- Email: [email protected]
Identification of impurities and the statistical classification of methamphetamine using headspace solid phase microextraction and gas chromatography–mass spectrometry
Forensic Science International 160 (2006) 44–52 www.elsevier.com/locate/forsciint
Identification of impurities and the statistical classification of methamphetamine using headspace solid phase microextraction and gas chromatography–mass spectrometry Kenji Kuwayama *, Kenji Tsujikawa, Hajime Miyaguchi, Tatsuyuki Kanamori, Yuko Iwata, Hiroyuki Inoue, Shuji Saitoh, Tohru Kishi National Research Institute of Police Science, First Chemistry Section, 6-3-1, Kashiwanoha, Kashiwa-shi, Chiba 277-0882, Japan Received 25 May 2005; received in revised form 17 August 2005; accepted 19 August 2005 Available online 26 September 2005
Abstract The profiling of impurities in methamphetamine (MA) using headspace solid phase microextraction (HS-SPME) and gas chromatography–mass spectrometry (GC–MS) is described. The extraction of the impurities with an SPME fiber was examined under varying conditions. Optimal chromatograms were obtained when a 50 mg MA sample at 85 8C for 30 min was extracted using a fiber coated with divinylbenzene/carboxen/polydimethylsiloxane. MA samples from nine different origins were analyzed under optimized extraction conditions. Compounds related to MA such as benzaldehyde, benzyl alcohol, amphetamine, benzyl methyl ketone, cis- and trans-1,2-dimethyl-3-phenylaziridine, dimethylamphetamine, N-acetylamphetamine, Nacetylmethamphetamine and N-formylmethamphetamine were detected in the chromatograms. Trace amounts of ethanol, diethyl ether and acetic acid were also detected in some of the chromatograms. The numbers and intensities of the peaks detected were different, depending on the sample. After the areas of the eight principal peaks were converted to their square root and logarithm, similarities among the samples were evaluated by Euclidian distance, cosine distance and correlation coefficient. The results showed that a combination of logarithmic conversion and cosine distance was the most suitable for discriminating and classifying the samples. HS-SPME/GC–MS is a simple and effective method for the extraction and identification of impurities. The present method, in combination with an appropriate statistical analysis, would be useful for developing a profile of impurities in MA. # 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Solid phase microextraction; Methamphetamine; Gas chromatography–mass spectrometry; Impurity profiling; Statistical analysis
1. Introduction Methamphetamine (MA) is currently the major drug of abuse in Japan [1]. The development of a profile of impurities in MA is an important approach for obtaining useful * Corresponding author. Tel.: +81 4 7135 8001; fax: +81 4 7133 9173. E-mail address: [email protected] (K. Kuwayama).
information in criminal investigations, such as the relationship between seized samples, traffic routes and sources of supply [2]. A number of methods have been reported for the impurity profiling of MA, including the use of gas chromatography (GC) with a flame ionization detector [3–9] and GC–mass spectrometry (MS) after liquid–liquid extraction (LLE) [10,11] with organic solvents under either basic or weakly acidic conditions. Although LLE is a versatile technique and requires only simple equipment, it has limitations, in
0379-0738/$ – see front matter # 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.forsciint.2005.08.013
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
terms of the extraction of compounds under specific pH conditions. Headspace solid phase microextraction (HS-SPME) is frequently used in conjunction with the analysis of volatile compounds in air, beverages, water, etc. In the forensic field, it can be applied to the determination of alcohol as well as illicit drugs in blood and urine [12–18]. Methods using HSSPME for the characterization of MA [19,20] and 4-methoxyamphetamine [21] have recently been reported. Unlike LLE, HS-SPME is simple, rapid and solvent-free, since it is based on the partitioning of analytes between a fiber coated with stationary phase and the sample matrix and would provide additional information in terms of characterizing MA samples. In this study, the impurity profiling of MA using HSSPME/GC–MS was investigated, in an attempt to develop a method that involves a simpler preparation and in which the discrimination and classification is more precise.
2. Materials and methods 2.1. Materials Authentic standards of d-MAHCl (Philopon) and l-ephedrineHCl were purchased from Dainippon Pharmaceutical Co. (Osaka, Japan). Two batches of d-MAHCl were synthesized using two different methods. One was the direct reduction of l-ephedrine with hydroiodic acid and red phosphorus. The other involved the preparation from l-ephedrine via chloroephedrine [22]. Six batches of MAHCl were obtained from the Ministry of Health, Labor and Welfare, Japan. They had previously been seized in Japan and had a purity of more than 95%. dl-Amphetamine sulfate, dldimethylamphetamineHCl, N-acetylamphetamine, N-acetyl ephedrine, N-formyl MA and cis- and trans-1,2-dimethyl-3phenylaziridine were synthesized in our laboratory, as previously reported [22–27]. All other chemicals were of analytical grade and were purchased from Wako Pure Chemical Co. (Osaka, Japan). An SPME holder and fibers coated with 100 mm polydimethylsiloxane (PDMS), divinylbenzene/carboxen/ polydimethylsiloxane (DVB/CAR/PDMS) or carbowax/ divinylbenzene (CW/DVB) were purchased from Supelco (PA, USA). 2.2. Sampling and extraction procedures For HS-SPME, varying amounts of d-MAHCl samples were placed in a 10-ml headspace vial, which was then sealed with a silicon septum attached to an aluminum cap. Sampling was done in a chemical-free room to minimize atmospheric background contamination. The vial was heated in a water bath at varying temperatures. An SPME fiber was inserted into the vial and exposed to the headspace inside the vial for varying periods of time. The fiber was conditioned
45
with a GC inlet at 260 8C for 30 min just prior to use to eliminate carry-over. After the extraction, the fiber was immediately inserted into a GC inlet and the compounds were desorbed and analyzed by GC–MS. For the LLE method, a 100 mg sample was dissolved in 1 ml of 0.1 M phosphate buffer pH 7.0 and 0.25 ml of 10% Na2CO3. The solution was extracted with 0.2 ml of ethyl acetate. The organic layer was diluted with ten volumes of ethyl acetate, and a 1 ml aliquot of the solution was injected to GC–MS. After optimizing the SPME conditions, a total of 32 samples, composed of 4 samples each from 7 batches (Philopon and 6 MA seizures) and 2 batches each from our 2 synthesized samples, were analyzed to evaluate intraand inter-batch variations in the impurity profiles. 2.3. GC–MS analysis The GC–MS instrument was an Agilent 6890 GC interfaced with an Agilent 5973N MSD. The column used was an Agilent HP-5MS capillary column (0.32 mm i.d. 30 m, film thickness 0.25 mm). The oven temperature was held at 50 8C for 3 min and ramped at 5 8C/min to 150 8C and 25 8C/min to 275 8C and then held for 2 min at the final temperature. Injection was involved in the splitless mode followed by a purge after 1 min at an inlet temperature of 260 8C. A Supelco inlet liner (0.75 mm i.d.) was used for the SPME. The temperatures of the interface and MS ion source were set at 280 and 230 8C, respectively. The MS was operated in the electron ionization mode with a scan range of m/z 33–400. The control for the GC–MS, data acquisition and data analysis were done using Agilent Chemstation software (G1701 DJ version). Peaks on chromatograms were identified by comparing their retention times and mass spectra with those of authentic standards. 2.4. Data processing Eight principal peaks were selected for discrimination and classification of the samples. Their peak areas on mass chromatograms obtained from the characteristic m/z of each compound were integrated. In cases where the peak was not detected, a value of 10,000, which was nearly the limit of peak area integrated automatically by the software, was assigned. The raw data were processed using Microsoft Excel and converted to their square roots and logarithms. Similarities among samples were calculated using the Euclidean distance, cosine distance and correlation coefficient, and normalized to the range of 0–1 (1 shows the highest similarity). The equations used are shown below (Eqs. (1)–(3)). P ðX X jk Þ2 hP ik i Euclidian distance ¼ 1 (1) Max ðXik X jk Þ2
46
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
P ðXik X jk Þ cosine distance ¼ h P 2 P 2 i1=2 Xik X jk
correlation coefficient ¼ h P
(2)
P ðXik Xi Þ ðX jk X j Þ i1=2 P ðXik Xi Þ2 ðX jk X j Þ2 (3)
Xik and Xi represent area of peak k in sample i and the averaged area of the selected peaks of sample i, respectively. Max[ ] represents the maximum in all pairs. Nine combinations of three types of data and three statistical indices were evaluated by the separation ratio between intra- and inter-batch of MA samples. The separation ratio is defined as the following equation (Eq. (4)). separation ratio ¼
Nðintra aloneÞ NðtotalÞ
(4)
N(intra alone) represents the number of pairs among intrabatches over the maximum of similarities of inter-batch pairs and N(total) represents the total number of pairs among intrabatches, respectively. When the separation ratio is 1, samples in the same batches can be separated completely from those of other batches. The classification of samples was visualized by hierarchical cluster analysis using the group average method under the most discriminative combination.
