CHEMICAL ANALYSIS, IDENTIFICATION AND DIFFERENTIATION OF GASTRODIA ELATA BLUME AND OTHER HERBAL MEDICINES WANG HUANSONG A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF PHARMACY NATIONAL UNIVERSITY OF SINGAPORE 2004 ACKNOWLEDGEMENTS This thesis would not have come about without the help of many people. First and foremost, I would especially like to express my most sincere and profound appreciation to my supervisor, Assistant Professor Koh Hwee Ling (Department of Pharmacy), for her constant guidance, invaluable advice and critical comments throughout the course of my research. I would also like to express my heartfelt thanks to Dr. Woo Soo On (Health Sciences Authority), my co-supervisor, for providing many valuable suggestions and constructive comments throughout this work. I would also like to thank the laboratory officers of the Department of Pharmacy, National University of Singapore, in particular, Mr. Tham Mun Chew, Ms Ng Sek Eng, Ms Oh Tang Booy, and Mr Tang Chong Wing for their superb technical assistance. I would like to thank my friends, Mr. Zhang Huihong, Mr. Shang Nianyong, Ms. Zhang Wenxia, Ms Gong Xianlin, and Ms Yang Zhimin, for their help in buying chemical standards and reference herbs. Many thanks are due to my friends and colleagues, especially, Luo Nan and Yiran, for sharing the laughter and tears throughout my work. The receipt of a Research Scholarship from National University of Singapore is gratefully acknowledged. Last but not least, I would like to thank my wife, Shen Ping, for her patience and support during my research course. I TABLE OF CONTENTS ACKNOWLEDGEMENTS I TABLE OF CONTENTS II SURMARY VII LIST OF TABLES IX LIST OF FIGURES X Chapter 1 Introduction 1 1.1 Importance of herbal medicines 1 1.2 An increasing need for quality control of herbal medicines 4 1.3 Analytical methods for quality control of herbal medicines 5 1.4 Objectives 8 Chapter 2 Analysis of Gastrodia elata Blume by HPLC-DAD 10 2.1 Introduction 10 2.1.1 Chemical constituents 10 2.1.2 Biological activities 15 2.1.2.1 Biological activities of G. elata 15 2.1.2.2 Biological activities of selected constituents of G. elata 16 2.1.3 Objectives 2.2 Preparation of gastrodin by HPLC 2.2.1 Introduction 18 20 20 II 2.2.2 Experimental 20 2.2.2.1 Material and reagents 20 2.2.2.2 HPLC conditions 20 2.2.2.3 Melting point 21 2.2.2.4 Infrared spectroscopy 21 2.2.2.5 Ultraviolet spectroscopy 21 2.2.2.6 Mass spectroscopy 22 2.2.2.7 Nuclear magnetic resonance spectroscopy 22 2.2.3 Results and discussion 22 2.2.4 Conclusion 30 2.3 Detection and Determination of Bioactive Constituents in Gastrodia elata by HPLC-DAD 31 2.3.1 Introduction 31 2.3.2 Experimental 32 2.3.2.1 Materials and regent 32 2.3.2.2 HPLC conditions 32 2.3.2.3 Optimization of extraction 33 2.3.2.4 Calibration curve 33 2.3.2.5 Validation 33 2.3.3 Results and discussion 34 2.3.3.1 Separation and detection of gastrodin, 4-hydroxybenzyl alcohol, 4-hydroxybenzylaldehyde and L-pyroglutamic acid 2.3.3.2 Quantitative analysis of gastrodin 34 39 III 2.3.4 Conclusion 44 2.4 Determination of Gastrodin in Chinese Proprietary Medicines Containing Gastrodia elata by Solid Phase Extraction and HPLC 45 2.4.1 Introduction 45 2.4.2 Experimental 46 2.4.2.1 Materials and reagents 46 2.4.2.2 HPLC conditions 46 2.4.2.3 Solid phase extraction method 48 2.4.3 Results and discussion 49 2.4.3.1 Solid phase extraction 49 2.4.3.2 Choice of separation conditions 51 2.4.3.3 Linearity and sensitivity 51 2.4.3.4 Recovery 51 2.4.3.5 Application 52 2.4.4 Conclusion 54 Chapter 3 HPLC Profiling of Herbal Medicines 55 3.1 Introduction 55 3.1.1 Chromatographic fingerprint analysis of herbal medicines 55 3.1.2 Common problem associated with naming of Chinese herbs 56 3.1.3 Cluster analysis 57 3.1.4 Objectives 63 3.2 Experimental 63 IV 3.2.1 Materials and reagents 63 3.2.2 HPLC conditions 65 3.2.3 Sample preparation 65 3.2.4 HPLC fingerprint analysis 66 3.2.4.1 Chromatographic analysis 66 3.2.4.2 Data analysis 67 3.2.5 Detection and identification of marker compounds 67 3.2.5.1 UV library of chemical standards 67 3.2.5.2 RT and RRT of chemical standards 67 3.2.5.3 Peak identification 67 3.3 Results and discussion 70 3.3.1 Choice of HPLC conditions 70 3.3.2 Chromatographic fingerprint analysis 70 3.3.2.1 HPLC profiling 70 3.3.2.2 Comparison of the similarity of chromatographic fingerprint 72 3.3.2.3 Hierarchical cluster analysis 72 3.3.3 Detection of marker compounds by HPLC-DAD 83 3.3.3.1 RT and RRT 83 3.3.3.2 Peak identification 84 3.4 Conclusion 86 Chapter 4. Differentiation and Authentication of Stephania tetrandra and Aristolochia fangchi by HPLC-DAD and LC-MS/MS 87 V 4.1 Introduction 87 4.2 Experimental 90 4.2.1 Chemicals and materials 90 4.2.2 Sample preparation 90 4.2.3 HPLC-DAD conditions 92 4.2.4 LC/MS conditions 92 4.3 Results and discussion 93 4.3.1 HPLC-DAD analysis 93 4.3.1.1 HPLC profiling of roots of Aristolochia fangchi and Stephania tetrandra 4.3.1.2 HPLC analysis of local Fangji samples 4.3.2 LC-MS/MS analysis 93 96 97 4.3.2.1 Optimization of APCI parameters 97 4.3.2.2 Mass spectra of the reference standards 105 4.3.2.3 LC-MS/MS analysis of Aristolochia fangchi (NICPBP) and Stephania tetrandra (Tongrentang) 4.3.2.4 LC-MS/MS analysis of local samples 109 114 4.4 Conclusion 116 Chapter 5 Conclusion 117 BIBLIOGRAPHY 121 VI SUMMARY Herbal medicine is gaining world wide popularity. The rapid increase in usage of herbal medicine has important medical and socio-economic implications. Assessment of the safety, quality and efficacy of herbal medicines becomes an important issue. In this study, HPLC and LC-MS methods for the analysis of complex mixtures of botanical origins are developed. A rapid and sensitive HPLC method for the detection of 4 biologically active chemical markers in Gastrodia elata, namely, gastrodin, 4hydroxbenzyl alcohol and 4-hydroxybenzaldehyde and L-pyroglutamic acid has been successfully developed. To the author’s knowledge, this is the first report of simultaneous detection of 4 active constituents in this medicinal plant. The contents of gastrodin, which is the main active ingredient, has also been determined in the plant and in the products (Chinese Proprietary Medicine, i.e. CPM). This method has been validated for linearity, precision, accuracy, LOD and LOQ. For the CPM, the presence of other plants further complicates the complex herbal matrix, resulting in the gastrodin peak being co-eluted with other herbal ingredients. Hence a SPE sample cleanup method is successfully developed and employed. In many medicinal plants, the active constituents may not be known. HPLC chromatographic fingerprinting can be used for quality control of herbal medicines, especially when the marker compounds are not available or not known. In the present work, a HPLC chromatographic fingerprinting method is successfully developed and applied to 34 botanical samples, constituting 11 different types of herbal medicines. VII Hierarchical analysis of the resulting chromatograms is performed. Results show that most of the Fangji samples in Singapore resemble the toxic Guangfangji samples. Successful differentiation between Fangji (Stephania tetrandra) and Guangfangji (A. fangchi) is achieved and results are confirmed by LC-MS/MS. This is the first report of comparative LC-MS/MS study of these two commonly mistaken herbs. This study indicates the urgent need for more stringent quality control of herbal medicine. Analysis of complex mixtures of botanical origins is very challenging. The work presented in this thesis has clearly demonstrated that methods for the analysis of some commonly used herbs in Singapore and around the region, have been developed and have been shown to be useful for the quality control of such botanicals and their products. The methods developed can be extended to other complex mixtures of botanical origins with appropriate optimization, when necessary. Academics/researchers, health