Anal Y Sis of Syn Thetic Drugs in Chi Nese Med I Cine by High Per For

Journal of the Chinese Chemi cal Soci ety , 2000, 47, 475-480 475

Analy sis of Synthetic Drugs in Chi nese Med i cine by High Per for mance Liq uid Chro ma tog ra phy/Mass Spec trom e try with In-Source Colli sion In duced Dis so ci a tion

Yi Songa ( ), Hsing-Ling Chenga ( ), Guor-Rong Hera* ( ) and Kuo-Ching Wenb ( ) aDepart ment of Chemis tr y, National Taiwan Univer sit y, Taipei, Taiwan, R.O.C. bNational Labo ra tories of Foods and Drugs, Depart ment of Health, Exec u tive Yuan, R.O.C.

Sixteen chemi cal drugs, often found in adulter ated Chinese medi cine, were studied by high perfor mance liquid chroma tog ra phy/at mo spheric ioniza tion mass spectrom e tr y. Under opti mal cond itions, three pairs of compounds were either coeluted or unre solved. The lack of chromato graphic reso lu tion and the lack of speci - ficity in UV detec tion were overcome by a method based on high perfor mance liquid chroma to gra phy/ electrospray mass spectrom e tr y. This method was capa ble of detect ing the adulterants based on their reten tion times, molec u lar ions, and charac ter is tic fragments result ing from in-source colli sion induced disso ci a tion.

INTRO DUC TION of mass spectrom e try and chroma tog ra phy proved to be a pow er ful method to identify the unknown compounds in a The use of Chinese med i cine to maintain human health compli cated mixture because abso lute identi fi ca tion is often and cure disease has a long and rich histor y. Because Chinese obtained with reten tion time, molec u lar ion and many struc- medi cine is a na ture based therap y, people believe that it is tural charac ter is tic fragment ions present in the mass spec tra. mild and less toxic than chemi cal drugs. However, to amelio - Since its devel op ment in the 60s, gas chroma tog ra phy/mass rate symptoms quickly, tradi tional Chinese medi cines are spectrom e try (GC/MS) has been proved to be a very power ful sometimes ille gally adulter ated with synthetic chemi cal ana lyt i cal tech nique. Unfor tu nately, without chemi cal deri- drugs. This practice can be harmful to health and may result in vatiza tion, this technique is limited to vola tile and thermally side effects or drug al lergy. Therefore, it is impor tant to pro- stable compounds. In recent years, HPLC/MS methods based vide the public with reli able infor ma tion about whether or not on atmo spheric pressure ioniza tion (API) have received con - chemi cal drugs are present in Chinese medi cine. sider able atten tion for the analy sis of nonvol a tile and ther- Conven tional ana lyt i cal method involves the sepa ra tion mally labile compounds. Therefore LC/MS is widely used in of the adulterants by Thin-Layer Chroma tog ra phy (TLC) fol- the fields of biol ogy, biochem is try, envi ron men tal science, lowing identi fi ca tion by retard ing factor and/or ultra vi o let pharma ceu ti cal chemis try and ana lyt i cal chemis try. spectros cop y. These methods are in general time consum ing, We have evalu ated HPLC/MS for the analy sis of syn- labor inten sive and more impor tantly lack the speci fic ity for a thetic drugs in Chinese medi cine. Prelim i nary results in di- high confi dence in identi fi ca tion. Re cently, methods based on cated that this method was capa ble of detect ing adulterants in high perfor mance liquid chroma tog ra phy (HPLC)1-3 and cap- Chinese medi cine. 7 The method has been explored further to illary electro pho re sis (CE)4-6 have been devel oped for the the analy sis of sixteen popu lar chemi cal drugs. The poten tial analy sis of synthetic drugs in tradi tional Chinese medi cine. A and the detailed ana lyt i cal condi tions of this method are dis - vari ety of detec tion tech niques, such as UV, fluo res cence, and cussed in this report. electro chem i cal methods have been devel oped for HPLC and CE. Although they have been shown to be supe rior to conven - tional TLC methods, these methods gener ally lack the speci - EXPER I MEN TAL ficity for high confi dence identi fi ca tion and have only limited capa bil i ties for struc ture eluci da tion. Moreover, analytes Chemi cals co-eluted or closely eluted can be identi fied solely based on All chemi cal drugs, acetaminophen, bucetin, caffeine, their reten tion times. diaz e pam, ethoxybenzamide, fenbufen, flufenamic acid, Mass spec trom e try is known for its high sensi tiv ity and indomethacine, ketoprofen, mefenamic acid, niflumic acid, high speci fic it y. In general, very high speci fic ity can be oxyphenbutazone, phenylbutazone, prednisolone, salicyl- achieved with min i mal quantity of samples. The combi na tion amide and sulindac were obtained from Labo ra tories of Foods 476 J. Chin. Chem. Soc., Vol. 47, No. 3, 2000 Song et al.

