Anal Bioanal Chem (2007) 387:551–559 DOI 10.1007/s00216-006-0950-z

ORIGINAL PAPER

Combined Fourier-transform infrared imaging and desorption electrospray-ionization linear ion-trap mass spectrometry for analysis of counterfeit antimalarial tablets Camilla Ricci & Leonard Nyadong & Facundo M. Fernandez & Paul N. Newton & Sergei G. Kazarian

Received: 4 October 2006 / Revised: 17 October 2006 / Accepted: 17 October 2006 / Published online: 29 November 2006 # Springer-Verlag 2006

Abstract This paper reports use of a combination of Fourier-transform infrared (FTIR) spectroscopic imaging and desorption electrospray ionization linear ion-trap mass spectrometry (DESI MS) for characterization of counterfeit pharmaceutical tablets. The counterfeit artesunate antimalarial tablets were analyzed by both techniques. The results obtained revealed the ability of FTIR imaging in non-destructive micro-attenuated total reflection (ATR) mode to detect the distribution of all components in the tablet, the identities of which were confirmed by DESI MS. Chemical images of the tablets were obtained with high spatial resolution. The FTIR spectroscopic imaging method affords inherent chemical specificity with rapid acquisition of data. DESI MS enables high-sensitivity detection of trace organic compounds. Combination of these two orthogonal surface-characterization methods has great potential for C. Ricci : S. G. Kazarian (*) Department of Chemical Engineering, Imperial College London, London SW7 2AZ, UK e-mail: [email protected] L. Nyadong : F. M. Fernandez School of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, GA 30332, USA P. N. Newton Wellcome Trust-Mahosot Hospital-Oxford Tropical Medicine Research Collaboration, Microbiology Laboratory, Mahosot Hospital, Vientiane, Laos P. N. Newton Centre for Clinical Vaccinology and Tropical Medicine, Churchill Hospital, Oxford University, Oxford OX3 7LJ, UK e-mail: [email protected]

detection and analysis of counterfeit tablets in the open air and without sample preparation. Keywords Fake drugs . Malaria . Chemical imaging . Mass spectrometry . Forensic . Artesunate

Introduction Counterfeiting of pharmaceutical formulations is a serious public health problem both in developed and in developing countries [1–3]. Although conventional chromatographic and spectroscopic methods can be used to detect counterfeit drugs, these usually suffer from low throughput, and researchers are working assiduously to develop faster and more sensitive detection methods. Examples of commonly used analytical techniques include near-infrared spectroscopy, in-situ Raman spectroscopy, tensiography, and chromatography [4]. As discussed by Kazarian et al. [5–10] and Rafferty et al. [11], FTIR spectroscopic imaging is a versatile tool in pharmaceutical research with a wide field of applications, ranging from characterization of drug formulations to elucidation of kinetic processes in drug delivery. In this work we examine the use of ATR-FTIR spectroscopic imaging as an alternative method for rapid characterization of counterfeit drugs. ATR-FTIR has the advantage of being non-destructive, thus requiring no sample preparation. Use of this method to study, in situ, the spatial distribution of the different components in a compacted tablet has been described elsewhere [7]. Here we use ATR-FTIR for analysis of artesunate antimalarial tablet samples previously found to be counterfeit by inspection of the packaging. An epidemic of sophisticated counterfeits of