professionals, consumers, regulators and people from industries must work together to ensure the safety and quality of herbal medicines. VIII LIST OF TABLES Table 2.1 Comparison of the 13C NMR spectral data of purified compound with those of gastrodin reported in literature 29 Table 2.2 Comparison of the retention times (RT) of the standards with those in the samples 35 Table 2.3 Intra-day and inter-day analytical precisions using five concentrations (20.0-200 µg/ml) of gastrodin 39 Table 2.4 Recoveries of gastrodin added into Gastrodia elata 42 Table 2.5 The contents of gastrodin in the samples and daily recommended dose 42 Table 2.6 CPM ingredients and their weight percentage 47 Table 2.7 Recoveries of gastrodin applied onto SPE cartridges 50 Table 2.8 Recoveries of gastrodin added into CPM samples 52 Table 2.9 The contents of gastrodin in four Chinese Proprietary Medicines 50 Table 3.1 Observations of five objects 61 Table 3.2 Similarity matrix (of observations in Table 3.1) 61 Table 3.3 Updated similarity matrix (from Table 3.2) 61 Table 3.4 Updated similarity matrix (from Table 3.3) 62 Table 3.5 Updated similarity matrix (from Table 3.4) 62 Table 3.6 List of Traditional Chinese Herbal Medicines and their sources 64 Table 3.7 Euclidean distance matrix of thirty-four Chinese herbs 73 Table 3.8 Retention time (RT) and relative retention time (RRT) of standards 83 Table 3.9 RT and RRT of target compound peak in the samples and their differences compared with the standard values 85 Table 4.1 Local sources of herbs bought as Fangji (Stephania tetrandra) 88 IX LIST OF FIGURES Figure 2.1 Chemical structures of some constituents of Gastrodia elata 12 Figure 2.2 HPLC chromatograms of (a) gastrodin without purification, and (b) gastrodin after purification. 24 Figure 2.3 UV spectrum of purified gastrodin 25 Figure 2.4 IR spectrum of purified gastrodin 25 Figure 2.5 Mass spectra of gastrodin 26 Figure 2.6 1 27 Figure 2.7 13 Figure 2.8 UV spectra of (a) gastrodin, (b) 4-hydroxybenzyl alcohol, (c) 4hydroxybenzaldehyde and (d) L-pyroglutamic acid Chromatograms of (a) gastrodin (GA), (b) 4-hydroxybenzyl alcohol (HA), (c) 4-hydroxybenzaldehyde (HD) and (d) extract of Gastrodia elata at 270nm Figure 2.9 H-NMR spectrum of gastrodin C-NMR spectrum of gastrodin 28 36 37 Figure 2.10 Chromatograms of (a) L-pyroglutamic acid (PGA) and (b) extract of Gastrodia elata at 195 nm Figure 2.11 Chromatograms of (a) gastrodin and (b) extract of Gastrodia elata at 40 224 nm Figure 2.12 Content of gastrodin determined in Gastrodia elata with four extraction methods 41 Figure 2.13 Solid phase extraction method for gastrodin in CPM samples. 48 Figure 2.14 Typical chromatograms of extracts of four Chinese herbal products: (a) Wild Gastrodia elata Tablet (CPM3), (b) Tian-Ma-Wan Pill (CPM4), (c) Capsules for the Stubborn Headache (CPM5) and (d) QiangLi-Tian-Ma-Du-Zhong Capsule (CPM6) 53 Figure 3.1 Euclidean distance between two data points in a two-dimentional measurement space defined by the measurement variables x1 and x2 58 Figure3.2 The distance between a data cluster and a point using single linkage, complete linkage and average linkage 60 38 X Figure 3.3 Average linkage dendrogram of the hypothetical data set (Table 3.1) 62 Figure 3.4 (a) Peak purity check by overlaying normalized spectra acquired in the upslope, apex, and downslope of the peak; (b) Overlaid UV spectra of peak at 35.6 min in Madouling (continuous line) and that of AAI standard (broken line); (c) typical chromatogram of an extract of MaDouLing Typical HPLC chromatogram of an extract of Huangqin 69 Figure 3.5 71 Figure 3.6 Dendrogram of thirty four Chinese herbs using average linkage method 76 Figure 3.7 Overlaid chromatograms of Fangji from (a) EYS, (b) WYN, (c) SINC and Guang FangJi from (d) Guangzhou and (e) Beijing 78 Figure 3.8 Overlaid chromatograms of extracts of Banlangen from (a) EYS, (b) WYN and (c) SINC 79 Figure 3.9 Chromatograms of extracts of DuZhong from (a) EYS, (b) WYN and (c) SINC 80 Figure 3.10 Overlaid chromatograms of extracts of Shudihuang from (a) EYS, 82 (b) WYN, and (c) SINC, and Shengdihuang from (d) EYS, (e) WYN and (f) SINC Figure 4.1 Chemical structures of tetrandrine, fangchinoline and aristolochic acid I 88 Figure 4.2 HPLC chromatograms of (a) reference standard of AAI and (b) reference herb of Guangfangji, and (c) Overlaid UV spectra of peak 1 (continuous line) and that of AAI (broken line). The RT and UV spectrum of peak 1 in Guangfangji match those of AAI 94 Figure 4.3 HPLC chromatograms of (a) reference standard of tetrandrine, (b) reference standard of fangchinoline, and (c) Fangji from Tongrentang, and UV spectra match of (d) peak 1 with reference standard of fangchinoline, (e) peak 2 with reference standard of tetrandrine 95 Figure 4.4 Overlaid chromatograms of reference Guangfangji, Guangfangji from Guangzhou and 10 local Fangji samples (Table 4.1) 98 XI Figure 4.6 (a) Overlaid chromatograms of the extracts of local Fangji sample from SINC and that of the reference Guangfangji. (b) Overlaid UV spectra of peak at 35.7 min of reference Guangfangji (continuous line) and that of AAI (broken line). (c) Overlaid UV spectra of peak 1 in Fangji sample from SINC (continuous line) and that of AAI (broken line). 100 Figure 4.7 Effect of vaporizer temperature on abundance of (a) [M+NH4]+ m/z 359 for AAI, and (b) [M+H]+ m/z 623 for tetrandrine and [M+H]+ m/z 609 for fangchinoline 101 Figure 4.8 Effect of capillary temperature on abundance of (a) [M+NH4]+ m/z 359 for AAI, and (b) [M+H]+ m/z 623 for tetrandrine and [M+H]+ m/z 609 for fangchinoline 103 Figure 4.9 Effect of tuber lens offset on abundance of (a) [M+NH4]+ for AAI, and (b) [M+H]+ for tetrandrine and fangchinoline 104 Figure 4.10 (a) Full scan mode MS spectrum of AAI; (b) MS/MS spectrum of AAI, with precursor ion, [M+NH4]+ m/z 359 106 Figure 4.11 Full scan mode MS spectra of (a) fangchinoline and (b) tetrandrine 107 Figure 4.12 MS/MS spectra of (a) fangchinoline with precursor ion m/z 609 and (b) tetrandrine with precursor ion m/z 623 108 Figure 4.13 LC-MS/MS analysis of reference Guangfangji: (a) Total ion 110 chromatogram (TIC), (b) Full scan mode mass spectrum, (c) Ion chromatogram of reference Guangfangji, [M+NH4]+ m/z 359, and (d) MS/MS spectrum. The data agrees with MS data from AAI (Figure 4.10), hence confirming the presence of AAI in the reference Guangfangji. Figure 4.14 LC-MS analysis of Fangji (Tongrentang): (a) Total ion chromatogram, (b) Mass spectrum of peak at 4.9 min, and (c) mass spectrum of peak at 4.4 min. The MS data agrees with those of the standards fangchinoline and tetrandrine (Figure 4.11), hence confirming the presence of these two standards in the extract of Fangji (Stephania tetrandra) from Tongrentang 112 XII Figure 4.15 LC-MS/MS analysis of Fangji (Tongrentang): (a) ion chromatogram, m/z 609; (b) MS/MS spectrum in full scan mode, precursor ion m/z 609; (c) ion chromatogram, m/z 623; and (d) MS/MS spectrum in full mode, precursor ion m/z 623. The data agrees with those obtained from the fangchinoline and tetrandrine standards (Figure 4.11 and 4.12), hence confirming the presence of these 2 standards in the extract of Fangji (Tongrentang). 113 Figure 4.16 LC-MS/MS analysis of local Fangji sample. (a) Total ion chromatogram; (b) full scan mode mass spectrum; (c) ion chromatogram, m/z 359; and (c) MS/MS spectrum, precursor ion m/z 359 115 XIII Chapter 1 Introduction Herbal medicines are plant-derived materials or preparations with therapeutic or other human health benefits which contains either raw or processed ingredients from one or more plants. In some traditions materials of inorganic or animal origin may also be present (WHO, 1998). 