Fig. 2. Total ion chromatograms obtained from varying the amount of sample used. Each chromatogram is shown on the same scale. Extraction conditions: a sample in a vial was heated at 85 8C. A DVB/CAR/PDMS fiber was exposed to the headspace inside the vial for 30 min and then subjected to GC–MS analysis.
Fig. 1. Total ion chromatograms obtained using a variety of different fibers. Each chromatogram is shown on the same scale. Extraction conditions: a 50 mg sample of MAHCl in a vial was heated at 85 8C. The fiber was exposed to the headspace inside the vial for 30 min and then subjected to GC–MS analysis.
Fig. 3. Total ion chromatograms obtained at different extraction temperatures. Each chromatogram is shown on the same scale. Extraction conditions: a 50 mg sample in a vial was heated at varying temperatures. A DVB/CAR/PDMS fiber was exposed to the headspace inside the vial for 30 min and then subjected to GC–MS analysis.
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
47
Fig. 5. Total ion chromatograms obtained from the same sample using LLE (the upper) and HS-SPME (the lower). Each chromatogram is shown on the same scale. Extraction and analytical conditions are described in the text. Fig. 4. Total ion chromatograms obtained for varying extraction times. Each chromatogram is shown on the same scale. Extraction conditions: a 50 mg sample in a vial was heated at 85 8C. A DVB/ CAR/PDMS fiber was exposed to the headspace inside the vial for varying times and then subjected to GC–MS analysis.
3. Results and discussion 3.1. Optimization of extraction conditions Changes in chromatograms between the types for SPME fiber were initially investigated. Fig. 1 shows the total ion
Table 1 Chromatographic peaks obtained from nine batches Peak number 1 2 3 4
5
6 7 8
9
10
Retention time (min)
Number of occurrence
Major m/z
Tentative or identified compound
1.41 1.51 1.59 1.86 8.41 8.44 9.07 9.56 9.80 10.16 10.45 10.80 11.05 11.39 11.60 11.95 11.98 13.40 13.57 13.59 13.73 13.82 14.77 14.90 15.00 15.37
1 2 2 3 9 4 6 1 1 1 1 7 2 2 2 1 2 8 5 8 1 2 5 5 1 8
36 45, 46 58 60 106, 105, 77 86, 44, 129 45, 75 103, 147, 117 94, 66 91, 126 117, 91 108, 79 118, 119 120, 121, 91, 44 42 108, 107 120 146, 105, 132 44, 91 91, 43, 134 44 92, 45 105, 77, 51 58 58, 138 58, 91
Hydrochloridea Ethanola Diethyl ethera Acetic acid a Benzaldehydea N-Butylbutanaminea 2,20 -Oxybis-ethanol Phenol Benzylchloridea Benzylalcohol a
2-Methylphenol 1,N-Dimethylphenylmethanamine cis-1,2-Dimethyl-3-phenylazirizinea Amphetaminea Benzyl methyl ketonea 1-Phenyl-2-propanol 1-Phenyl-1,2-propanedionea
Methamphetamine a
48
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
Table 1 (Continued ) Peak number
Retention time (min)
Number of occurrence
Major m/z
Tentative or identified compound
11
15.90 16.37 16.29 16.60 16.75 16.81 16.89 17.00 17.20 17.45 17.50 17.63 17.69 18.44 18.62 18.93 18.97 19.28 19.79 19.80 20.40 20.50 20.64 20.80 20.82 20.93 21.36 21.47 21.61 21.74 22.87 22.98 23.05 23.10 23.26 23.42 23.71 23.77 23.82 23.83 23.91 24.00 24.16 24.35 24.54 24.90 25.30 25.40 25.66 26.30 26.33 28.09
3 2 1 3 1 2 1 2 8 1 1 2 1 2 2 4 2 2 4 2 1 3 0 1 1 1 8 1 4 1 1 3 1 4 2 2 0 4 3 2 3 8 4 4 9 1 2 4 4 3 6 2
146, 105 103, 146, 77 94, 138, 77 107, 79 88 72, 44 150, 151, 122 88, 102 72 150, 105, 77 42 86, 44 150, 165 120 132, 133, 91 86 132, 133, 148, 163 120 159, 130, 90, 77 84, 133, 159 160, 58 71, 56, 43 58 132, 133 58 91, 98, 119 156, 141 186, 166, 150, 70 151, 109, 43 138, 56, 181 163, 106, 120, 91 58, 165, 180 105, 169, 133 163, 43 114 124, 167 72, 44, 118 191, 206 205, 220 44, 86, 118 99 91, 148 99, 117 43, 98, 111, 166 86, 58 58, 71, 100, 43 146, 56 99, 155, 211 129, 185 58, 100 119, 91, 132, 154 105, 58, 162, 191
trans-1,2-Dimethyl-3-phenylazirizinea 1-Phenyl-1-methene 2-propanone 2-Phenoxyethanol 1-Phenylpropan-1-ol 2-one
12
13
14
15
a
Ethamphetamine
Dimethylamphetaminea
Ethylmethylamphetaminea Benzylephedrine
Ephedrinea
1,4-Dimethylnaphthalene
N-Formylamphetamine a
N-Acetylamphetaminea
N-Formylmethamphetamine a
N-Acetylephedrinea
Indicates a compound identified by both retention time and mass spectrum compared with those of authentic standard.
chromatograms for the same MA sample using different SPME-fiber coatings, i.e., PDMS, DVB/CAR/PDMS and CW/DVB, which have a low, intermediate and high polarity, respectively. A DVB/CAR/PDMS fiber, which has a wide
range of polarity, could be used to extract many volatile compounds with various polarities, and the maximum number of peaks were found among the three fiber types tested.