552

artesunate tablets, used for the treatment of Plasmodium falciparum malaria, has spread through mainland SE Asia since 1998 [2, 12–15]. Malaria is a potentially fatal disease and, with the spread of drug resistance, highly active artemisinin derivatives, for example artesunate, are crucial for efficacious treatment. The ATR-FTIR results were complemented and validated by use of a new open-air screening technique called desorption electrospray ionization (DESI) mass spectrometry [16]. DESI makes use of a high-speed liquid spray directed at a sample held or deposited on a surface at atmospheric pressure. The ions generated during the desorption/ionization process are then sampled by the atmospheric pressure interface of a mass spectrometer. DESI enables rapid analysis of samples of different shapes and sizes under atmospheric pressure and is thus especially well suited to detection of analytes exposed on the surface of counterfeit pharmaceutical tablets. DESI validation of the ATR-FTIR results was performed in two modes, conventional mode, in which a mixture of water and an organic solvent was used as the spray solution, and “reactive DESI” mode, in which a linear primary alkylamine was added to the spray solution to prevent artesunate fragmentation during ionization [17]. DESI provides the m/z of the species detected and spatially located by ATR-FTIR, thus aiding their identification. Compared with chromatographic approaches, which require lengthy sample preparation and analysis steps which often result in throughputs of a few samples per hour, DESI has the added advantage of much

Fig. 1 Micro-ATR-FTIR images collected on a genuine Guilin Pharmaceutical Co., Ltd. (Guilin Pharma) 50-mg artesunate tablet, showing the distribution of (a) artesunate, (b) avicel, and (c) talc. The size of each image is approximately 64×64 μm2. The representative spectra at the location indicated by arrows are also shown and compared with the reference spectra obtained from pure avicel and pure talc

Anal Bioanal Chem (2007) 387:551–559

higher sample throughput (several samples per minute), because of the lack of sample-preparation steps. In summary, in this work we demonstrate, for the first time, the combined potential of DESI and ATR-FTIR as orthogonal and complementary tools for rapid screening and characterization of counterfeit antimalarial tablets.

Experimental All reagents were used as purchased, without additional purification. HPLC-grade acetonitrile (Fisher, Hampton, NH, USA) and dodecylamine (DDA) (Sigma–Aldrich, St Louis, MO, USA) were used for reactive DESI experiments. Ultrapure water (18.2 MΩ cm−1) was obtained from a Nanopure purification unit (Barnstead, San Jose, CA, USA). Counterfeit and genuine artesunate tablets were collected in a wide area of SE Asia encompassing Laos, Myanmar (Burma), and Thailand, and kept refrigerated (4 °C) until analysis [18]. FTIR spectroscopic imaging An FPA detector (Santa Barbara, USA) comprising 16384 pixels arranged in a 128×128 grid format was used to measure FTIR spectra with a spectrometer operating in continuous scan mode. Spectra were collected with 8 cm−1 spectral resolution in the 4000–900 cm−1 range using 32 scans. The chemical images were obtained by attributing a

Anal Bioanal Chem (2007) 387:551–559

colour to each pixel according to the absorbance of a spectral band characteristic of a given compound. In the micro ATR configuration the spectrometer and the FPA detector were coupled to an infrared microscope with a 20× cassegrainian objective and a Ge ATR crystal. The total area imaged was 64×64 μm2. The ATR imaging spectrometer has been patented by Varian [19]. Desorption electrospray ionization mass spectrometry (DESI MS) DESI MS was performed with laboratory-built equipment, using a 75:25 CH3CN–H2O solution as the desorption spray, at a flow rate of 5 μL min−1. The angle between the spray and the sample surface was 55°, whereas the collection angle was approximately 0°. The sprayer tip was positioned 5–6 mm from the mass spectrometer capillary inlet and 1–2 mm from the surface of the tablet. The spray was pneumatically assisted by a coaxial spray of N2 gas at a flow rate of 0.3 L min−1. DESI was performed in the positive-ion mode for all the samples, using +3000 V as the spray voltage, and in the negative-ion mode (−3000 V) for one of the samples (sample code 11 KHA P 7/1 #2). Detection was performed with an LTQ linear iontrap mass spectrometer (Thermo Finnigan, San Jose, CA, USA), tuned for optimum detection of the analyte of interest. The ion-trap mass spectrometer was operated with a maximum injection time of 200 ms, and two microscans per spectrum. Tablets were mounted on an x–y stage and exposed to the DESI jet for 6 s, resulting in a maximum sample throughput of approximately 6 samples min−1, if the time taken to place a tablet in the sample holder is taken into account. Artesunate analysis by liquid chromatogra-