and Drugs, Dep. of Health, Taipei, Taiwan. Water was puri - ation) exper i ments, extrac tion lens voltage was set between fied with a Mill-Q system (Millipore Inc., Bedford, MA, 40 and 70 eV. U.S.A.). Etha nol, metha nol and acetonitrile were of HPLC/ Spectro grade, purchased from Merck (Darmstadt, Germany), LAB- SCAN (Labscan Ltd. Dublin, Ire land) and TEDIA RESU LTS AND DIS CUS SION (Fairfield, Ohio, U.S.A.), respec tively. Acetic acid of ana lyt i - cal grade was purchased from Baker (Deventer, Nether lands). Analy sis of Chemical Drugs by HPLC/UV Chemi cal drugs were dissolved in eth a nol and the concen tra - Both narrow bore (2 mm i.d.) and microbore (1 mm i.d.) tion for each was 100 ppm. columns 8,9 were used as sepa ra tion columns in this study. For narrow bore columns with higher flow rate (about 200 Sample Prepa ra tion l/min), it is easier to do the gradi ent elution without pre- Chinese medi cine pills were ob tained from Na tional column split ting. However, post col umn splitting is often Labo ra tories of Foods and Drugs, Dep. of Health, Taipei, Tai- needed to preserve the op ti mum ESI perfor mance. Special wan. Each sample was pounded into powde r. One hundred equipment is often needed to do gradi ent elution with a mil li grams of sample was dissolved in 10 mL etha nol, ex- microbore column. The advan tage 10 of using microbore col - tracted for 15 min by ul tra sound and centri fuged for 1min to umns is that the lower flow rate (about 50 l/min) makes it remove the undis solved sample from the solu tion. possi ble to connect HPLC directly with the mass spec trom e try without splitting. HPLC Sixteen chemi cal drugs (Fig. 1), often found in adulter - All HPLC sepa ra tions were performed on a Microtech ated Chinese medi cine, were sepa rated with a Nucleosil 2mm Prodigix4 HPLC system (Microtech Scien tific Inc., U.S.A.). C18 column and a GSK 1mm ODS-2 microbore column. The A custom packed GSK ODS-2 microbore column (5 m, 1mm best chromatogram obtained from the 2 mm column is shown

15cm) with Nucleosil 100-5 C18 (Macherey-Nagel, Düren, in Fig. 2a. Fourteen instead of sixteen peaks were observed