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
Fig. 2 shows chromatograms obtained from the same lot, but using different amounts of sample (10, 50 and 100 mg) using a DVB/CAR/PDMS fiber. When 10 mg of the sample was used, it was difficult to detect smaller impurity peaks, and the reproducibility of peak detection was insufficient. To examine the influence of the size of MAHCl crystals on the chromatographic profiles, samples were powdered in a mortar and then analyzed. This resulted in a low peak intensity, especially for volatile compounds, although the variation in the chromatographic profile from the same batch samples was decreased (data not shown). The use of a larger amount of sample increased the reproducibility of peak detection and peak intensity. Little differences in the number of peaks detected and the peak intensities between chromatograms from 50 and 100 mg samples were noted. Therefore, a sample quantity of 50 mg was used in subsequent experiments. Fig. 3 shows the effect of extraction temperature on the chromatograms. The use of a higher temperature increased the peak intensities and the numbers of peaks detected, while some peaks, especially those corresponding to highly volatile compounds, such as ethanol and diethyl ether, decreased because volatile compounds are easily desorbed from the fiber as well as compounds in the crystals are volatilized at higher temperatures. The use of an extraction temperature in excess of 85 8C was not possible, due to the difficulty of temperature control. Therefore, an extraction temperature at 85 8C was used for profiling analysis of MA. The extraction time had less effect on the chromatograms than the other parameters (Fig. 4), and a 30-min extraction time was deemed to be sufficient. Chromatographic profiles obtained by LLE were compared with those by HS-SPME. The relative intensity of MA to impurities, as determined by HS-SPME was much smaller than that by LLE (Fig. 5). HS-SPME enabled the effective extraction of impurities without an overload of MA to GC– MS.
49
Fig. 6. Total ion chromatograms of MA samples from three different batches. Each chromatogram is shown on the same scale. Extraction conditions: optimal condition described in the text. Peaks are listed in Table 1. The compound and the retention time (min) of peak number displayed: 1, were ethanol (1.51); 2, diethyl ether (1.59); 3, acetic acid (1.86); 4, benzaldehyde (8.41); 5, benzyl alcohol (10.80); 6, cis-1,2-dimethyl-3-phenylaziridine (13.40); 7, amphetamine (13.57); 8, benzyl methyl ketone (13.59); 9, 1-phenyl1,2-propanedione (14.77); 10, MA (15.37); 11, trans-1,2-dimethyl3-phenylaziridine (15.90); 12, dimethylamphetamine (17.20); 13, Nacetylamphetamine (23.83); 14, N-formyl MA (24.54); 15, N-acetylephedrine (26.30).
polarity. Optimal conditions for the extraction and analysis of volatile compounds in MA samples are currently under investigation. 3.3. Discrimination/classification
3.2. Identification of impurities MA samples from nine batches (Philopon, six seizures and two synthetics) were analyzed under the optimized extraction conditions. The Peaks observed are summarized in Table 1. Chromatograms of MA samples from three different batches are shown in Fig. 6. They differed from each other in the number of peaks detected and their intensities. At early retention times, traces of volatile compounds such as ethanol (peak 1), diethyl ether (peak 2) and acetic acid (peak 3) were detected. Diethyl ether was detected only on the chromatograms obtained from the two synthetics prepared in our laboratory. This suggests that the solvent used in preparing MA is retained, to some extent, in the crystals. The synthetics were actually recrystallized with diethyl ether. Unfortunately, these peaks were too small and broad to permit their quantification. This problem could be resolved by using a column with a higher
A total of 32 samples from 9 batches were analyzed, to examine variations in the identified peaks. Eight peaks with less intra-batch variation and those related to MA were selected for a statistical discrimination and classification of MA samples. The peaks selected and m/z of the mass chromatograms for area calculation were benzaldehyde (peak 4, m/z 106), benzylalcohol (peak 5, m/z 108), cis1,2-dimethyl-3-phenylaziridine (peak 6, m/z 146), benzyl methyl ketone (peak 8, m/z 134), 1-phenyl-1,2-propanedione (peak 9, m/z 105), trans-1,2-dimethyl-3-phenylaziridine (peak 11, m/z 146), dimethylamphetamine (peak 12, m/z 72) and N-formyl MA (peak 14, m/z 86). The raw data on peak areas were converted to their square roots and logarithms, to prevent the influence of large numerical values. Similarities among samples were evaluated by three indices; Euclidian distance, cosine distance and correlation coefficient [28–31]. Nine combinations were
50
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
Fig. 7. Distribution of similarities between chromatograms from inter-batch pairs (open column) and those from intra-batch pairs (closed column) using statistical indices of similarity (Euclidian distance (EUC), cosine distance (COS) and correlation coefficient (COR)) after mathematical conversions (raw data (RAW), square root (SQR) or logarithm (LOG)). The similarity becomes higher with a shift to the right.
evaluated by calculating the separation ratios between intraand inter-batches. Fig. 7 shows the distribution of similarities obtained from intra-batches (44 pairs) and interbatches (452 pairs) using 9 combinations. The separation ratio denotes the number of pairs in the same batches that are separated from the other batch pairs. Table 2 shows the separation ratios for nine combinations between the three types of data and the three statistical indices of similarity. The similarity calculated using the cosine distance after logarithmic conversion separated intra-batch variations from inter-batch ones well, compared with the other combinations. Table 2 Separation ratios calculated using various statistical analyses Index of similarity
Conversion of peak area Raw data
Square root
Logarithm
Euclidian distance Cosine distance Correlation coefficient
0.34 0.32 0.31
0.42 0.32 0.32
0.26 0.50 0.45
To classify the nine batches (A–I), a cluster analysis was performed using the combination with the highest separation ratio in this study. Four samples from seven batches (A–G) were analyzed except for two samples from two batches (H and I) because of a shortage of sample. A dendrogram obtained from a cluster analysis of 9 batches (32 samples) showed that the same batch profiles were well grouped with the exception of batches D and E (Fig. 8). A visual comparison indicated that the profile of the samples from batches D and E were very similar. In a comparison between this approach and the method using GC (FID) as reported previously [3–9], both extraction procedures are simple and inexpensive. The results of grouping using both methods were almost similar for the nine batches used in this study. This approach, however, has the advantage that it permits the peaks to be identified. Therefore, it is not necessary to compensate for the retention times of impurity peaks using an internal standard. In an impurity profiling with HS-SPME/GC–MS reported previously [19,20], detailed evaluations concerning the statistical analysis [8,28–31], as reported here, were not performed. Using
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
[3]
[4]
[5] [6]
[7] Fig. 8. Dendrogram obtained for a cluster analysis of 32 impurity profiles from 9 batches (A–I). The raw data were converted to their logarithm and similarly among samples was calculated using the cosine distance.
this approach, in addition to a GC (FID) analysis after LLE, which is currently used for impurity profiling in our laboratory [5–8], supplementary information related to MA samples can be readily obtained.
