at 1 g/mLa (alkaloid group)
Interfering peaks present (alkaloid group)
Area of interfering peak to analyte peak (%)
Anisodine (solanaceous tropane alkaloid)
Scopolamine (solanaceous tropane alkaloid) Matrine/sophoridine (sophora alkaloid) Mesaconitine (aconitum alkaloid)
0.6
Oxymatrine (sophora alkaloid) Aconitine (aconitum alkaloid) Yunaconitine (aconitum alkaloid)
Mesaconitine (aconitum alkaloid) Crassicauline A (aconitum alkaloid)
0.4 0.3 0.02 0.05
a No interfering peaks were produced by hypaconitine, mesaconitine, crassicauline A, benzoylaconine, benzoylmesaconine, benzoylhypaconine, deacetylyunaconitine, deacetylcrassicauline A, atropine, anisodamine, scopolamine, matrine, sophoridine, cytisine, N-methylcytisine, brucine, strychnine and colchicine at 1 g/mL.
of mixed mode cation exchange and reverse phase sorbent solid phase extraction was also evaluated but the extraction recovery was lower (data not shown). The chromatographic conditions in the present method provided good separation of the target toxic plant alkaloids, apart from matrine and sophoridine which are diastereomers and could not be separated by our system. The relative ion intensity of MRM transitions was included as identification criteria for all analytes except N-methylcytisine. The ion ratio of N-methylcytisine was not reproducible because even the second abundant ion was less than 5% of the base peak, hence identification criteria of this analyte included only the retention time and the two MRM transitions. The extracted ion chromatograms of all the analytes were shown in Fig. 1. The method was validated for selectivity, LOD, matrix effect, extraction recovery, carry-over effect, false positive rate and false negative rate at around the cut-off level, precision of retention time, precision of relative ion intensity, and post-preparative stability. The method selectivity was demonstrated by the absence of interfering peaks in the tested 20 blank urine and 10 blank herbal matrices. Besides, analysis of individual toxic plant alkaloid standards at high concentration (1 g/mL) showed no or very low level of interference with other analytes. Among the 22 toxic plant alkaloids, only anisodine, oxymatrine, aconitine, and yunaconitine produced small interfering peaks in other analytes, of which all were of the same plant alkaloid group as the standard tested. The areas of these non-specific peaks were 0.02–0.6% of those of the target analytes. As a result, there was no significant interference and identification of the class of toxic plant alkaloids was not affected. The details were shown in Table 2. The LOD of all target analytes except cytisine in both urine and herbal matrices were estimated to be 5 ng/mL (or levels equivalent of 5 ng/mL corresponding aconitum alkaloids for aconitum deacetylated derivatives), and the LOD of cytisine was 50 ng/mL. The matrix effect and extraction recovery results at LOD levels were summarized in Table 3. The herbal matrix effects for all analytes ranged from 89% to 118%, whereas the urine matrix effects for all analytes, except cytisine and N-methylcytisine, were from 91% to 109%. The urine matrix effects of cytisine and N-methylcytisine were 57% and 67% respectively, which were still fit for the purpose for this qualitative method. The urine extraction recovery ranged from 74% to 110% for all analytes except cytisine (15%) and oxymatrine (30%). The low extraction recovery of cytisine and oxymatrine could be related to their low molecular weights, being 190.24 and 264.36 respectively, rendering them easily lost during the drying step. The drying time has been optimized to 2 min to minimize
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
67
Fig. 1. Extracted ion chromatograms of blank urine spiked with 6 ng/mL of N-methylcytisine, cytisine, matrine, sophoridine, oxymatrine, anisodine, anisodamine, scopolamine, brucine, strychnine, atropine, benzoylmesaconine, benzoylaconine, colchicine, benzoylhypaconine, methyllyaconitine (internal standard), benzoylaconine, mesaconitine, deactylcrassicauline A, yunaconitine, aconitine, hypaconitine and crassicauline A.
68
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
Table 3 Matrix effect and extraction recovery of target analytes at LOD levels. Plant alkaloid group
Analyte
Matrix effect in herbs (%)
Matrix effect in urine (%)
Extraction recovery in urine (%)
Aconitum alkaloid
Aconitine Benzoylaconine Benzoylhypaconine Benzoylmesaconine Crassicauline A Deacetyl-crassicauline A Deacetyl-yunaconitine Hypaconitine Mesaconitine Yunaconitine
102 97 95 102 93 103 97 103 103 100
96 96 100 106 101 102 103 102 103 97
97 106 105 103 98 104 101 95 94 98
Solanaceous tropane alkaloid
Anisodamine Anisodine Atropine Scopolamine
97 118 98 110
96 92 103 93
76 96 93 96
Strychnos alkaloid
Brucine Strychnine
89 91
103 96
90 89
106 114 103 107 89
105 57 101 67 97
95 15 98 110 30
Colchicum alkaloid Sophora alkaloid
Colchicine Cytisine Matrine/sophoridine N-methylcytisine Oxymatrine
the loss. Despite the relatively low extraction recovery of cytisine and oxymatrine, which are minor components of sophora alkaloids, identification of sophora alkaloids as a group was not affected with the good extraction efficiency of the other major sophora alkaloids such as matrine and sophoridine. The carry-over effects were found to be less than 0.1% for all analytes. The precision at around the cut-off levels was assessed by the false positive rate and false negative rate. The false positive rates in blank urine and blank herbal samples were 0% for all analytes. The false negative rates at around LOD in urine matrix were 6.7% (one out of 15 samples tested) for both anisodine and oxymatrine and were 0% for the remaining 20 analytes. For herbal matrix,
the false negative rates at around LOD were 13.3% (two out of 15 samples tested) for cytisine; 6.7% (one out of 15 samples tested) for atropine, scopolamine, anisodamine, anisodine, brucine, matrine, and sophoridine; and 0% for the remaining 14 analytes. The withinday and between-day precisions of retention time were determined to be below 1.5% for all analytes. Regarding the relative ion intensity of MRM transitions, the within-day precisions ranged from 3.1 to 12.7% for various analytes, and between-day precisions were from 5.6% to 17.5%. The variability of relative ion intensities in 10 different matrices were tested, and the relative standard deviations ranged from 0.9% to 13.2% for different analytes. N-methylcytisine was not included in the testing of relative ion intensity precision, as
Fig. 2. Extracted ion chromatogram of a representative patient urine sample.