Germany) and a Nucleosil 120-5 C18 column (5 m, 2mm (Fig. 2). Coinjection with authen tic standards suggested that 25cm; Macherey-Nagel, Düren, Germany) were used ketoprofen coeluted with oxyphenbutazone and indomethacine throughout the study. A Rheodyne 7125 sampling valve coeluted with niflumic acid. In addi tion, flufenamic acid was (Rheodyne, U.S.A.) with a 5- l loop was used as the in jec tor. unre solved from mefenamic acid. The chromatogram ob- A linear gradi ent was employed for LC sepa ra tions begin ning tained with a GSK 1mm ODS-2 microbore column is shown in with 85% of 0.1% acetic acid in acetonitrile followed by Fig. 2b. The two chromatograms were very simi lar except for ramping the acetonitrile content to 100 % in a period of 30 the elution order of diaz e pam. Diaz e pam was the 10th peak in min. UV detec tion was performed by using a Dynamax UV-C Fig. 2a and was the last peak in Fig. 2b. UV/Visi ble detec tor (Rainin MA, U.S.A.) with a flow cell (1.2 l) at 235 nm. Analy sis of Chemi cal Drugs by HPLC/MS It is ille gal to add synthetic drugs into Chinese medi - HPLC/MS cine.11,12 People commit ting the crime may face a maxi mum Mass spectra were acquired using a Fisons Plat form penalty of seven years in prison. To avoid any possi bil ity of mass spectrom e ter (VG BioTech, Altrincham, United King- mistrial, an a lyt i cal data with the highest quality should be dom) equipped with either an electrospray ion source or an at - pursued. The HPLC/UV data, as shown in Fig. 2, may not mospheric pressure chemi cal ioniza tion source. The HPLC have the speci fic ity needed to avoid a false posi tive erro r. was cou pled with an ultra vi o let (UV) detec tor before connect - Mass spec trom e try is known for its sensi tiv ity and speci fic it y. ing to the MS system. The 2 mm column was oper ated at a Further more, by using mass spec trom e try as the chroma tog ra - flow rate of 200 l/min. The flow enter ing the ESI source was phy detec tor, co-eluting peaks or peaks with very simi lar re- reduced to 15 l/min using a post-column splitte r. The 1 mm tention time in HPLC/UV can often be deconvoluted by the re- microbore column was oper ated at a flow rate of 50 l/min solving power of mass spectrom e tr y. and connected directly into the ion source. In ESI, the flow Two of the most pop u lar ioniza tion methods in the cou- rates of nitro gen gas were set at 30 l/hr for nebulization and pling of HPLC with MS are atmo spheric pres sure chemi cal 400 l/hr for drying. In APCI operartion, the sheath gas was set ioniza tion (APCI) and ESI. The study of the 16 drugs by at 150 l/hr and drying gas was set at 350 l/hr. Vapor izer tem- APCI and ESI suggested that ESI was supe rior to APCI be- pera ture was 400 C and the corona discharge pin was main - cause the sensi tiv ity and the abundance of molec u lar ion in tained at 3.5 kV. For in-source CID (colli sion induced disso ci - ESI were gener ally higher than those in APCI. The TIC and Analy sis of Drugs by HPLC/MS J. Chin. Chem. Soc., Vol. 47, No. 3, 2000 477 mass chromatograms of the 16 compounds studied by ESI are butazone. A simi lar condi tion was observed on indomethacine shown in Fig. 3. The two co-eluting peaks were deconvoluted and niflumic acid. Having the same molec u lar weight (M.W. based on their differ ence in molec u lar weight. As can be seen 254), ketoprofen and fenbufen result in the appear ance of two in Fig. 3, the ketoprofen elutes a little earlier than oxyphen - peaks at the trace of m/z 255 (19.80 and 20.40 min).

Fig. 1. Struc tures of the six teen chemi cal drugs. 478 J. Chin. Chem. Soc., Vol. 47, No. 3, 2000 Song et al.

In-Source CID instru ment used in this study is a single quadrupole mass spec- For electrospray ioniza tion, the analyte mole cules are trome ter. The major fragment ions result ing from in-source presented as preformed ions in solu tion. ESI is just a process CID were listed in Table 1. Except for diaz e pam, fragment of ejecting the preformed ion from liquid phase to gas ions were observed for the other fifteen drugs. phase.13-16 Therefore, in compar i son with other ioniza tion methods, ESI produces ions with a mini mum of inter nal en- Analy sis of Adulter ants in Chinese Med i cine ergy and thus little or no fragment ions are observed. Based Several samples were ana lyzed by HPLC/ESIMS; the on reten tion time as well as molec u lar ion, HPLC/MS is more result of one sample is presented. The TIC and mass chroma- specific than HPLC/UV. However, speci fic ity can be fur ther tograms (Fig. 4) suggested that four synthetic drugs were ille - improved if fragment ions are also included. gally added into the prepa ra tion. The m/z 195 and m/z 138 In general, it takes tandem mass spectrom e try to pro- ions at 9.07 minutes corre spond to the protonated molec u lar