1.1 Importance of herbal medicines Over the past decades, herbal medicines have become a topic of increasing global importance, with both medical and economic implications (WHO, 1998a; Tyler, 1999). In many developing countries, herbal medicines have always played a central role in healthcare. Although modern medicine may be available in these countries, it is estimated that 65 to 80% of the populations depend on traditional medicines as the primary source of healthcare (Bannerman et al, 1983). In many circumstances, herbal medicines may be the only medicine available to populations in developing nations, due to the fact that modern pharmaceutical drugs are in great demand and are costly. Historically, all medical systems, including the medical system in developed countries, were once botanically based. In fact, herbal medicines were practiced worldwide relatively successfully before the advent of new synthetic drugs. Due to a lack of clinical data to establish the safety and efficacy of herbal medicine, the use of herbal medicines declined in the developed countries only during the 1940s to 1950s (Gail et al, 2001). In 1 recent years, the usage of complementary and alternative medicine (CAM), including herbal medicine, has increased (Koh et al, 2003 and Eisenberg et al, 1998). It has also been estimated that about 25% of all prescription drugs are derived either directly or indirectly from natural sources such as plants, bacteria, and fungi (Farnsworth and Morris, 1976; WHO, 2002). In developed countries, including Australia, Canada, Europe and the United States, a resurgence of interest in herbal medicines has resulted from the preference of many consumers for products of natural origin. In the U.S. alone, it is estimated that herbal usage increased by 380%, in the period between 1990 and 1997 (Eisenberg et al, 1998). According to WHO report (2002a), in USA, herbal sales increased by 101%, from US$ 292 million to US$ 587 million in mainstream markets between May 1996 and May 1998. This herbal renaissance has been fueled by strong consumer interest in natural therapies, preventative medicine, coupled with a disappointment with allopathic medicine, and the perception that herbal medicines are relatively safe and free from side effects. In France, 75% of the population has used complementary medicine at least once. In German, 77% of pain clinics provide acupuncture. In the United Kingdom, expenditure on complementary or alternative medicine stands at US$ 2.3 billion per year (WHO, 2002). In Australia, research has indicated that 48.5% of the population used at least one non– medically prescribed alternative medicine in 1993. The estimated national expenditure on alternative medicines and alternative practitioners is close to A$1 000 million per annum, of which A$621 million is spent on alternative medicines (Maclennan, et al, 1996). An Australian government report in 1996 estimated that there were at least 2.8 million traditional Chinese medicine consultations in 1996, representing an annual turnover of 2 A$84 million within the health economy. This growth was also reflected in a four–fold increase in the importation of Chinese herbal medicines since 1992 (WHO, 2000). The world market for herbal medicines based on traditional knowledge was estimated at US$ 60 billion a year, some 20 percent of the overall drug market (WHO, 2002a). In China, Traditional Chinese Medicine is fully integrated into China’s health system. 95% of Chinese hospitals have units for traditional medicine (WHO, 2002). Over 100 thousand herbal preparations have been recorded that are still in clinical use (WHO, 2000). Traditional medicine accounts for 30-50% of total consumption. There are 800 manufacturers of herbal products with a total annual output of US$ 1.8 billion (WHO, 2002). In 2000, the herbal medicine market in Japan was worth US$ 2.4 billion. An October 2000 survey showed that 72% of registered western-style doctors use kampo medicine (the Japanese adaptation of Chinese medicine) in their clinical services (WHO, 2002). A study conducted in Tokyo in 1990 showed that 91% of the survey population considered that oriental medicine was effective for chronic diseases, 49% had used herbal medicines. Japan now produces 210 Kampo herbal formulas according to strict quality controls, 147 of which are covered by health insurance (WHO, 2000). In Singapore, some people make use of herbal medicines as their alternative form of healthcare. A report (Singapore, 1995) by a review committee on Traditional Chinese Medicine showed that about 45% of Singaporeans had consulted Traditional Chinese Medicine practitioners in the past. A much smaller proportion of Malays (8%) and Indians (16%) had also consulted a Traditional Chinese Medicine practitioner. About 3 12% of daily outpatient attendance is estimated to be seen by Traditional Chinese Medicine Practitioners. It is estimated that there are about 8000 to 10000 CPM products in the market. They are readily available in the medical halls, supermarkets, and pharmacies etc. The number of Chinese medical halls that are registered with Registry of Companies and Businesses is estimated to be 900. 1.2 An increasing need for quality control of herbal medicines With the tremendous expansion in the use of herbal medicines worldwide, assurance of their quality, safety and efficacy have become an important concern for both health authorities and the public (Gail et al, 2001; Koh and Woo, 2000). For instance, the herb Ma Huang (ephedra) is traditionally used in China to treat short-term respiratory congestion. In the United States, the herb was marketed as a dietary aid, whose long-term use led to at least a dozen deaths, heart attacks and strokes. The U.S. Food and Drug Administration (2003) issued a consumer alert on the safety of dietary supplements containing ephedra. In Belgium, at least 70 people required renal transplant or dialysis for interstitial fibrosis of the kidney after taking the wrong herb from the Aristolochiaceae family, again as a dietary aid (Vanherwegnem et al, 1993). Unlike conventional drugs, herbal medicines present some unique problems in their quality standardization. These variables are caused by many factors such as species difference, organ specificity, seasonal variation, cultivation, harvest, storage, transportation, adulteration, substitution, contamination, post harvest treatment and manufacturing practices (Eskinazi, et al, 1999). Therefore, product variation in herbal 4 medicines can be significant. For example, the content of ginsenosides was examined in 50 commercial brands of ginseng sold in 11 countries (Cui et al, 1994). In forty-four of these products, the concentration of ginsenosides ranged from 1.9% to 9% (w/w); 6 products contained no ginsenosides and one of these contained large amounts of ephedrine. To guarantee batch-to-batch reproducibility of plant material and herbal products, standardization of herbal medicine is necessary. Active components of herbal medicines are often used as markers for quality control of herbal medicine (Thompson and Morris, 2001). 