S.W. Ng et al. / J. Chromatogr. B 942–943 (2013) 63–69
the ion ratio was not used for compound identification. The postpreparative stability of the analytes was demonstrated by the stable signals within ±20% in five replicates for seven days. This validated method was successfully applied to the analysis of 154 urine samples and 115 herbal samples from a total of 170 patients with suspected herbal poisoning referred to our center in 2011 and 2012. A representative chromatogram of a patient urine sample was shown in Fig. 2. Among the 170 cases, aconitum alkaloids, sophora alkaloids, solanaceous tropane alkaloids, and strychnos alkaloids were detected in 53 cases (involving 47 urine and 42 herbal samples), 42 cases (involving 33 urine and 27 herbal samples), 18 cases (involving 17 urine and 15 herbal samples), and 6 cases (involving 1 urine and 5 herbal samples) respectively. Mild to potentially fatal poisoning features were seen in these patients. Laboratory identification of toxic plant alkaloids in urine and/or herbal samples, together with other clinical information such as clinical presentation and herbal formula ingredients, played an indispensable role in confirmation of herbal poisoning cases. Unequivocal detection of these toxic plant alkaloids was particularly important for cases of which culprit herb is not clear from the history, or in “hidden” poisoning cases due to herb contamination or misidentification in our experience [1,2]. Though our established method only provided identification, a urine quantitative method was deemed not necessary for confirmation of toxic plant alkaloid poisonings by the authors, taking into account that the urine alkaloid concentration could be affected by multiple factors including time of collection and degree of urine dilution, and that there was little evidence about correlation of urine level with poisoning severity. 4. Conclusion A sensitive and specific method for simultaneous detection of 22 toxic plant alkaloids (aconitum alkaloids, solanaceous tropane
69
alkaloids, sophora alkaloids, strychnos alkaloids and colchicine) in human urine and herbal samples using LC–MS/MS was developed and validated. The method was successfully applied in real clinical samples for identification of toxic plant alkaloids and facilitated investigation of suspected herbal poisoning cases. References [1] S.P.L. Chen, S.W. Ng, W.T. Poon, C.K. Lai, T.M.S. Ngan, M.L. Tse, T.Y.K. Chan, A.Y.W. Chan, T.W.L. Mak, Drug Saf. 35 (2012) 575. [2] K.L. Cheng, Y.C. Chan, T.W.L. Mak, M.L. Tse, F.L. Lau, Hong Kong Med. J. 19 (2013) 38. [3] Z. Wang, J. Zhao, J. Xing, Y. He, D. Guo, J. Anal. Toxicol. 28 (2004) 141. [4] The State Pharmacopoeia Commission of the People’s Republic of China, Pharmacopoeia of the People’s Republic of China, Beijing, 2010. [5] TOXBASE, Strychnine, http://toxbase.org/Poisons-Index-A-Z/S-Products/ Strychnine/ (accessed 16.04.13). [6] A.K. Drew, A. Bensoussan, I.M. Whyte, A.H. Dawson, X. Zhu, S.P. Myers, J. Toxicol. Clin. Toxicol. 40 (2002) 173. [7] Y. Finkelstein, S.E. Aks, J.R. Hutson, D.N. Juurlink, P. Nguyen, G. DubnovRaz, U. Pollak, G. Koren, Y. Bentur, Clin. Toxicol. (Phila.) 48 (2010) 407. [8] K. Ito, Y. Ohyama, Y. Konishi, S. Tanaka, M. Mizugaki, Planta Med. 63 (1997) 75. [9] Z. Yu, Z. Wu, F. Gong, R. Wong, C. Liang, Y. Zhang, Y. Yu, J. Sep. Sci. 35 (2012) 2773. [10] E. Aehle, B. Drager, J. Chromatogr. B 878 (2010) 1391. [11] C.K. Lai, W.T. Poon, Y.W. Chan, J. Anal. Toxicol. 30 (2006) 426. [12] K. Usui, Y. Hayashizaki, M. Hashiyada, A. Nakano, M. Funayama, Leg. Med. (Tokyo) 14 (2012) 126. [13] P. Qiu, X. Chen, X. Chen, L. Lin, C. Ai, J. Chromatogr. B 875 (2008) 471. [14] X. Wu, W. Huang, L. Lu, L. Lin, X. Yang, Anal. Bioanal. Chem. 398 (2010) 1319. [15] Clinical and Laboratory Standards Institute, C50-A, Mass Spectrometry in the Clinical Laboratories: General Principles and Guidance; Approved Guideline, Wayne, Pennsylvania, 2007. [16] AOAC Peer Verified Methods Program, Manual on Policies and Procedures, Arlington, VA, 1993. [17] EURACHEM, The Fitness for Purpose of Analytical Methods, a Laboratory Guide to Method Validation and Related Topics, Teddington, 1998.