Fig. 2 Genuine Guilin Pharmaceutical artesunate tablet surface mass spectrum produced by DESI. The signal at 407.2 corresponds to the artesunate [M+Na]+ ion. The signal at 790.8 corresponds to the [2M+Na]+ ion

553

phy–MS requires approximately one hour per sample, if the time required by the sample preparation steps are included in the calculation [14]. “Reactive” DESI was performed with 100 μmol L−1 dodecylamine in 75:25 CH3CN–H2O solution as the desorption spray. Under the spray conditions used, a circular 1.8 mm diameter area of the tablet surface was probed. This was determined by spraying a dye solution on to a piece of filter paper, and measuring the average diameter of the spot.

Results and discussion Artemisinin, an antimalarial agent isolated from the Chinese herb Artemisia annua, is an unusual sesquiterpene lactone which contains an epidioxide function [20]. Previous reports have shown that the FTIR spectrum of the δ-lactone contains a signal at 1755 cm−1 arising from carbonyl absorption [21]. Artesunate is a semi-synthetic derivative of

Fig. 3 Micro-ATR-FTIR image of a Type 8 counterfeit artesunate (12 PAS P 64/1) sample showing the distribution of the band at 1403 cm−1. The size of the image is approximately 64×64 μm2. The representative spectrum at the location indicated by the arrow is also shown and compared with the reference spectrum obtained from pure calcium carbonate

554

artemisinin, the spectrum of which also contains this carbonyl absorption. The strong ν(C=O) band at ca. 1750 cm−1 was therefore used to generate the chemical image representative of the distribution of this drug in the imaged area. The images obtained from a genuine Guilin Pharmaceutical Co., Ltd. (Guilin Pharma) (Guangxi, China) artesunate tablet are shown in Fig. 1. It was also observed that genuine Guilin Pharma artesunate tablets give a characteristic band at 1003 cm−1, corresponding to the Fig. 4 (a) Surface DESI spectrum obtained from a fake artesunate sample, Type 8 (12 PAS P 64/1). The two peaks observed correspond to pyrimethamine (m/z=249.2) and sulfadoxine (m/z=311.2); (b) Surface DESI spectrum obtained from sample 12 PAS P 64/1 at different x–y positions showing the inhomogeneous distribution of sulfadoxine and pyrimethamine

Anal Bioanal Chem (2007) 387:551–559

main vibrational mode of talc (Si–O stretching). The presence of a total of four bands in that spectral region, at 1030, 1055, 1100, and 1160 cm−1, suggests the presence of avicel, a common excipient used to enhance or control tablet dissolution [21]. The integrated absorbance of the bands at 1055 cm−1 and 1003 cm−1 were been used to create images showing distribution of avicel (Fig. 1b) and talc (Fig. 1c), respectively. The collected images show that active ingredient (artesunate) domains in avicel and dusting

Anal Bioanal Chem (2007) 387:551–559

powder can easily be identified by use of FTIR imaging; this is not possible with ordinary optical microscopy. The DESI mass spectrum of this genuine Guilin Pharma artesunate tablet is shown in Fig. 2. Mass spectral peaks at m/z 407.2 and 790.8 were observed corresponding to the sodiated artesunate adducts [M+Na]+ and [2M+Na]+, respectively. Peaks below m/z 407.2 corresponding to artesunate fragments were also apparent [17]. Four different fake artesunate samples from the Lao PDR and Burma (Myanmar) were then analyzed—Type 4 (sample code 11 KHA P 7/1#2), Type 8 (sample code 12 PAS P 64/1), Type 9 (sample code S 12/2005), and Type 13 (sample code S