1.3 Analytical methods for quality control of herbal medicines Quality control directly impacts the safety and efficacy of herbal medicines (WHO, 2003). The entire process of product of herbal medicines, from raw materials to finished herbal products, need to be controlled. World Health Organization has developed a series of technical guidelines relating to the quality control of herbal medicines. These include: Guidelines for the assessment of herbal medicines (WHO, 1991), Good manufacturing practice (GMP): supplementary guidelines for the manufacture of herbal medicinal products (WHO, 1996), and Guidelines on good agricultural and collection practices [GACP] for medicinal plants (WHO, 2003). “Markers” was defined by WHO (1996) as “constituents of medicinal plant material which are chemically defined and of interest for control purposes”. Analysis of maker compounds in herbal medicines can be accomplished by colorimetric, spectroscopic or 5 chromatographic methods (Gail et al, 2001). Colorimetric and spectroscopic methods are older analytical procedures quantifying the absorption of structurally related compounds at a specific wavelength of light, and expressed as concentration of a reference compound, which is normally the active or major chemical constituent in that plant material. Since other unrelated plant constituents absorbing at the same wavelength will also be included in the measurement, a higher concentration can be erroneously ascribed to the test. There has been a decline in the use of these procedures in recent years (Gail et al, 2001). In recent years, chromatographic procedures have become the method of choice for the analysis of secondary chemical constituents. Thin-layer chromatographic (TLC) procedures have the advantage of being simple, rapid, can provide useful characteristic profile patterns, and are inexpensive to use (Wagner and Bladt, 1996). However, their resolving power is limited. Gas chromatography (GC) can provide high resolution of the more volatile complex mixtures, but is of limited value in the case of non-volatile polar compounds, especially the polar polyhydroxylated and glycosidic compounds (Gail et al, 2001). High-performance liquid chromatography (HPLC) is capable of resolving complex mixtures of polar and non-polar compounds, and has become the method of choice for the qualitative and quantitative analysis of herbal extracts and products. The use of liquid chromatography-mass spectrometry (LC-MS) and liquid chromatography mass spectrometry - mass spectrometry (LC-MS/MS) has rapidly increased in the last few years. In the past decade, the development of simple, reliable, LC/MS interfaces, most notably electrospray (ESI) and atmospheric pressure chemical ionization (APCI), has spurred the development and acceptance of LC/MS methods, such that, today, there are many laboratories that routinely use LC/MS as the primary analytical method (Tiller 6 et al, 2003; Hayen and Karst, 2003; Niessen, 2003). The advantage of these methods is that as each compound is being eluted, it is captured by the mass spectrometer and provides an immediate molecular ion and/or major mass fragment, which allows for positive identification of the eluting “peak” (Gail et al, 2001). Other hyphenated techniques, such as liquid chromatography combined with nuclear magnetic resonance spectrometry (LC-NMR) and LC-NMR-MS, have been developed in recent years for on-line determination of natural products (Wolfender et al, 2003). The advent of the diode array detection (DAD) has brought compound identification to HPLC. Previously the sole domain of gas or liquid chromatography-mass spectrometry (GC-MS or LC-MS) peak identification and purity checking can now be done as part of HPLC analysis, and at a lower cost (Ludwig and Stephan, 1993). HPLC techniques often use absolute or relative retention times to identify chromatographic peaks. With the DAD, spectra can be acquired automatically as each peak elutes. Under well-defined conditions, UV spectra are useful data for the confirmation of peak identity (Ludwig and Stephan, 1993). In recent years, capillary electrophoresis (CE) has expanded its scope and range of both instrumentation and application (Altria and Elder, 2004). Many CE methods have been developed for analysis of raw herbal materials and herbal products (Chen et al, 2004; Jiang et al, 2004; Feng et al, 2003). It was thought by many that CE would rapidly replace HPLC, but considerable previous investment in HPLC equipment purchases and training has created an analytical inertia that CE has found a difficult to change. CE often 7 offered only an alternative to HPLC, not an improvement, and therefore CE was not widely implemented (Altria and Elder, 2004). Besides quantifying marker compounds and chromatographic fingerprinting, quality control of herbal medicine also includes screening for western drug adulterants (Liu et al, 2001, Koh and Woo, 2000) and determination of toxic heavy metals (Koh and Woo, 2000). A combination of analytical methods is often required in the quality control of herbal medicine. 1.4 Objectives The broad objective of this project is to develop analytical methods for the analysis of complex mixtures of botanical origins, in particular, for the quality control of some commonly used herbs in Singapore and around the region. Due to their complexity, quality control of herbal medicines is a challenge. The combination of several of analytical methods is often needed. In this study, we use bioactive components as markers for quality control of herbal medicines (e.g. Gastrodia elata) (Chapter 2). However, when the bioactive components are not available or not known, or they do not adequately reflect the total effects of the herbs, the whole chromatographic fingerprints of the extracts will be very useful for the differentiation of botanicals (Chapter 3). For Stephania tetrandra and Aristolochia fangchi herbs, a combination of HPLC-DAD and LC-MS/MS methods will be developed in this study (Chapter 4). 8 Thence, the specific objectives of this study are to 1) develop HPLC methods for the chemical analysis, and identification of G. elata; 2) develop HPLC methods for chromatographic fingerprinting analysis of some Chinese herbs which are commonly used in Singapore; and 3) develop HPLC-DAD and LC-MS/MS methods for differentiation of some Chinese herbs. 4) apply the developed methods to the analysis of selected herbal medicines and their products. 9 Chapter 2 Analysis of Gastrodia elata by HPLCDAD 2.1 Introduction Rhizoma Gastrodiae, Tianma （天麻）, is the dried tuber of Gastrodia elata Blume (Orchidaceae). The tuber is collected from winter to early spring, washed clean, steamed thoroughly and then dried at a low temperature (Tang and Eisenbrand, 1992). G. elata was listed in the ancient Shennong Bencao Jing (ca. 100 A.D.) and was later classified by Tao Hong as a superior herb, meaning that it could be taken for a long time to protect the health and prolong life (as well as treating illnesses). Now it is officially listed in the Chinese Pharmacopoeia (2000). The traditional use of G. elata is to calm internal wind and dispel invading wind, and invigorate circulation in the meridians; thereby used as an anticonvulsant, analgesic, and sedative against vertigo, general paralysis, epilepsy, and tetanus. In this thesis, G. elata will be used to refer to Rhizoma Gastrodiae, unless otherwise specified. 2.1.1 Chemical constituents The tuber of G. elata contains mainly phenolic compounds (Figure 2.1). Gastrodin (1) was isolated as the first active principle from G. elata (Zhou et al, 1979; Feng et al, 1979). 4-hydroxybenzyl alcohol (2), which is the aglycone of gastrodin, and 4hydroxybenzaldehyde (3) were also isolated from the ethanol extracts (Zhou et al, 1979a). From the fresh tuber of G. elata, another 5 phenolic compounds were isolated and identified as bis(4-hydroxyphenyl)methane (4), bis(4-hydroxybenzyl)ether (5), 3,410 dihydroxybenzaldehyde (6), 4-ethoxymethyl phenyl 4’-hydroxybenzyl ether (7), and 4hydroxybenzyl ethyl ether (8) (Zhou et al, 1980). Taguchi et al also reported gastrodioside (9), 4-hydroxybenzyl methyl ether (10), 4-(4’-hydroxybenzyl)benzyl methyl ether (11), and parishin (12) were isolated from the tube of G. elata (Taguchi et al 1981). Another bis(4-hydroxybenzyl) derivative, 2,4-bis(4-hydroxybenzyl)phenol (13), was isolated from the methanolic extract (Naoki et al, 1995). S-(4- hydroxybenzyl)glutathione (14) was isolated from water extracts (Andersson, et al, 1995). Parishin B (15) and parishin C (16) were isolated from 70% methanol extract (Lin et al 1996). From the ethyl ether fraction prepared from the methanol extract, 4,4’dihydroxybenzyl sulfoxide (17) (Yun-Choi and Pyo, 1997) and cirsiumaldehyde (18) (Yun-Choi, et al, 1997) were isolated. Another two 4-hydroxybenzyl alcohol derivatives, 3-O-(4’-hydroxybenzyl)-β-sitosterol (19) and 4-[4’-(4’’- hydroxybenzyloxy)benzyloxy]benzyl methyl ether (20) were isolated from the methanol extract of fresh tuber (Yun-Choi, et al, 1998). Gastrol (21) was isolated from the MeOH extract of the rhizomes of G. elata (Hayashi, et al, 2002). The tuber of G. elata also contains other compounds (Tang and Eisenbrand, 1992), such as succinic acid (24), sucrose (25), β-sitosterol (26), daucosterol (27), citric acid, 2methyl ester (28), palmitic acid (29), citric acid (30), and citric acid 1,5-dimethyl ester (31). L-pyroglutamic acid (32) was isolated from the tuber of G. elata collected in Guizhou (Hao, et al, 2000). α-acetylamino-phenylpropyl α-benzoylamino- phenylpropionate (33) was recently isolated (Xiao, et al, 2002). 