4. Conclusion HS-SPME is simple and rapid for preparation, effective for volatile compounds and nondestructive to the sample. In addition, the advantage of HS-SPME in impurity profiling of MA is that it is possible to introduce extracts into the GC– MS without any overload of the MA to the GC–MS, unlike the LLE method. The extraction ratio of MA to impurities with HS-SPME was found to be much smaller than that for LLE method. A variety of statistical methods were evaluated for the discrimination and classification of samples. An appropriate combination of mathematical conversion and the statistical index of similarity led to a more precise discrimination and classification of samples. This approach would be useful and effective for the impurity profiling of MA and it would also provide supplementary information relative to results obtained by a conventional analysis using LLE method.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
References [1] Present status of firearms and drug problems and countermeasures, in: National Police Agency, Government of Japan (Ed.), White Paper on Police 2000 (Excerpt), The Japan Times, Japan, pp. 43–48. [2] B. Remberg, A.H. Stead, Drug characterization/impurity profiling, with special focus on methamphetamine: recent work of
[17]
51
the United Nations International Drug Control Programme, Bull. Narcotics 51 (1999) 97–117. M. Perkal, Y.L. Ng, J.R. Pearson, Impurity profiling of methamphetamine in Australia and the development of a national drugs database, Forensic Sci. Int. 69 (1994) 77–87. United Nations International Drug Control Programme (Scientific Section), Scientific and Technical Notes: A Practical Guide to Methamphetamine Characterization/Impurity Profiling: Method Procedures, Mass Spectral Data of Selected Impurities, and Literature References, New York, NY, 2000. K. Tanaka, T. Ohmori, T. Inoue, Analysis of impurities in illicit methamphetamine, Forensic Sci. Int. 56 (1992) 157–165. K. Tanaka, T. Ohmori, T. Inoue, S. Seta, Impurity profiling analysis of illicit methamphetamine by capillary gas chromatography, J. Forensic Sci. 39 (1994) 500–511. T. Inoue, K. Tanaka, T. Ohmori, Y. Togawa, S. Seta, Impurity profiling analysis of methamphetamine seized in Japan, Forensic Sci. Int. 69 (1994) 97–102. H. Inoue, T. Kanamori, Y. Iwata, Y. Ohmae, K. Tsujikawa, S. Saitoh, T. Kishi, Methamphetamine impurity profiling using a 0.32 mm i.d. nonpolar capillary column, Forensic Sci. Int. 135 (2003) 42–47. F.M. Dayrit, M.C. Dumlao, Impurity profiling of methamphetamine hydrochloride drugs seized in the Philippines, Forensic Sci. Int. 144 (2004) 20–36. M. Lambrechts, T. Klemetsrud, K.E. Rasmussen, H.J. Storesund, Analysis of Leuckart-specific impurities in amphetamine and methamphetamine, J. Chromatogr. 284 (1984) 99– 502. J.H. Liu, W.W. Ku, J.T. Tsay, M.P. Fitzgerald, S. Kim, Approaches to drug sample differentiation Part III. A comparative study of use of chiral and achiral capillary column gas chromatography/mass spectrometry for the determination of methamphetamine enantiomers and possible impurities, J. Forensic Sci. 27 (1982) 39–48. W.E. Brewer, R.C. Galipo, S.L. Morgan, K.H. Habben, The confirmation of volatiles by solid-phase microextraction and GC–MS in the investigation of two traffic fatalities, J. Anal. Toxicol. 21 (1997) 286–290. N. Nagasawa, M. Yashiki, Y. Iwasaki, K. Hara, T. Kojima, Rapid analysis of amphetamines in blood using head spacesolid phase microextraction and selected ion monitoring, Forensic Sci. Int. 78 (1996) 95–102. M. Yashiki, T. Kojima, T. Miyazaki, N. Nagasawa, Y. Iwasaki, K. Hara, Detection of amphetamines in urine using head spacesolid phase microextraction and chemical ionization selected ion monitoring, Forensic Sci. Int. 76 (1995) 169–177. J. Liu, K. Hara, S. Kashimura, M. Kashiwagi, T. Hamanaka, A. Miyoshi, M. Kagerua, Headspace solid-phase microextraction and gas chromatographic-mass spectrometric screening for volatile hydrocarbons in blood, J. Chromatogr. B 748 (2000) 401–406. D.J. Tranthim-Fryer, R.C. Hansson, K.W. Norman, Headspace/ solid-phase microextraction/gas chromatography-mass spectrometry: a screening technique for the recovery and identification of volatile organic compounds (VOC’s) in postmortem blood and viscera samples, J. Forensic Sci. 46 (2001) 934–946. K.G. Furton, Y. Hong, Y. Hsu, T. Luo, S. Rose, J. Walton, Identification of odor signature chemicals in cocaine using solid-phase microextraciton-gas chromatography and detector-
52
[18]
[19]
[20]
[21]
[22]
[23]
[24]
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52 dog response to isolated compounds spiked on U.S. paper currency, J. Chromatogr. Sci. 40 (2002) 147–155. S. Gentile, A. Torresi, R. Marsili, M. Chiarotti, T. Macchia, Simultaneous detection of amphetamine-like drugs with headspace solid-phase microextraction and gas chromatographymass spectrometry, J. Chromatogr. B 780 (2002) 183–192. C.J. Koester, B.D. Andresen, P.M. Grant, Optimum methamphetamne profiling with sample preparation by solid-phase microextraction, J. Forensic Sci. 47 (2002) 1002–1007. D.T. Vu, SPME/GC–MS characterization of volatiles associated with methamphetamine: toward the development of a pseudomethamphetamine training material, J Forensic Sci. 46 (2001) 1014–1024. J.C. Coumbaros, K.P. Kirkbride, G. Klass, Application of solid-phase microextraction to the profiling of an illicit 4methoxyamphetamine, J. Forensic Sci. 44 (1999) 1237–1242. T. Kishi, T. Inoue, S. Suzuki, T. Yasuda, T. Oikawa, T. Niwaguchi, Analysis of impurities in methamphetamine, Eisei Kagaku 29 (1983) 400–406 (Japanese). V. Lekskulchai, K. Carter, A. Poklis, W. Soine, GC–MS analysis of methamphetamine impurities: reactivity of (+) or ()-chloroephedrine and cis- or trans-1,2-dimethyl-3-phenylaziridine, J. Anal. Toxicol. 24 (2000) 602–605. J. Caldwell, L.G. Dring, R.T. Williams, Metabolism of (14C) methamphetamine in man, the guinea pig and the rat, Biochem. J. 129 (1972) 11–22.
[25] R.P. Barron, A.V. Kruegel, J.M. Moore, T.C. Kram, Identification of impurities in illicit methamphetamine samples, J. Assoc. Off. Anal. Chem. 57 (1974) 1147–1158. [26] C. Conn, M. Dawson, A.T. Baker, J. Keegan, B. Fryirs, Identification of N-acetylmethamphetamine in a sample of illicitly synthesized methamphetamine, J. Forensic Sci. 41 (1996) 645–647. [27] T. Inoue, S. Suzuki, Studies on metabolism of methamphetamine analogs 6. Analysis of dimethylamphetamine and its metabolites, Rep. Natl. Res. Inst. Police Sci. 39 (1986) 228–234 (in Japanese). [28] S. Jonson, Identification and forensic classification of amphetamine, in: Linko¨ing Studies in Science and Technology Dissertations, vol. 641, Linko¨ing, Sweden, 2000. [29] V. Puthaviriyakorn, N. Siriviriyasomboon, J. Phorachata, W. Pan-ox, T. Sasaki, K. Tanaka, Identification of impurities and statistical classification of methamphetamine tablets (Ya–Ba) seized in Thailand, Forensic Sci. Int. 126 (2002) 105–113. [30] P. Gimeno, F. Besacier, H. Chaudron-Thozet, J. Girard, A. Lamotte, A contribution to chemical profiling of 3,4-methylenedioxymethamphetamine (MDMA) tablets, Forensic Sci. Int. 127 (2002) 1–44. [31] P. Esseiva, L. Dujourdy, F. Anglada, F. Taroni, P. Margot, A methodology for illicit heroin seizures comparison in a drug intelligence perspective using large databases, Forensic Sci. Int. 132 (2003) 139–152.