Fig. 5 Micro-ATR-FTIR images collected from a fake artesunate Type 4 (11 KHA P 7/1 #2) tablet sample showing the distribution of the bands (a) at 1008 cm−1 and (b) 1655 cm−1. The size of each image is approximately 64×64 μm2. The representative spectra at the locations indicated by the letters are also shown and compared with the reference spectra of pure talc and pure dipyrone

555

40-1). The packaging of these samples has been discussed elsewhere [11]. The main component of fake artesunate Type 8 (12 PAS P 64/1), as observed by ATR-FTIR, is calcium carbonate, as is apparent from Fig. 3. All the spectra on the surface of this tablet are characterized by the presence of a strong band at 1403 cm−1 assigned to the asymmetric stretching mode (ν3) of the carbonate ion. As expected from the fake packaging, artesunate was not detected. Although calcium carbonate cannot be detected by DESI MS, the DESI mass spectrum for this tablet (Fig. 4a) revealed two additional major peaks, at m/z 249.2 and 311.2 corresponding to the

556

Anal Bioanal Chem (2007) 387:551–559

protonated pyrimethamine and sulfadoxine adducts, respectively. As with FTIR, no artesunate was detected in this sample by DESI. The DESI jet was subsequently directed at different positions on the tablet in the x–y plane. The different spectra obtained (Fig. 4b) indicated different proportions of pyrimethamine (m/z 249.2) and sulfadoxine (m/z 311.2), suggesting inhomogeneities in the composition of the tablet (Fig. 4b). Pyrimethamine or sulfadoxine were not detected by FTIR, because of the large calcium carbonate signal. Results obtained for fake artesunate Type 4 (11 KHA P 7/1#2) tablet are presented in Fig. 5. The spectra extracted from these images show the presence of dipyrone (metamizol) and talc. Again, artesunate was not detected in this tablet. Figure 6a shows the positive-ion mode DESI mass

spectrum obtained from this sample. The most intense signal observed at m/z 218.2 corresponds to the methylaminoantipyrine proton adduct, a dypirone fragment with the elemental formula (C12H16N3O). The same sample was analyzed by DESI in negative-ion mode, and the presence of dypirone was verified by the appearance of the [M−H]− ion (m/z=310.3) in the corresponding spectrum (Fig. 6b). No artesunate was present in this sample, as is evident from the absence of the m/z 407.2 peak in positive-ion mode DESI MS. The presence of dypirone and the absence of artesunate in this sample was again consistent with the results obtained by ATR-FTIR imaging. The main characteristic of all the ATR-FTIR spectra collected on the surface of fake artesunate Type 9 (S 12/ 2005), which has very sophisticated counterfeit packaging, is a sharp band at 1008 cm−1 (Fig. 7). A band at 970 cm−1 also arises from some areas of the sample surface (Fig. 7). The band at approximately 1008 cm−1 is ascribable to the presence of talc whereas that at 970 cm−1 suggests the use

Fig. 6 (a) Surface DESI spectrum obtained from fake artesunate Type 4 (11 KHA P 7/1 #2) tablet sample in positive-ion mode. The peak at m/z=218.2 is a dipyrone fragment. The identity of this peak was confirmed by negative-ion mode DESI experiments shown in (b) The intense signal at 310.3 in this case represents the [M−H]− dipyrone ion

Fig. 7 Micro-ATR-FTIR images collected from fake artesunate Type 9 (S 12/2005) tablet showing the distribution of the bands at (a) 1008 and (b) 1148 cm−1. The size of each image is approximately 64×64 μm2. The representative spectra at the location indicated by the letters are also shown and compared with the reference spectrum of talc

Anal Bioanal Chem (2007) 387:551–559

of dusting powder as excipient. No artesunate was detected in this tablet by FTIR. Figure 8a shows the DESI mass spectrum obtained from this sample. The peak at m/z 152.1 was identified as arising from the acetaminophen protonated adduct ion. The artesunate ion precursor peak at 407.2 was not detectable, even though this sample had previously been shown by HPLC to contain 10 mg artesunate per tablet [17]. DESI analysis was then continued with use of an alternative, higher sensitivity “reactive” DESI mode entailing addition of dodecylamine to the DESI spray (Experimental). Primary amines can form stable noncovalent complexes with artemisinins during electrospray ionization [22]. Thus, in reactive DESI mode, DDA