11 O CH2 CH2OH O OH HO HO HO CH2OH CHO OH gastrodin (1) 4-hydroxybenzyl alcohol (2) 4-hydroxybenzaldehyde (3) CHO CH2 CH2 O HO OH bis(4-hydroxyphenyl)methane (4) O EtO CH2 HO HO OH OH bis(4-hydroxbenzyl)ether (5) 3,4-dihydroxybenzaldehyde (6) CH2 OEt CH2 CH2 OH HO 4-ethoxymehtyl phenyl 4’-hydroxybenyl ether (7) 4-hydroxybenzyl ethyl ether (8) CH2OH OO CH2OCH2 OH OH HO HO CH2 OCH3 OH gastrodioside (9) 4-hydroxybenzyl methyl ether (10) OH O CH3O CH2 CH2 CH2 OH 4-(4’-hydroxybenzyl)benzyl methyl ether (11) CH2 HO OH 2,4-bis(4-hydroxybenzyl) phenol (13) Figure 2.1 Chemical structures of some constituents of G. elata 12 CH2 HO C COOR1 R: COOR2 CH2 CH2 COOR3 O HO O CH2 S R N H O OH HO HN (12) (15) (16) HO OH O CH2 S CO 2 H O OH parishin parishin B parishin C NH 2 S HO R1=R2=R, R3=H R1=R3=R, R2=H R1=R2=R3=R CO 2 H S-(4-hydroxybenzyl)glutathione (14) OHC O CH2 O CH2 4,4’-dihydroxybenzyl sulfoxide (17) O CH2 CHO cirsiumaldehyde (18) Et Me H Me Me Pri H H CH2OCH3 HO H CH2O O CH2O HO 3-O-(4’-hydroxybenzyl)-β-sitosterol (19) 4-[4’-(4’’-hydroxybenzyloxy)benzyloxyl benzyl methyl ether (20) OMe OMe HO HO O HO OH gastrol (21) OH CHO vanillin (22) CH2OH vanillin alcohol (23) Figure 2.1 (continued) 13 Et Me HO OH OH HO H Me OH O Me Pri H OH HO HO2C CH2 O O CH2 CO2H H OH succinic acid (24) H HO β-sitosterol (26) sucrose (25) Et Me Me Me Pri H H H H O O HO CH2 C O OH HO2C HO CH2 C CH2 CO2H HO2C OH OH daucosterol (27) citric acid, 2-methyl ester (28) O CO2H HO2C OMe CH2 C CH2 CO2H MeO C CH2 C palmitic acid (29) O CH2 C OMe O H N CO2H OH OH citric acid (30) citric acid 1,5-dimethyl ester (31) O C CO2H (CH2)14 CH3 L-pyroglutamic acid (32) O NH O CH C HN O CH2 CCH3 CH2 CH CH2 α-acetylamino-phenylpropyl α-benzoylamino-phenylpropionate (33) Figure 2.1 (continued) 14 2.1.2 Biological activities 2.1.2.1 Biological activities of G. elata Koang et al (1958), first reported that G. elata may be effective in the treatment of epilepsy. The aqueous extract of G. elata, given intravenously, increased the threshold of electroshock-induced convulsions and inhibited the occurrence of seizures in experimentally epileptic guinea pigs. It also exhibited sedative effects in both the rat and mouse (Tang and Eisenbrand, 1992). G. elata administered orally for 1 week could improve the scopolamine-induced learning and memory deficit in rats (Hsieh et al, 2000). G. elata had anticonvulsive and free radical scavenging activity. It could significantly reduce kainic acid (KA)-induced lipid peroxide level in vitro, and delay the onset of wet dog shakes, paw tremor and facial myoclonia in KA-treated rats (Hsieh et al, 2001). Kim et al(2001), reported that the ether fraction of methanol extracts of G. elata had anticonvulsive effect and putative neuroprotective effect against excitotoxicity induced by kainic acid. Huh et al (1998) also reported the methanol extracts of G. elata could significantly inhibited the convulsion state as well as the level of lipid peroxidition in pentylenetetrazole (PTZ)-treated rat brain. Ha et al (2000), reported that the ethyl ether fraction of a methanol extract of G. elata shortened the recovery time from and inhibited the severity of pentylenetetrazole (PTZ)-induced convulsions in rats. Pretreatment with this fraction prevented the decrease of brain GABA in rats given subconvulsive doses of PTZ. It was also reported that the extract of G. elata had muscle relaxant effect in isolated guinea pig ileum (Lin, et al, 1997). G. elata could also improve the response of condition 15 taste aversion, increase spontaneous locomotion, and enhance the ability of learning and memory in water-maze in mice after the rotation. The symptoms of motion sickness induced by rotation could be improved by G. elata treatment (Wang, et al, 1999). The ethyl ether fraction of G. elata could dramatically reduce amyloid beta-peptide induced neuronal cell death in vitro (Kim, et al, 2003). 2.1.2.2 Biological activities of selected constituents of G. elata The activities of gastrodin, 4-hydroxylbenzyl alcohol, L-pyroglutamic acid and others have also been reported. Gastrodin and 4-hydroxybenzyl alcohol Gastrodin and its genin, 4-hydroxybenzyl alcohol, showed sedative activity in mice, monkeys, rabbits, and human subjects. In addition, intravenously administered gastrodin and its genin had anticonvulsant activity in treatment of experimental epilepsy of the guinea pig (Tang and Eisenbrand, 1992). Both compounds were not toxic to mice when given orally or intravenously at doses below 5 g/kg. It was also reported that gastrodin and 4-hydroxybenzyl alcohol could facilitate memory consolidation and retrieveal (Hsieh, et al, 1997) and the facilitating effects of 4-hydroxybenzyl alcohol on learning and memory were better than those of gastrodin. Further studies showed that 4hydroxybenzyl alcohol might act through suppressing dopaminergic and serotonergic activities and thus improve learning. The physiological disposition of 3H-labeled gastrodin was investigated in rats. The decline in radioactivity from the gastrointestinal tract was rapid following oral 16 administration of gastrodin and only 1.1% of the dose was recovered from the gastrointestinal tract after 8 h. In rats given gastrodin intragastrically, the radioactivity level in blood was moderate at 5 min and reached its peak at 50 min after administration. Radioactivity was highest in kidneys, moderate in liver, lung, and uterus, and relatively lower in the brain, reaching a maximum at 2 h in the brain. The main metabolite of gastrodin detected by thin-layer chromatography was its genin, 4-hydroxybenzyl alcohol. The pharmacokinetics of gastrodin in rats reflected a circadian rhythm (Tang and Eisenbrand, 1992). 4-hydroxybenzaldehyde 4-hydroxybenzaldehyde showed an inhibitory effect on the lipid peroxidation. In the brains of pentylenetetrazole (PTZ) treated rats, the brain lipid peroxidation was significantly increased, while it recovered to the control level after treatment with 4hydroxybenzaldehyde. Furthermore, 4-hydroxybenzaldehyde showed an inhibitory effect on the GABA (γ-aminobutyric acid) transaminase (Ha et al, 2000). A recent study (Kim et al, 2003) showed that the ethyl ether fraction of G. elata dramatically reduced amyloid β-peptide induced neuronal cell death in vitro. HPLC analysis demonstrated that this fraction contained mainly 4-hydroxybenzaldehyde. L-pyroglutamic acid It is reported that L-pyroglutamic acid has anticonvulsive effect (Dusticier et al, 1985). A competitive inhibition of the high affinity transport of glutamic acid by L-pyroglutamic acid was found in vitro with no effect on the uptake of γ-aminobutyric acid (GABA). 17 Other ingredients Vanillin (22) and vanillyl alcohol (23) have been reported to have anti-oxidant and free radical scavenging activities. Vanillin is a powerful scavenger of 1,1-dipheny-2-picry hydrazyl, superoxide and hydroxyl radicals and inhibits iron-dependent lipid peroxidation in rat brain homogenate, microsomes and mitochondria (Liu and Mori, 1993). Vanillyl alcohol has anticonvulsive and suppressive effects on seizures and lipid peroxidation induced by ferric chloride in rats. Vanillyl alcohol also has free radical scavenging activities, which may be responsible for its anticonvulsive properties (Hsieh et al, 2000). S-(4-hydroxybenzyl)glutathione was isolated as the major principle responsible for the inhibition of the in vitro binding of kainic acid to brain glutamate receptors by water extracts of G. elata (Andersson et al, 1994). 