Identification of impurities and the statistical classification of methamphetamine using headspace solid phase microextraction and gas chromatography–mass spectrometry Kenji Kuwayama *, Kenji Tsujikawa, Hajime Miyaguchi, Tatsuyuki Kanamori, Yuko Iwata, Hiroyuki Inoue, Shuji Saitoh, Tohru Kishi National Research Institute of Police Science, First Chemistry Section, 6-3-1, Kashiwanoha, Kashiwa-shi, Chiba 277-0882, Japan Received 25 May 2005; received in revised form 17 August 2005; accepted 19 August 2005 Available online 26 September 2005
Abstract The profiling of impurities in methamphetamine (MA) using headspace solid phase microextraction (HS-SPME) and gas chromatography–mass spectrometry (GC–MS) is described. The extraction of the impurities with an SPME fiber was examined under varying conditions. Optimal chromatograms were obtained when a 50 mg MA sample at 85 8C for 30 min was extracted using a fiber coated with divinylbenzene/carboxen/polydimethylsiloxane. MA samples from nine different origins were analyzed under optimized extraction conditions. Compounds related to MA such as benzaldehyde, benzyl alcohol, amphetamine, benzyl methyl ketone, cis- and trans-1,2-dimethyl-3-phenylaziridine, dimethylamphetamine, N-acetylamphetamine, Nacetylmethamphetamine and N-formylmethamphetamine were detected in the chromatograms. Trace amounts of ethanol, diethyl ether and acetic acid were also detected in some of the chromatograms. The numbers and intensities of the peaks detected were different, depending on the sample. After the areas of the eight principal peaks were converted to their square root and logarithm, similarities among the samples were evaluated by Euclidian distance, cosine distance and correlation coefficient. The results showed that a combination of logarithmic conversion and cosine distance was the most suitable for discriminating and classifying the samples. HS-SPME/GC–MS is a simple and effective method for the extraction and identification of impurities. The present method, in combination with an appropriate statistical analysis, would be useful for developing a profile of impurities in MA. # 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Solid phase microextraction; Methamphetamine; Gas chromatography–mass spectrometry; Impurity profiling; Statistical analysis
1. Introduction Methamphetamine (MA) is currently the major drug of abuse in Japan [1]. The development of a profile of impurities in MA is an important approach for obtaining useful * Corresponding author. Tel.: +81 4 7135 8001; fax: +81 4 7133 9173. E-mail address: [email protected] (K. Kuwayama).
information in criminal investigations, such as the relationship between seized samples, traffic routes and sources of supply [2]. A number of methods have been reported for the impurity profiling of MA, including the use of gas chromatography (GC) with a flame ionization detector [3–9] and GC–mass spectrometry (MS) after liquid–liquid extraction (LLE) [10,11] with organic solvents under either basic or weakly acidic conditions. Although LLE is a versatile technique and requires only simple equipment, it has limitations, in
0379-0738/$ – see front matter # 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.forsciint.2005.08.013
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
terms of the extraction of compounds under specific pH conditions. Headspace solid phase microextraction (HS-SPME) is frequently used in conjunction with the analysis of volatile compounds in air, beverages, water, etc. In the forensic field, it can be applied to the determination of alcohol as well as illicit drugs in blood and urine [12–18]. Methods using HSSPME for the characterization of MA [19,20] and 4-methoxyamphetamine [21] have recently been reported. Unlike LLE, HS-SPME is simple, rapid and solvent-free, since it is based on the partitioning of analytes between a fiber coated with stationary phase and the sample matrix and would provide additional information in terms of characterizing MA samples. In this study, the impurity profiling of MA using HSSPME/GC–MS was investigated, in an attempt to develop a method that involves a simpler preparation and in which the discrimination and classification is more precise.
2. Materials and methods 2.1. Materials Authentic standards of d-MAHCl (Philopon) and l-ephedrineHCl were purchased from Dainippon Pharmaceutical Co. (Osaka, Japan). Two batches of d-MAHCl were synthesized using two different methods. One was the direct reduction of l-ephedrine with hydroiodic acid and red phosphorus. The other involved the preparation from l-ephedrine via chloroephedrine [22]. Six batches of MAHCl were obtained from the Ministry of Health, Labor and Welfare, Japan. They had previously been seized in Japan and had a purity of more than 95%. dl-Amphetamine sulfate, dldimethylamphetamineHCl, N-acetylamphetamine, N-acetyl ephedrine, N-formyl MA and cis- and trans-1,2-dimethyl-3phenylaziridine were synthesized in our laboratory, as previously reported [22–27]. All other chemicals were of analytical grade and were purchased from Wako Pure Chemical Co. (Osaka, Japan). An SPME holder and fibers coated with 100 mm polydimethylsiloxane (PDMS), divinylbenzene/carboxen/ polydimethylsiloxane (DVB/CAR/PDMS) or carbowax/ divinylbenzene (CW/DVB) were purchased from Supelco (PA, USA). 2.2. Sampling and extraction procedures For HS-SPME, varying amounts of d-MAHCl samples were placed in a 10-ml headspace vial, which was then sealed with a silicon septum attached to an aluminum cap. Sampling was done in a chemical-free room to minimize atmospheric background contamination. The vial was heated in a water bath at varying temperatures. An SPME fiber was inserted into the vial and exposed to the headspace inside the vial for varying periods of time. The fiber was conditioned
45
with a GC inlet at 260 8C for 30 min just prior to use to eliminate carry-over. After the extraction, the fiber was immediately inserted into a GC inlet and the compounds were desorbed and analyzed by GC–MS. For the LLE method, a 100 mg sample was dissolved in 1 ml of 0.1 M phosphate buffer pH 7.0 and 0.25 ml of 10% Na2CO3. The solution was extracted with 0.2 ml of ethyl acetate. The organic layer was diluted with ten volumes of ethyl acetate, and a 1 ml aliquot of the solution was injected to GC–MS. After optimizing the SPME conditions, a total of 32 samples, composed of 4 samples each from 7 batches (Philopon and 6 MA seizures) and 2 batches each from our 2 synthesized samples, were analyzed to evaluate intraand inter-batch variations in the impurity profiles. 2.3. GC–MS analysis The GC–MS instrument was an Agilent 6890 GC interfaced with an Agilent 5973N MSD. The column used was an Agilent HP-5MS capillary column (0.32 mm i.d. 30 m, film thickness 0.25 mm). The oven temperature was held at 50 8C for 3 min and ramped at 5 8C/min to 150 8C and 25 8C/min to 275 8C and then held for 2 min at the final temperature. Injection was involved in the splitless mode followed by a purge after 1 min at an inlet temperature of 260 8C. A Supelco inlet liner (0.75 mm i.d.) was used for the SPME. The temperatures of the interface and MS ion source were set at 280 and 230 8C, respectively. The MS was operated in the electron ionization mode with a scan range of m/z 33–400. The control for the GC–MS, data acquisition and data analysis were done using Agilent Chemstation software (G1701 DJ version). Peaks on chromatograms were identified by comparing their retention times and mass spectra with those of authentic standards. 2.4. Data processing Eight principal peaks were selected for discrimination and classification of the samples. Their peak areas on mass chromatograms obtained from the characteristic m/z of each compound were integrated. In cases where the peak was not detected, a value of 10,000, which was nearly the limit of peak area integrated automatically by the software, was assigned. The raw data were processed using Microsoft Excel and converted to their square roots and logarithms. Similarities among samples were calculated using the Euclidean distance, cosine distance and correlation coefficient, and normalized to the range of 0–1 (1 shows the highest similarity). The equations used are shown below (Eqs. (1)–(3)). P ðX X jk Þ2 hP ik i Euclidian distance ¼ 1 (1) Max ðXik X jk Þ2
46
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
P ðXik X jk Þ cosine distance ¼ h P 2 P 2 i1=2 Xik X jk
correlation coefficient ¼ h P
(2)
P ðXik Xi Þ ðX jk X j Þ i1=2 P ðXik Xi Þ2 ðX jk X j Þ2 (3)
Xik and Xi represent area of peak k in sample i and the averaged area of the selected peaks of sample i, respectively. Max[ ] represents the maximum in all pairs. Nine combinations of three types of data and three statistical indices were evaluated by the separation ratio between intra- and inter-batch of MA samples. The separation ratio is defined as the following equation (Eq. (4)). separation ratio ¼
Nðintra aloneÞ NðtotalÞ
(4)
N(intra alone) represents the number of pairs among intrabatches over the maximum of similarities of inter-batch pairs and N(total) represents the total number of pairs among intrabatches, respectively. When the separation ratio is 1, samples in the same batches can be separated completely from those of other batches. The classification of samples was visualized by hierarchical cluster analysis using the group average method under the most discriminative combination.