duce fragment ions. Fortu natel y, with the devel op ment of ion and the loss of CONCH3 from the molec u lar ion of caf - in-source CID, fragment ions can be produced with a single feine (MW 194). The m/z 166 and m/z 149 ions at 14.45 quadrupole mass spectrom e ter. In the in-source CID, a poten - minutes corre spond to the protonated molec u lar ion and the

tial differ ence is applied between the sample cone and the loss of NH2 from the molec u lar ion of ethoxybenzamide skimme r. When the molec u lar ions pass through the sample cone, they are accel er ated by the poten tial differ ence. These fast moving ions are then collided with gas mole cules and many structur ally specific fragment ions are pro duced. In-source CID was used to produce fragment ions because the

Fig. 2. Gra di ent elu tion of a mixture of six teen chem i- cal drugs, UV detec tion at 235 nm. A linear gradi ent of 85-100% acetonitrile contain ing 0.1% ace tic acid over 30 min. (a) col umn, Nucleosil 120-5 C18 col umn (5 m, 2mm 25cm). (b) column, GSK ODS-2 C18 (5 m, 1mm 15cm). Peaks: (1) acetaminophen; (2) caffeine; (3) salicylamide; (4) bucetin; (5) ethoxybenzamide; (6) prednisolone; (7) sulindac; (8) ketoprofen; (9) oxyphenbutazone; (10) fenbufen; (11) diaz e - pam; (12) indomethacine; (13) niflumic acid; (14) phenylbutazone; (15) flufenamic acid; (16) Fig. 3. The TIC and mass chromatograms of the six teen Mefenamic acid. chem i cal drugs studied by LC/ESI/MS. Analy sis of Drugs by HPLC/MS J. Chin. Chem. Soc., Vol. 47, No. 3, 2000 479

Table 1. Molecular Ions and the Major in Source CID Fragment to the protonated molec u lar ion and the loss of Ions of the Sixteen Chemical Drugs CH3O(C6H3)NC(CH2CO2H)C(CH3) from the molec u lar ion of chemical drug molecular ion (m/z) major fragment ions indomethacine (MW 357). As can be seen in Fig. 4, un like caffeine, ethoxy- acetaminophen 152 110 bucetin 224 206 164 136 benzamide and indomethacine, diaz e pam was identi fied by caffeine 195 138 reten tion time and molec u lar ion but not fragment ions. To diazepam 285 none further enhance the speci fic ity of diaz e pam, a “two columns” ethoxybenzamide 166 149 121 approach was used. It is gen er ally believed that spec i fic ity fenbufen 255 237 181 209 can be improved if the target compound displays differ ent elu - flufenamic acid 282 264 tion behav ior in two differ ent columns. As mentioned earlier, indomethacine 358 139 ketoprofen 255 209 the elu tion order of diaz e pam is quite differ ent between the mefenamic acid 242 224 GSK and the Nucleosil column. The data of the same sample niflumic acid 283 265 ana lyzed by the custom packed GSK ODS-2 microbore col- oxyphenbutazone 325 204 umn is shown in Fig. 5. The chromatogram is simi lar to the phenylbutazone 309 211 120 result from the Nucleosil column (Fig. 4) except the elu tion prednisolone 361 343 salicylamide 138 121 order of the tenta tively assigned diaz e pam. This behav ior is sulindac 357 340 294 248 233 consis tent with the data obtained with an authen tic di az e pam standard as shown in Fig. 2. Conse quentl y, speci fic ity of diaz - epam was improved. (MW 165). The m/z 285 ion at 22.42 minutes corre sponds to the protonated molec u lar ion of diaz e pam (MW 284). The m/z 358 and m/z 139 ions at 22.85 minutes corre spond