4-hydroxy-3-methoxybenzaldehyde inhibited potently the activity of GABA transaminase (Ha JH et al, 2001). The antithrombotic effect of citric acid 1,5 di-methyl ester was observed with prolonging the bleed time in thrombin-induced thrombosis model of mice (Pyo et al, 2000). 2.1.3 Objectives The objectives of the study are to develop high performance liquid chromatography (HPLC) and solid phase extraction (SPE) methods using some bioactive constituents as markers for indication of the authenticity of G. elata, and to determinate the major constituent, gastrodin, in G. elata and in Chinese Proprietary Medicines (CPMs) containing G. elata. It has been reported (Huang, 2000) that some faked samples of G. elata have been prepared from some other herbs or plants, such as Colocasia esculenta (L.) Schott (Yunai, 芋 艿 ), Dahlia pinnata Cav. (Daliju, 大 理 菊 ), Mirabilis jalapa 18 L.(Zimoli, 紫茉莉), Cacalia tangutica (Franch) Hand.-Mazz. (Yuliexiejiacao, 羽裂蟹甲 草), Solanum tuberosum L. (Malingshu, 马铃薯), Ipomoea batatas (L.) Lam. (Ganshu, 甘 薯 ), Canna edulis Ker-Gawl. (Bajiaoyu, 芭 蕉 芋 ), Trichosanthes kirilowii Maxim. (Gualou, 栝楼), and Dobinca delavayi (Baill.) Engl. (Duobinqi, 多槟槭). Although the faked samples resemble like G. elata, their ingredients are quite different from those of G. elata. Therefore chemical profiling of G. elata would be useful for discerning it from its fakes and for quality control of this herb. 19 2.2 Preparation of gastrodin by HPLC 2.2.1 Introduction The reference standards of herbal ingredients are often unavailable commercially. Even when available, the cost is often prohibitive, as it is difficult to obtain the pure standard in large quantity. In this study, gastrodin was obtained from Kunming Pharmacy Industry (Kunming). However, HPLC analysis showed that the compound is not of acceptable purity. The purpose of this section is to develop a semi-preparative HPLC method to purify this chemical and to confirm its structural characteristics based on UV, IR, ESIMS, 1H-NMR and 13C-NMR. The purity was checked by its m.p. and HPLC analysis. 2.2.2 Experimental 2.2.2.1 Materials and reagents Gastrodin (raw material, 63.9%) was provided by Kunming Pharmacy Industry (Kunming, China); Methanol and acetonitrile were of HPLC grade. Milli Q water (Millipore, France) was used. Methanol-d4 was from Sigma (St. Louis, MO, USA). 2.2.2.2 HPLC conditions HPLC was performed using a Agilent 1100 HPLC series chromatograph equipped with Chem-Station Software version Rev.A.08.03, a degasser model G1322A, a QuatPump model G1311A, a column oven model G1316A, an autosampler model G1313A and a diode array detector model G1315B. 20 Gastrodin was prepared on a semi-preparative column, Inertsil Agilent ZORBAX SBC18 (9.4 × 250mm, 5µm), guarded by a ZORBAX cartridge column (9.4 mm ID × 15 mm). A mixture of Milli Q water and methanol (90:10) was used as mobile phase. Flow rate was 2.5ml/min. Detection wavelength was 270nm. To evaluate the purity of the gastrodin prepared, an Inertsil ODS-3 column (4.6 mm × 250mm, 5 µm) was used, operated at 35°C. The mobile phase was Milli Q water (A) and acetonitrile (B) with a gradient program as follows: 5%B to 30% B (15 min), 30% B to 60% B (10 min), 60% B (5 min), post-run (10 min) at a flow rate of 1ml/min. The injection volume was 5 µl. Detection wavelength was 224nm. 2.2.2.3 Melting point A Bausch and Lamb & hot-stage microscopy was used to determine melting points (uncorrected). 2.2.2.4 Infrared spectrometry IR spectrum (KBr disc) was recorded on JASCO FT/IR-430 spectrometer (Japan). 2.2.2.5 Ultraviolet spectrometry UV spectra were recorded on an Agilent G1315B diode array spectrometer (USA). Scan range was set from 190 to 400 nm. 21 2.2.2.6 Mass spectrometry MS was measured on a Finnigan LCQ ion trap mass spectrometer (San Jose, CA). Electrospray interface was used. 2.2.2.7 Nuclear magnetic resonance spectrometry NMR spectra were recorded by Bruker Advance DPX300 NMR Spectrometer (Germany) [300 MHz (1H) and 75 MHz (13C)] using methanol-d4 solution with tetramethylsilane (TMS) as an internal standard. 2.2.3 Results and discussion Gastrodin (1) was prepared by the semi-preparative HPLC method as white powder. The melting point was measured as 156-157°. This agreed with the reported value (Taguchi et al, 1981). The purity was also checked by HPLC. Figure 2.2b shows the chromatogram of gastrodin after purification. The percentage area of gastrodin peak with respect to the total area was 63.9% before purification (Figure 2.2a) and increased to 98.2% after purification (Figure 2.2b). The ultraviolet spectrum exhibited absorption maxima λmax at 195, 220 and 270 nm (Figure 2.3). Such absorption was due to “OR” substituted benzene ring. Benzene absorbs at 184, 203 and 254 nm. When benzene ring is substituted by lone pair donating system, the wavelength of the absorption maximum will increase. Absorption bands in the IR spectrum (Figure 2.4) indicated that hydroxyl group (36003200 cm-1) was present. Due to the hydrogen bond, O-H stretching frequency is shown as 22 the broad band in the range of 3600-3200 cm-1. The absorption band at 1240 can be due to O-H bending; while the band at 1076 cm-1 is due to C-O stretching (Nyquist, 2001). The two bands at 1612 and1513 cm-1 indicated the presence of aromatic ring system. The band at 829 cm-1 is due to the out-of-plane C-H bending vibration. For para-disubstituted benzene ring system, the frequency of out-of-plane C-H bending is in the range of 860 to 800 cm-1. The ammonium adduct molecular ion [M+NH4]+, m/z 304, can be detected by ESI-MS (Figure 2.5a). This ion was selected and further fragmented by collision induced dissociated (CID) to product ions. Figure 2.5b shows the characteristic product ions: [(M+NH4)-NH3]+ 287, [(M+NH4)-NH3-H2O]+ 269, and [(M+NH4)-NH3-glc]+ 107. The 1H NMR spectrum (Figure 2.6) showed signals due to four aromatic protons (δH 7.07, 7.26, each 2H, d), the protons on the sugar unit (δH, 3.3-3.5, 4H, m, and 3.6-4.2, 2H, m), and the CH2 protons (δH, 4.53, 2H, s). The 13 C NMR spectrum of 1 (Figure 2.7) showed signals due to an aromatic ring (δC, 137.4, 130.0 [2×C], 118.5 [2×C], 159.3), a sugar unit (δC, 103.0, 78.9, 78.8, 75.7, 72.2, 63.3) and (δC, 66.7). Comparison of 13C NMR of compound 1 with those of the literature (Table 2.1) revealed that compound 1 was gastrodin. 23 mAU 7.036 100 (a) 9.151 80 60 40 20 0 0 5 10 15 20 25 10 15 20 25 min 7.035 mAU (b) 120 80 40 0 0 5 min Figure 2.2 HPLC chromatograms of (a) gastrodin without purification, and (b) gastrodin after purification. 24 mAU 1750 1500 1250 1000 750 500 250 0 200 225 250 275 300 325 350 375 nm Figure 2.3 UV spectrum of purified gastrodin Figure 2.4 IR spectrum of purified gastrodin 25 (a) 304 100 90 Relative Abundance 80 70 60 50 40 30 20 10 0 100 (b ) 100 150 200 250 300 m/z 350 400 450 500 107 304 90 80 70 Relative Abundance 26 9 60 50 40 30 287 20 10 0 100 150 200 250 300 m /z 350 400 450 500 Figure 2.5 Mass spectra of gastrodin 26 Figure 2.6 1H-NMR spectrum of gastrodin 27 Figure 2.7 13C-NMR spectrum of gastrodin 28 6' CH2OH O O 5' 4' OH HO 4 1 7 CH2OH 1' 3' 2' OH Table 2.1 Comparison of the 13C NMR spectral data of purified compound with those of gastrodin reported in literature (Taguchi et al, 1981). Chemical Shift (δ ppm) a Carbon Number Measureda Reference data 1 137.4 136.6 2, 6 130.0 129.4 3, 5 118.5 117.7 4 159.3 158.4 7 66.7 65.7 1’ 103.0 102.4 2’ 75.7 74.9 3’ 78.9 78.0 4’ 72.2 71.3 5’ 78.8 77.9 6’ 63.3 62.5 75 MHz, methanol-d6 29 2.2.4 Conclusion A semi-preparative HPLC method has been successfully developed to purify gastrodin from the raw material purchased. The structure of the purified substance was confirmed by UV, IR, MS and NMR, and the spectroscopic data was consistent with the literature data for gastrodin. The melting point was consistent with reported values and the percent area of the purified peak to the total area was 98.2%. The purified gastrodin was therefore found suitable to be used as a standard in subsequent studies. 