Fig. 2. Total ion chromatograms obtained from varying the amount of sample used. Each chromatogram is shown on the same scale. Extraction conditions: a sample in a vial was heated at 85 8C. A DVB/CAR/PDMS fiber was exposed to the headspace inside the vial for 30 min and then subjected to GC–MS analysis.
Fig. 1. Total ion chromatograms obtained using a variety of different fibers. Each chromatogram is shown on the same scale. Extraction conditions: a 50 mg sample of MAHCl in a vial was heated at 85 8C. The fiber was exposed to the headspace inside the vial for 30 min and then subjected to GC–MS analysis.
Fig. 3. Total ion chromatograms obtained at different extraction temperatures. Each chromatogram is shown on the same scale. Extraction conditions: a 50 mg sample in a vial was heated at varying temperatures. A DVB/CAR/PDMS fiber was exposed to the headspace inside the vial for 30 min and then subjected to GC–MS analysis.
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
47
Fig. 5. Total ion chromatograms obtained from the same sample using LLE (the upper) and HS-SPME (the lower). Each chromatogram is shown on the same scale. Extraction and analytical conditions are described in the text. Fig. 4. Total ion chromatograms obtained for varying extraction times. Each chromatogram is shown on the same scale. Extraction conditions: a 50 mg sample in a vial was heated at 85 8C. A DVB/ CAR/PDMS fiber was exposed to the headspace inside the vial for varying times and then subjected to GC–MS analysis.
3. Results and discussion 3.1. Optimization of extraction conditions Changes in chromatograms between the types for SPME fiber were initially investigated. Fig. 1 shows the total ion
Table 1 Chromatographic peaks obtained from nine batches Peak number 1 2 3 4
5
6 7 8
9
10
Retention time (min)
Number of occurrence
Major m/z
Tentative or identified compound
1.41 1.51 1.59 1.86 8.41 8.44 9.07 9.56 9.80 10.16 10.45 10.80 11.05 11.39 11.60 11.95 11.98 13.40 13.57 13.59 13.73 13.82 14.77 14.90 15.00 15.37
1 2 2 3 9 4 6 1 1 1 1 7 2 2 2 1 2 8 5 8 1 2 5 5 1 8
36 45, 46 58 60 106, 105, 77 86, 44, 129 45, 75 103, 147, 117 94, 66 91, 126 117, 91 108, 79 118, 119 120, 121, 91, 44 42 108, 107 120 146, 105, 132 44, 91 91, 43, 134 44 92, 45 105, 77, 51 58 58, 138 58, 91
Hydrochloridea Ethanola Diethyl ethera Acetic acid a Benzaldehydea N-Butylbutanaminea 2,20 -Oxybis-ethanol Phenol Benzylchloridea Benzylalcohol a
2-Methylphenol 1,N-Dimethylphenylmethanamine cis-1,2-Dimethyl-3-phenylazirizinea Amphetaminea Benzyl methyl ketonea 1-Phenyl-2-propanol 1-Phenyl-1,2-propanedionea
Methamphetamine a
48
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
Table 1 (Continued ) Peak number
Retention time (min)
Number of occurrence
Major m/z
Tentative or identified compound
11
15.90 16.37 16.29 16.60 16.75 16.81 16.89 17.00 17.20 17.45 17.50 17.63 17.69 18.44 18.62 18.93 18.97 19.28 19.79 19.80 20.40 20.50 20.64 20.80 20.82 20.93 21.36 21.47 21.61 21.74 22.87 22.98 23.05 23.10 23.26 23.42 23.71 23.77 23.82 23.83 23.91 24.00 24.16 24.35 24.54 24.90 25.30 25.40 25.66 26.30 26.33 28.09
3 2 1 3 1 2 1 2 8 1 1 2 1 2 2 4 2 2 4 2 1 3 0 1 1 1 8 1 4 1 1 3 1 4 2 2 0 4 3 2 3 8 4 4 9 1 2 4 4 3 6 2
146, 105 103, 146, 77 94, 138, 77 107, 79 88 72, 44 150, 151, 122 88, 102 72 150, 105, 77 42 86, 44 150, 165 120 132, 133, 91 86 132, 133, 148, 163 120 159, 130, 90, 77 84, 133, 159 160, 58 71, 56, 43 58 132, 133 58 91, 98, 119 156, 141 186, 166, 150, 70 151, 109, 43 138, 56, 181 163, 106, 120, 91 58, 165, 180 105, 169, 133 163, 43 114 124, 167 72, 44, 118 191, 206 205, 220 44, 86, 118 99 91, 148 99, 117 43, 98, 111, 166 86, 58 58, 71, 100, 43 146, 56 99, 155, 211 129, 185 58, 100 119, 91, 132, 154 105, 58, 162, 191
trans-1,2-Dimethyl-3-phenylazirizinea 1-Phenyl-1-methene 2-propanone 2-Phenoxyethanol 1-Phenylpropan-1-ol 2-one
12
13
14
15
a
Ethamphetamine
Dimethylamphetaminea
Ethylmethylamphetaminea Benzylephedrine
Ephedrinea
1,4-Dimethylnaphthalene
N-Formylamphetamine a
N-Acetylamphetaminea
N-Formylmethamphetamine a
N-Acetylephedrinea
Indicates a compound identified by both retention time and mass spectrum compared with those of authentic standard.
chromatograms for the same MA sample using different SPME-fiber coatings, i.e., PDMS, DVB/CAR/PDMS and CW/DVB, which have a low, intermediate and high polarity, respectively. A DVB/CAR/PDMS fiber, which has a wide
range of polarity, could be used to extract many volatile compounds with various polarities, and the maximum number of peaks were found among the three fiber types tested.