Fig. 8 (a) Surface mass spectrum obtained from fake artesunate Type 9 (S 12/2005) tablet by conventional DESI in positive-ion mode showing signals arising from the presence of acetaminophen (ACMP). The signal at 407.2 corresponding to artesunate is not detectable. (b) Surface mass spectrum obtained from fake artesunate Type 9 (S 12/2005) tablet by “reactive” DESI. The signal at 570.3 corresponds to the [artesunic acid+DDA+H]+ ion. Artesunate could be detected because of the greater sensitivity of the technique

557

dissolved in the charged spray microdroplets reacts directly with artesunate molecules exposed on the surface tablet. Fake artesunate Type 9 (sample S 12/2005) was reanalyzed using reactive DESI. Figure 8b shows the spectrum obtained. A peak at m/z 570.3 corresponding to the proton-bound non-covalent artesunate complex with dodecylamine [M+DDA+H]+ was readily observed. This peak assignment was verified by DESI MS–MS experiments (data not shown). A second peak at m/z=337.1 corresponding to [acetaminophen+DDA+H]+ was also detected. Detailed characterization of the variables affecting sensitivity in reactive DESI mode will be reported shortly [23]. An interesting feature of ATR-FTIR is that it enables analysis of localized areas of the sample where a particular active ingredient is present. This is crucial in applications such as the screening of counterfeit drugs, samples of which can be highly inhomogeneous because of poor manufacturing practices, as we observed for sample 12 PAS P 64/1. As a further demonstration of this useful feature, the ATR-FTIR image representing distribution of the absorbance of the carbonyl band on the surface of fake artesunate Type 13 (sample S 40-1) was created by plotting the integrated absorbance of the band in the range 1740– 1720 cm−1 over the whole imaged area. This image is presented in Fig. 9a. Detailed inspection of the spectrum extracted from the active ingredient-rich domain showed that the carbonyl group has a characteristic vibration at 1735 cm−1, and not at 1750 cm−1, as is observed for artesunate tablets. According to Kapetanaki and Varostsis this is the characteristic δ-lactone carbonyl ν(C=O) vibration mode of artemisinin, the artesunate precursor [20]. The ATR-FTIR spectrum extracted from the artemisinin-rich domain is shown in Fig. 9b. This spectrum contained signals not only from the artemisinin carbonyl but also from the main band of talc, in accordance with our previous findings. To further verify these results, the reactive DESI spectrum for this counterfeit drug sample was recorded in several positions along one of the tablet’s axes. An intense signal at m/z=469.1 was always detected, corresponding to the DDA complex with artemisinin (Fig. 9c). We have previously detected artemisinin in similar counterfeit antimalarial samples, and its identity has been further confirmed by direct analysis in real time (DART) and accurate mass measurements [17]. Signals corresponding to [DDA+H]+ (m/z=186.2) and [2artemisinin+DDA+H]+ (m/z= 750.5) were also detected.

Conclusions It has been shown that micro ATR-FTIR spectroscopic imaging and DESI MS are promising new methods for characterization of counterfeit pharmaceutical tablets. With

558

Anal Bioanal Chem (2007) 387:551–559

Fig. 9 Micro-ATR-FTIR images collected from fake artesunate Type 13 (S 40-1) tablet showing the distribution of the absorbance of the bands (a) at 1735 cm−1 and (b) at 1008 cm−1. The size of each image is approximately 64× 64 μm2. The representative spectra at the location indicated by the arrows are also shown and compared with the reference spectrum obtained from pure talc. Figure 9c shows a representative reactive DESI-MS spectrum obtained on the sample surface, showing the artemisinin–dodecylamine complex at m/z=469.1

both methods the distribution of different components in counterfeit antimalarial drugs was studied in a totally non destructive way without the need of staining, breaking, or dissolving the tablets. ATR-FTIR imaging is a non-

destructive approach to analysis of drug domains whereas DESI enables high-sensitivity drug detection. These techniques are therefore complementary in the analysis of counterfeit tablets.