Fig. 4. The TIC and mass chromatograms of a real sam- Fig. 5. The TIC and mass chromatograms of the same ple. The LC sep a ra tion was performed on a sample as in Fig. 4. The LC sepa ra tion was per- Nucleosil 120-5 C18 column. The HPLC con di- formed on a GSK ODS-2 C18 col umn. The tions were the same as in Fig. 2. HPLC con di tions were the same as in Fig. 2. 480 J. Chin. Chem. Soc., Vol. 47, No. 3, 2000 Song et al.

CONCLU SIONS REFER ENCES

Whereas many methods can be used in the anal y sis of 1. Kearnns, G. L.; Wilson, J. T. J. Chromatagr. 1981, 226, chemi cal drugs in adulter ated Chinese medi cine, these meth - 183. ods require exces sive time and effort, and lack the speci fic ity 2. Omile, C. I.; Tebbett, I. R. Chromatographia. 1986, 26, for high confi dence identi fi ca tion. HPLC/ESIMS in com bi- 187. nation with in-source CID proved to be a useful method for the 3. Ku, Y.-R.; Tsai, M.-J.; Wen, K.-C. Yaowu Shipin Fenxi anal y sis of synthetic drugs in Chinese medi cine. One disad - 1995, 3(1), 51. vantage of this single quadrupole mass spectro met ric ap- 4. Nishi, H.; Fukuyama, T.; Matsuo, M.; Terabe, S. J. proach is that, unlike the CID in tandem mass spectrom e tr y, Pharm. Sci. 1990, 79(6), 519. there is no mass selec tiv ity before the in-source CID. There- 5. Donato, M. G.; Eeckhout, E. V. D.; Bossche, W. V. D.; fore, the speci fic ity would be inad e quate if the coeluting com- Sandra, P. J. Pharm. and Biomed. Anal. 1993, 11(3), 197. pounds produce the same fragment ions. Another disad van - 6. Ku, Y.-R.; Tsai, M.-J.; Wen, K.-C. Yaowu Shipin Fenxi tage of this approach is that the in-source CID may produce 1995, 3(3), 185. fewer fragment ions than the conven tional CID. These disad - 7. Song, Y.; Her, G.-R.; Wen, K.-C. Journal of Food and vantages may, however, well be offset by the much cheaper Drug Analy sis 1997, 5(4), 295-302. price of a single quadrupole mass spectrom e ter. 8. Horvath, C. G.; Preiss, B. A.; Lipsky, S. R. Anal. Chem. 1967, 39(39), 1422. 9. Ishii, D.; Sakurai, K. Toko Confer ence of Applied Spec- ACKNOWL EDGE MENTS trome tr y, Tokyo, Japan, Octo ber 1973; abstracts 1B05 pp 73. We thank the National Sci ence Council of the Repub lic 10. Scott, R. P. W.; Kucera, P. J. Chromatagr. 1976, 125, 251. of China for finan cial support. 11. Ku, Y.-R.; Tsai, M.-J.; Wen, K.-C. Yaowu Shipin Fenxi 1996, 4 (2) 141. 12. Tsai, M.-J.; Wen, K.-C. Yaowu Shipin Fenxi 1996, 4 (1), Received Novem ber 1, 1999. 49. 13. Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451. Key Words 14. Dole, M.; Hines, R. L.; Mack, L. L.; Mobley, R. C.; Fergu - Synthetic drugs in Chinese medi cine; HPLC; son, L. D.; Alice, M. B. J. Chem. Phys. 1968, 49, 2240. ESI-MS; In-source CID. 15. Kebarle, P.; Tang, L. Anal. Chem. 1993, 65, 972A. 16. Ikonomon, M. G.; Blades, A. T.; Kebarle, P. Anal. Chem. 1991, 63, 1989.