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
Fig. 2 shows chromatograms obtained from the same lot, but using different amounts of sample (10, 50 and 100 mg) using a DVB/CAR/PDMS fiber. When 10 mg of the sample was used, it was difficult to detect smaller impurity peaks, and the reproducibility of peak detection was insufficient. To examine the influence of the size of MAHCl crystals on the chromatographic profiles, samples were powdered in a mortar and then analyzed. This resulted in a low peak intensity, especially for volatile compounds, although the variation in the chromatographic profile from the same batch samples was decreased (data not shown). The use of a larger amount of sample increased the reproducibility of peak detection and peak intensity. Little differences in the number of peaks detected and the peak intensities between chromatograms from 50 and 100 mg samples were noted. Therefore, a sample quantity of 50 mg was used in subsequent experiments. Fig. 3 shows the effect of extraction temperature on the chromatograms. The use of a higher temperature increased the peak intensities and the numbers of peaks detected, while some peaks, especially those corresponding to highly volatile compounds, such as ethanol and diethyl ether, decreased because volatile compounds are easily desorbed from the fiber as well as compounds in the crystals are volatilized at higher temperatures. The use of an extraction temperature in excess of 85 8C was not possible, due to the difficulty of temperature control. Therefore, an extraction temperature at 85 8C was used for profiling analysis of MA. The extraction time had less effect on the chromatograms than the other parameters (Fig. 4), and a 30-min extraction time was deemed to be sufficient. Chromatographic profiles obtained by LLE were compared with those by HS-SPME. The relative intensity of MA to impurities, as determined by HS-SPME was much smaller than that by LLE (Fig. 5). HS-SPME enabled the effective extraction of impurities without an overload of MA to GC– MS.
49
Fig. 6. Total ion chromatograms of MA samples from three different batches. Each chromatogram is shown on the same scale. Extraction conditions: optimal condition described in the text. Peaks are listed in Table 1. The compound and the retention time (min) of peak number displayed: 1, were ethanol (1.51); 2, diethyl ether (1.59); 3, acetic acid (1.86); 4, benzaldehyde (8.41); 5, benzyl alcohol (10.80); 6, cis-1,2-dimethyl-3-phenylaziridine (13.40); 7, amphetamine (13.57); 8, benzyl methyl ketone (13.59); 9, 1-phenyl1,2-propanedione (14.77); 10, MA (15.37); 11, trans-1,2-dimethyl3-phenylaziridine (15.90); 12, dimethylamphetamine (17.20); 13, Nacetylamphetamine (23.83); 14, N-formyl MA (24.54); 15, N-acetylephedrine (26.30).
polarity. Optimal conditions for the extraction and analysis of volatile compounds in MA samples are currently under investigation. 3.3. Discrimination/classification
3.2. Identification of impurities MA samples from nine batches (Philopon, six seizures and two synthetics) were analyzed under the optimized extraction conditions. The Peaks observed are summarized in Table 1. Chromatograms of MA samples from three different batches are shown in Fig. 6. They differed from each other in the number of peaks detected and their intensities. At early retention times, traces of volatile compounds such as ethanol (peak 1), diethyl ether (peak 2) and acetic acid (peak 3) were detected. Diethyl ether was detected only on the chromatograms obtained from the two synthetics prepared in our laboratory. This suggests that the solvent used in preparing MA is retained, to some extent, in the crystals. The synthetics were actually recrystallized with diethyl ether. Unfortunately, these peaks were too small and broad to permit their quantification. This problem could be resolved by using a column with a higher
A total of 32 samples from 9 batches were analyzed, to examine variations in the identified peaks. Eight peaks with less intra-batch variation and those related to MA were selected for a statistical discrimination and classification of MA samples. The peaks selected and m/z of the mass chromatograms for area calculation were benzaldehyde (peak 4, m/z 106), benzylalcohol (peak 5, m/z 108), cis1,2-dimethyl-3-phenylaziridine (peak 6, m/z 146), benzyl methyl ketone (peak 8, m/z 134), 1-phenyl-1,2-propanedione (peak 9, m/z 105), trans-1,2-dimethyl-3-phenylaziridine (peak 11, m/z 146), dimethylamphetamine (peak 12, m/z 72) and N-formyl MA (peak 14, m/z 86). The raw data on peak areas were converted to their square roots and logarithms, to prevent the influence of large numerical values. Similarities among samples were evaluated by three indices; Euclidian distance, cosine distance and correlation coefficient [28–31]. Nine combinations were
50
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
Fig. 7. Distribution of similarities between chromatograms from inter-batch pairs (open column) and those from intra-batch pairs (closed column) using statistical indices of similarity (Euclidian distance (EUC), cosine distance (COS) and correlation coefficient (COR)) after mathematical conversions (raw data (RAW), square root (SQR) or logarithm (LOG)). The similarity becomes higher with a shift to the right.
evaluated by calculating the separation ratios between intraand inter-batches. Fig. 7 shows the distribution of similarities obtained from intra-batches (44 pairs) and interbatches (452 pairs) using 9 combinations. The separation ratio denotes the number of pairs in the same batches that are separated from the other batch pairs. Table 2 shows the separation ratios for nine combinations between the three types of data and the three statistical indices of similarity. The similarity calculated using the cosine distance after logarithmic conversion separated intra-batch variations from inter-batch ones well, compared with the other combinations. Table 2 Separation ratios calculated using various statistical analyses Index of similarity
Conversion of peak area Raw data
Square root
Logarithm
Euclidian distance Cosine distance Correlation coefficient
0.34 0.32 0.31
0.42 0.32 0.32
0.26 0.50 0.45
To classify the nine batches (A–I), a cluster analysis was performed using the combination with the highest separation ratio in this study. Four samples from seven batches (A–G) were analyzed except for two samples from two batches (H and I) because of a shortage of sample. A dendrogram obtained from a cluster analysis of 9 batches (32 samples) showed that the same batch profiles were well grouped with the exception of batches D and E (Fig. 8). A visual comparison indicated that the profile of the samples from batches D and E were very similar. In a comparison between this approach and the method using GC (FID) as reported previously [3–9], both extraction procedures are simple and inexpensive. The results of grouping using both methods were almost similar for the nine batches used in this study. This approach, however, has the advantage that it permits the peaks to be identified. Therefore, it is not necessary to compensate for the retention times of impurity peaks using an internal standard. In an impurity profiling with HS-SPME/GC–MS reported previously [19,20], detailed evaluations concerning the statistical analysis [8,28–31], as reported here, were not performed. Using
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52
[3]
[4]
[5] [6]
[7] Fig. 8. Dendrogram obtained for a cluster analysis of 32 impurity profiles from 9 batches (A–I). The raw data were converted to their logarithm and similarly among samples was calculated using the cosine distance.
this approach, in addition to a GC (FID) analysis after LLE, which is currently used for impurity profiling in our laboratory [5–8], supplementary information related to MA samples can be readily obtained.