Anal Bioanal Chem (2007) 387:551–559 Acknowledgements We thank EPSRC for support (Grant EP/ C532678/1). L.N. thanks the US Pharmacopeia for a scholarship that supports his participation in this project. FMF thanks the Society of Analytical Chemists of Pittsburgh for a starter grant. We thank all those who collected samples, the Food and Drug Department and Food and Drug Quality Control Centre of the Ministry of Health, Lao PDR. The collection of samples was funded by the British Embassy, Bangkok, and the Wellcome Trust (UK). We thank Professors Nicholas White and Nicholas Day for their help.

References 1. Cockburn R, Newton PN, Agyarko EK, Akunyili D, White NJ (2005) Plos Med 2:302–308 2. Newton PN, Green MD, Fernández FM, Day NJP, White NJ (2006) Lancet Infect Dis 6:602–613 3. Coghlan A (2006) New Sci 191:8–9 4. Deisingh AK (2005) Analyst 130:271 5. Kazarian SG, Chan KLA (2003) Macromolecules 36:9866–9872 6. van der Weerd J, Kazarian SG (2004) J Controlled Release 98:295–305 7. Chan KLA, Hammond SV, Kazarian SG (2003) Anal Chem 75:2140–2146 8. Chan KLA, Kazarian SG (2005) J Comb Chem 7:185–189 9. Ricci C, Chan KLA, Kazarian SG (2006) Appl Spectrosc 60:1013–1021 10. Chan KLA, Elkhider N, Kazarian SG (2005) Chem Eng Res Des 83:1303–1310

559 11. Rafferty DW, Koenig JL (2002) J Controlled Release 83:29– 39 12. Newton PN, McGready R, Fernandez F, Green MD, Sunjio M, Bruneton C, Phanouvong S, Millet P, Whitty CJM, Talisuna AO, Proux S, Christophel EM, Malenga G, Singhasivanon P, Bojang K, Kaur H, Palmer K, Day NPJ, Greenwood BM, Nosten F, White NJ (2006) Plos Med 3:752–755 13. De Veij M, Vandenabeele P, Alter Hall K, Fernandez FM, Green MD, White NJ, Dondorp AM, Newton PN, Moens L (2006) J Raman Spectrosc, accepted for publication 14. Alter Hall K, Newton PN, Green MD, De Veij M, Vandenabeele P, Pizzanelli D, Mayxay M, Dondorp A, White NJ, Fernandez FM (2006) Am J Trop Med Hyg, accepted for publication 15. Newton P (2001) Lancet 358:246 16. Takats Z, Wiseman JM, Gologan B, Cooks RG (2004) Science 306:471–473 17. Fernández FM, Cody RB, Green MD, Hampton CY, McGready R, Sengaloundeth S, White NJ, Newton PN (2006) Chem Med Chem 1:702–705 18. Dondorp AM, Desakorn V, Pongtavornpinyo W, Sahassananda D, Silamut K, Chotivanich K, Newton PN, Pitisuttithum P, Smithyman AM, White NJ, Day NPJ (2005) Plos Med 2:788–797 19. Burka EM, Curbelo R (2000) US Patent 6, 141, 100 20. Kapetanaki S, Varotsis C (2001) J Med Chem 44:3150–3156 21. Kalasinsky VF, Jenkins HM, Johnson FB (2002) Vib Spectrosc 28:199–207 22. Stefansson M, Sjoberg PJR, Markides KE (1996) Anal Chem 68:1792–1797 23. Nyadong L, Newton PN, Green M, De Jesus V, Fernandez FM (2006) Anal Chem, in preparation