30 2.3 Detection and Determination of Bioactive Constituents in Gastrodia elata by HPLC-DAD 2.3.1 Introduction The quality control of herbs is very important to ensure the quality of the final products. Active constituents are often used as markers to indicate the authenticity of a herb or its product. Using one or several components as markers, several methods have been reported to analyze G. elata, such as high-performance liquid chromatography (HPLC) (Wei et al, 2001; Li et al, 2001), Gas chromatography (GC-MS) (Li et al, 2001a) micellar electrokinetic capillary chromatography (MEKC) (Ku et al, 1998 and 1998a) and capillary electrophoresis (CE) (Zhao et al, 1999; Wang et al, 2002). In today’s analytical laboratories, HPLC is commonly used, and diode array detection (DAD) is also used for routine analysis. The DAD can provide multiple-signal detection and peak identification. To date, no report on simultaneous determination of the four bioactive constituents as markers in G. elata can be found. In this study, a rapid and sensitive HPLC-DAD method has been developed, using the four active constituents as markers for analysis of G. elata. Quantitative analysis of gastrodin was performed at 224 nm in G. elata and two CPMs containing 100% G. elata. 31 2.3.2 Experimental 2.3.2.1 Materials and reagents 4-hydroxybenzyl alcohol, 4-hydroxybenzaldehyde and L-pyroglutamic acid of analytical grade were obtained from Sigma (St. Louis, United States). Gastrodin was provided by Kunming Pharmacy Industry (Kunming, China) and purified (Section 2.2) before use. Methanol and acetonitrile were of HPLC grade. Milli Q water (Millipore, France) was used. Three herbs of dried and sliced G. elata were purchased from three local medical stores, and labeled as GE1, GE2 and GE3. Two different Chinese Proprietary Medicines (CPMs) in the form of capsules, containing 100% of G. elata, were bought locally and labeled as CPM1 and CPM2. 2.3.2.2 HPLC conditions The analysis was performed using an Agilent 1100 HPLC series. The column used was an Inertsil ODS-3 column (4.6 mm × 250mm, 5 µm), maintained at 35°C. The mobile phase was Milli Q water (A) and acetonitrile (B) with a gradient program as follows: 5%B to 30% B (15 min), 30% B to 60% B (10 min), 60% B (5 min), post-run (10 min) at a flow rate of 1ml/min. The injection volume for all samples was 5 µl. Simultaneous monitoring was performed at 195 nm (for L-pyroglutamic acid), 270 nm (for gastrodin, 4hydrobenzyl alcohol, 4-hydrobenzaldehyde) and 224 nm (for gastrodin). Spectra were recorded from 190 to 400 nm. 32 2.3.2.3 Optimization of extraction Extractions were carried out either with 100% (v/v) or 70% (v/v) of methanol using a Soxhlet extractor or with triple extractions in an ultrasonic bath. 2 g of powdered samples for each group were accurately weighed. Soxhlet extractions were performed using 75 ml of solvent for 6 hours. In the ultrasonic bath extraction, samples were extracted for 20 min with 30 ml of solvent. The extract was filtered, and fresh solvent was added. This procedure was performed in triplicate. The filtrates were combined and the solvents were evaporated off under reduced pressure. The residue was dissolved in 70% methanol, quantitatively transferred to 10 ml volumetric flasks and made up to volume. The solution was filtered through 0.45µm nylon membrane filter before HPLC analysis. 2.3.2.4 Calibration curve Gastrodin standard was dissolved in methanol-water (30:70, v/v) yielding concentrations of 20, 50, 80, 140 and 200µg/ml. The solutions were filtered through a 0.45 µm membrane filter. Three different sets of solutions were prepared and analyzed each day and in three continuous days. Each calibration curve was fitted by linear regression. 2.3.2.5 Validation The linearity was determined for the calibration curve obtained by LC analysis of gastrodin. The slop and other statistics of the calibration curve were determined by linear regression. 33 The intermediate precision was evaluated on the same day, while the inter-day precision was assessed for 3 consecutive days. The data were expressed as the relative standard deviation (RSD %). A standard solution of gastrodin was sequentially diluted and analyzed using HPLC. The limit of detection (LOD) was based on the signal-to-noise (S/N) ratio of 3, while the limit of quantification (LOQ) was base on the signal-to-noise (S/N) ratio of 10 (Snyder et al, 1997). The accuracy was evaluated through recovery studies by adding known amounts of gastrodin at three different levels to the sample. The unspiked samples and each of the spiked samples were analyzed in triplicate and the recoveries were determined. 2.3.3 Results and discussion 2.3.3.1 Separation and detection of gastrodin, 4-hydroxybenzyl alcohol, 4hydroxybenzyladehyde, and L-pyroglutamic acid Figure 2.8 shows the UV spectra of the four compounds, GA, HA, HD and PGA. Three wavelengths have been selected for optimal sensitivity and selectivity: 195nm for PGA, 270 nm for GA, HA and HD, and 224 nm for quantitative analysis of GA. The method included extraction of gastrodin from CPM samples, followed by a separation on HPLC using a reversed-phase column and detection by a DAD. Three signals were recorded simultaneously over the range from 190 to 400 nm. Gastrodin, 4-hydroxybenzyl alcohol and 4-hydroxybenzaldehyde can be detected at 270 nm (Figure 2.9), and L-pyroglutamic acid can be detected at 195 nm (Figure 2.10). To identify peaks in chromatograms, 34 retention times of chromatographic peaks were compared with those of the four reference standards (Table 2.2). Table 2.2. Comparison of the retention times (RT) of the standards with those in the samples RT [Mean±SD (min)] (n=3) Sample GA HA HD PGA Standards 7.016±0.018 10.128±0.012 16.698±0.024 3.773±0.006 GE1 7.015±0.025 10.126±0.026 16.698±0.040 3.773±0.006 GE2 7.018±0.022 10.129±0.017 16.699±0.043 3.774±0.006 GE3 7.010±0.039 10.120±0.034 16.698±0.034 3.772±0.007 CPM1 7.018±0.020 10.130±0.015 16.689±0.045 3.774±0.006 CPM2 7.016±0.025 10.128±0.011 16.694±0.047 3.772±0.006 35 (a) (b) mAU mAU 150 1500 100 1000 50 500 0 0 200 250 300 350 nm (c) 200 250 300 350 250 300 350 nm (d) mAU mAU 400 150 300 100 200 50 100 0 0 200 250 300 350 nm 200 nm Figure 2.8 UV spectra of (a) gastrodin, (b) 4-hydroxybenzyl alcohol, (c) 4hydroxybenzaldehyde and (d) L-pyroglutamic acid 36 (a) mAU GA 80 0 0 5 10 15 20 25 min 15 20 25 min 20 25 min 20 25 min (b) mAU 40 HA 20 0 0 5 10 (c) mAU 40 HD 20 0 0 5 10 15 (d) mAU 20 GA 15 HD 10 HA 5 0 0 5 10 15 Figure 2.9 Chromatograms of (a) gastrodin (GA), (b) 4-hydroxybenzyl alcohol (HA), (c) 4-hydroxybenzaldehyde (HD) and (d) extract of G. elata at 270 nm 37 (a) mAU 300 200 PGA 100 0 0 5 10 15 20 25 min 10 15 20 25 min (b) mAU 600 PGA 400 200 0 0 5 Figure 2.10 Chromatograms of (a) L-pyroglutamic acid (PGA) and (b) extract of G. elata at 195 nm 38 2.3.3.2 Quantitative analysis of gastrodin Quantitative HPLC determination of gastrodin was performed at its maximum absorbance wavelength, 224 nm (Figure 2.11). Extractions with 70% methanol by Soxhlet extraction was found to afford the highest yield of gastrodin in the samples (Figure 2.12). Calibration curves for gastrodin were linear in the concentration range 20200 µg/ml. The LOD and LOQ were 0.70µg/ml and 2.57µg/ml respectively. The precision of the method was evaluated. The intra-day and inter-day RSD values were less than 1.31% (Table 2.3). Table 2.3 Intra-day and inter-day analytical precisions using five concentrations (20.0200 µg/ml) of gastrodin RSD (%) (n=3) Concentration (µg/ml) Intra-day Inter-day 20.0 0.22 0.61 50.0 0.06 0.78 80.0 0.47 1.31 140.0 0.92 0.86 200.0 0.35 0.85 39 (a) mAU 1600 GA 1200 800 400 0 0 5 10 15 20 25 min 15 20 25 min (b) mAU GA 120 80 40 0 0 5 10 Figure 2.11 