4. Conclusion HS-SPME is simple and rapid for preparation, effective for volatile compounds and nondestructive to the sample. In addition, the advantage of HS-SPME in impurity profiling of MA is that it is possible to introduce extracts into the GC– MS without any overload of the MA to the GC–MS, unlike the LLE method. The extraction ratio of MA to impurities with HS-SPME was found to be much smaller than that for LLE method. A variety of statistical methods were evaluated for the discrimination and classification of samples. An appropriate combination of mathematical conversion and the statistical index of similarity led to a more precise discrimination and classification of samples. This approach would be useful and effective for the impurity profiling of MA and it would also provide supplementary information relative to results obtained by a conventional analysis using LLE method.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
References [1] Present status of firearms and drug problems and countermeasures, in: National Police Agency, Government of Japan (Ed.), White Paper on Police 2000 (Excerpt), The Japan Times, Japan, pp. 43–48. [2] B. Remberg, A.H. Stead, Drug characterization/impurity profiling, with special focus on methamphetamine: recent work of
[17]
51
the United Nations International Drug Control Programme, Bull. Narcotics 51 (1999) 97–117. M. Perkal, Y.L. Ng, J.R. Pearson, Impurity profiling of methamphetamine in Australia and the development of a national drugs database, Forensic Sci. Int. 69 (1994) 77–87. United Nations International Drug Control Programme (Scientific Section), Scientific and Technical Notes: A Practical Guide to Methamphetamine Characterization/Impurity Profiling: Method Procedures, Mass Spectral Data of Selected Impurities, and Literature References, New York, NY, 2000. K. Tanaka, T. Ohmori, T. Inoue, Analysis of impurities in illicit methamphetamine, Forensic Sci. Int. 56 (1992) 157–165. K. Tanaka, T. Ohmori, T. Inoue, S. Seta, Impurity profiling analysis of illicit methamphetamine by capillary gas chromatography, J. Forensic Sci. 39 (1994) 500–511. T. Inoue, K. Tanaka, T. Ohmori, Y. Togawa, S. Seta, Impurity profiling analysis of methamphetamine seized in Japan, Forensic Sci. Int. 69 (1994) 97–102. H. Inoue, T. Kanamori, Y. Iwata, Y. Ohmae, K. Tsujikawa, S. Saitoh, T. Kishi, Methamphetamine impurity profiling using a 0.32 mm i.d. nonpolar capillary column, Forensic Sci. Int. 135 (2003) 42–47. F.M. Dayrit, M.C. Dumlao, Impurity profiling of methamphetamine hydrochloride drugs seized in the Philippines, Forensic Sci. Int. 144 (2004) 20–36. M. Lambrechts, T. Klemetsrud, K.E. Rasmussen, H.J. Storesund, Analysis of Leuckart-specific impurities in amphetamine and methamphetamine, J. Chromatogr. 284 (1984) 99– 502. J.H. Liu, W.W. Ku, J.T. Tsay, M.P. Fitzgerald, S. Kim, Approaches to drug sample differentiation Part III. A comparative study of use of chiral and achiral capillary column gas chromatography/mass spectrometry for the determination of methamphetamine enantiomers and possible impurities, J. Forensic Sci. 27 (1982) 39–48. W.E. Brewer, R.C. Galipo, S.L. Morgan, K.H. Habben, The confirmation of volatiles by solid-phase microextraction and GC–MS in the investigation of two traffic fatalities, J. Anal. Toxicol. 21 (1997) 286–290. N. Nagasawa, M. Yashiki, Y. Iwasaki, K. Hara, T. Kojima, Rapid analysis of amphetamines in blood using head spacesolid phase microextraction and selected ion monitoring, Forensic Sci. Int. 78 (1996) 95–102. M. Yashiki, T. Kojima, T. Miyazaki, N. Nagasawa, Y. Iwasaki, K. Hara, Detection of amphetamines in urine using head spacesolid phase microextraction and chemical ionization selected ion monitoring, Forensic Sci. Int. 76 (1995) 169–177. J. Liu, K. Hara, S. Kashimura, M. Kashiwagi, T. Hamanaka, A. Miyoshi, M. Kagerua, Headspace solid-phase microextraction and gas chromatographic-mass spectrometric screening for volatile hydrocarbons in blood, J. Chromatogr. B 748 (2000) 401–406. D.J. Tranthim-Fryer, R.C. Hansson, K.W. Norman, Headspace/ solid-phase microextraction/gas chromatography-mass spectrometry: a screening technique for the recovery and identification of volatile organic compounds (VOC’s) in postmortem blood and viscera samples, J. Forensic Sci. 46 (2001) 934–946. K.G. Furton, Y. Hong, Y. Hsu, T. Luo, S. Rose, J. Walton, Identification of odor signature chemicals in cocaine using solid-phase microextraciton-gas chromatography and detector-
52
[18]
[19]
[20]
[21]
[22]
[23]
[24]
K. Kuwayama et al. / Forensic Science International 160 (2006) 44–52 dog response to isolated compounds spiked on U.S. paper currency, J. Chromatogr. Sci. 40 (2002) 147–155. S. Gentile, A. Torresi, R. Marsili, M. Chiarotti, T. Macchia, Simultaneous detection of amphetamine-like drugs with headspace solid-phase microextraction and gas chromatographymass spectrometry, J. Chromatogr. B 780 (2002) 183–192. C.J. Koester, B.D. Andresen, P.M. Grant, Optimum methamphetamne profiling with sample preparation by solid-phase microextraction, J. Forensic Sci. 47 (2002) 1002–1007. D.T. Vu, SPME/GC–MS characterization of volatiles associated with methamphetamine: toward the development of a pseudomethamphetamine training material, J Forensic Sci. 46 (2001) 1014–1024. J.C. Coumbaros, K.P. Kirkbride, G. Klass, Application of solid-phase microextraction to the profiling of an illicit 4methoxyamphetamine, J. Forensic Sci. 44 (1999) 1237–1242. T. Kishi, T. Inoue, S. Suzuki, T. Yasuda, T. Oikawa, T. Niwaguchi, Analysis of impurities in methamphetamine, Eisei Kagaku 29 (1983) 400–406 (Japanese). V. Lekskulchai, K. Carter, A. Poklis, W. Soine, GC–MS analysis of methamphetamine impurities: reactivity of (+) or ()-chloroephedrine and cis- or trans-1,2-dimethyl-3-phenylaziridine, J. Anal. Toxicol. 24 (2000) 602–605. J. Caldwell, L.G. Dring, R.T. Williams, Metabolism of (14C) methamphetamine in man, the guinea pig and the rat, Biochem. J. 129 (1972) 11–22.
[25] R.P. Barron, A.V. Kruegel, J.M. Moore, T.C. Kram, Identification of impurities in illicit methamphetamine samples, J. Assoc. Off. Anal. Chem. 57 (1974) 1147–1158. [26] C. Conn, M. Dawson, A.T. Baker, J. Keegan, B. Fryirs, Identification of N-acetylmethamphetamine in a sample of illicitly synthesized methamphetamine, J. Forensic Sci. 41 (1996) 645–647. [27] T. Inoue, S. Suzuki, Studies on metabolism of methamphetamine analogs 6. Analysis of dimethylamphetamine and its metabolites, Rep. Natl. Res. Inst. Police Sci. 39 (1986) 228–234 (in Japanese). [28] S. Jonson, Identification and forensic classification of amphetamine, in: Linko¨ing Studies in Science and Technology Dissertations, vol. 641, Linko¨ing, Sweden, 2000. [29] V. Puthaviriyakorn, N. Siriviriyasomboon, J. Phorachata, W. Pan-ox, T. Sasaki, K. Tanaka, Identification of impurities and statistical classification of methamphetamine tablets (Ya–Ba) seized in Thailand, Forensic Sci. Int. 126 (2002) 105–113. [30] P. Gimeno, F. Besacier, H. Chaudron-Thozet, J. Girard, A. Lamotte, A contribution to chemical profiling of 3,4-methylenedioxymethamphetamine (MDMA) tablets, Forensic Sci. Int. 127 (2002) 1–44. [31] P. Esseiva, L. Dujourdy, F. Anglada, F. Taroni, P. Margot, A methodology for illicit heroin seizures comparison in a drug intelligence perspective using large databases, Forensic Sci. Int. 132 (2003) 139–152.