Chromatograms of (a) gastrodin and (b) extract of G. elata at 224 nm 40 Content of gastrodin determined . (mg/g) 5 4.5 4 Ultrasonic Extraction Soxhlet Extraction 3.5 3 2.5 2 1.5 1 0.5 0 100%MeOH extraction 70% MeOH extraction Figure 2.12 Contents of gastrodin determined in G. elata using four extraction methods 41 The accuracy was investigated with recovery test. The recovery of the method was determined with the standard addition method for gastrodin in G. elata with results ranging from 91.4% to 101.9% (Table 2.4). Table 2.4 Recoveries of gastrodin added into G. elata Amount added Amount measured Recovery Mean±SD RSD (mg) (mg) (%) (%) (n=3) (%) 4.91 4.91 100.0 99.3±1.0 1.01 4.89 4.88 99.8 4.85 4.85 98.2 10.04 10.23 101.9 98.9±2.6 2.63 9.99 9.67 96.8 10.10 9.91 98.1 15.09 15.06 99.8 95.5±4.2 4.40 15.52 14.18 91.4 15.53 14.81 95.4 Table 2.5 The contents of gastrodin in the samples and daily recommended dose Sample Concentration mean±SD Daily recommended dose (g) Calculated daily dose of gastrodin (mg) (mg/g) (n=3) Herb GE1 4.07±0.01 3 to 9* 12.21 to 36.63 Herb GE2 3.48±0.01 3 to 9* 10.44 to 31.32 Herb GE3 4.49±0.02 3 to 9* 15.87 to 40.41 CPM1 2.97±0.02 2 capsules three times a day 7.07 CPM2 6.17±0.12 2 capsules three times a day 18.70 * Daily recommended dose from reference (Pharmacopoeia of the People’s Republic of China, 2000) 42 In this study, the contents of gastrodin in the three herbal samples were from 3.48 to 4.49 mg/g (Table 2.5). The gastrodin contents in G. elata could vary, depending on the collection seasons and culturing areas. A previous report showed that the average content of gastrodin was 3.1 mg/g in September samples, 2.3 mg/g in December samples and 9.3 mg/g in July samples. It was also reported (Zhang and Liu, 1983) that gastrodin content in G. elata ranged from 1.6 mg/g to 11.8 mg/g, as determined in the samples from various areas. CPM2 was found to contain twice as much gastrodin as CPM1 (Table 2.5), although both claimed to contain 100% G. elata powder. Nowadays, G. elata has been widely cultured. However, the content of gastrodin in cultured G. elata is normally less than those in the G. elata grown wild (Zhu, 1998). CPM2 was claimed to be prepared from G. elata grown wild. The Chinese Pharmacopoeia (2000) states that the recommended daily dose of G. elata is 3 to 9g. The concentrations of gastrodin found in the 3 herbal samples range from 3.48 to 4.49 mg/g. This corresponds to a daily dose of gastrodin ranging from 12.2 to 15.9 mg per day (Table 2.5). The two CPMs are capsules containing 100% of powdered G. elata as labeled. The frequency of administration is 2 capsules, three times a day for each of the CPM. From the concentration of gastrodin found in the CPM samples and the weight of contents per capsule, the daily dose of gastrodin can be calculated (Table 2.5). Consuming CPM1 delivers 7 mg of gastrodin per day while consuming CPM2 delivers 18.7 mg of gastrodin. The calculated dose of gastrodin from CPM2 falls within the calculated daily dose of gastrodin according to the dosage of G. elata given in the Chinese Pharmacopoeia. 43 2.3.4 Conclusion Using four bioactive components as markers, a rapid and sensitive HPLC-DAD method has been developed for analysis of G. elata and for the determination of gastrodin, its major bioactive constituent. This validated method was applied on three locally bought G. elata and two CPMs containing 100% G. elata. The CPM claimed to be prepared from G. elata grown wild has more than twice the content of gastrodin than the one from cultured G. elata. 44 2.4 Determination of Gastrodin in Chinese Proprietary Medicines Containing Gastrodia elata by Solid Phase Extraction and HPLC 2.4.1 Introduction G. elata has been traditionally used in oriental countries for a few thousand of years. And there are several traditional formulas with this herb that come to us today. Most of the ancient formulas with G. elata are designed to treat convulsions, such as occurs with tetanus or epilepsy, stroke, and headaches. Based on these formulas, some Chinese Proprietary Medicines (CPMs) were developed in the form of tablets, pills and capsules. In these formulas, G. elata is the principal ingredient, and plays the major role in treating diseases. Gastrodin is the major active constituents of G. elata. Hence, for quality control of these CPMs, gastrodin is selected as a marker. Due to the complexity of the matrix in the CPMs, cleanup of the extracts prior to chromatography is very important for both precise measurement and preservation of chromatographic system. Solid Phase Extraction (SPE) has been widely used to cleanup biological samples (Poole, 2003; Simpson, 2000; Hennion, 1999). However, there are relatively few reports about its use on herbal products. Many methods have been developed for analysis of G. elata (Zhao et al, 1999; Wei et al, 2001; Li et al, 2001 and 2001a). However, little has been reported for its products. The purpose of this study is to develop an SPE HPLC method for determination of gastrodin in the CPMs containing G. elata. 45 2.4.2 Experimental 2.4.2.1 Materials and reagents Gastrodin was provided by Kunming Pharmacy Industry (Kunming, China) and purified (Section 2.2) before use. The acetonitrile and methanol were of HPLC grade. Milli Q water (Millipore, France) was used. Four different CPMs were bought in local medical halls. Their ingredients and the percentage of each herb used as claimed in the label are shown in Table 2.6. 2.4.2.2 HPLC conditions The analysis was performed using an Agilent 1100 HPLC series. The column used was an Inerstil ODS-3 column (4.6 mm × 250 mm, 5 µm), maintained at 35°C. The mobile phase was Milli Q water-acetonitrile-methanol (90:5:5) at a flow rate of 1ml/min. The injection volume for all samples was 5 µl. The detection wavelength was set at 224 nm. 46 Table 2.6 CPM ingredients and their weight percentage CPM Name Ingredients of CPM Percentage Wild G. elata Rhizoma Gastrodiae（Tianma 天麻） 20% Tablet (CPM3) Radix Angelicae Sinensis (Danggui 当归) 20% 野生天麻片 Cortex Eucommiae (Duzhong 杜仲) 14% Radix Rehmanniae (Dihuang 地黄) 32% Radix Scrophulariae (Xuanshen 玄参) 12% Radix Angelicae Pubescentis (Duhuo 独活) 10% Tian-Ma-Wan Pill Rhizoma Gastrodiae（Tianma 天麻） 8.2% (CPM4) Radix Rehmanniae (Dihuang 地黄) 21.9% 天麻丸 Rhizoma et Radix Notopterygii （Qianghuo 羌活） 13.7% Radix Angelicae Sinensis (Danggui 当归) 13.7% Radix Achyranthis Bidentatae （Huainiuxi 怀牛膝） 8.2% Rhizoma Dioscoreae Hypoglaucae （Bixie 荜薤） 8.2% Cortex Eucommiae (Duzhong 杜仲) 9.6% Radix Scrophulariae (Xuanshen 玄参) 8.2% Radix Angelicae Pubescentis (Duhuo 独活) 6.9% Aconiti Tuber Laterale （Fuzi 附子） 1.4% Capsules for the Stubborn Headache Rhizoma Gastrodiae（Tianma 天麻） 30% Radix Polygoni Multiflori （Heshouwu 何首乌） 15% (CPM5) Fructus Ligustri Lucidi（Nüzhenzi 女贞子） 20% Fructus Schisandrae （Wuweizi 五味子） 15% Rhizoma Ligustici Chuanxiong （Chuanxiong 川芎） 10% Radix Angelicae Dahuricae （Baizhi 白芷） 10% Qiangli-Tian-Ma-Du- Rhizoma Gastrodiae（Tianma 天麻） 9.6% Zhong Capsule Cortex Eucommiae (Duzhong 杜仲) 10.2% (CPM6) Radix Aconiti Kusnezoffii Preparata （Zhicaowu 制草乌） 1.2% 强力天麻杜仲胶囊 Radxi Aconiti Lateralis Preparata （Zhifuzi 制附子） 1.2% Radix Angelicae Pubescentis (Duhuo 独活) 6.0% Rhizoma Ligusticum（Haoben 蒿本） 7.1% Radix Scrophulariae (Xuanshen 玄参) 7.1% 顽固头痛胶囊 Radix Angelicae Sinensis (Danggui 当归) 12.8% Radix Rehmanniae (Dihuang 地黄) 19.4% Radix Cyathulae （Chuanniuxi 川牛膝） 7.1% Herba Visci （Hujisheng 槲寄生） 7.1% Rhizoma et Radix Notopterygii（Qianghuo 羌活） 12.0% 47 2.4.2.3 Solid phase extraction method Waters Oasis® HLB SPE cartridges (1cc/30mg) were used. The cartridge was conditioned with 1 ml methanol, and equilibrated with 2 ml Milli Q water. 1 ml sample was loaded, and eluted with 1 ml 10% methanol. The eluate was collected in a vial for HPLC analysis. The SPE scheme is illustrated in Figure 2.13. Condition 1 ml methanol Equilibrate 2 ml Milli Q Water Load 1 ml CPM extract Elute 1 ml 10% MeOH Inject into HPLC system Figure 2.13 Solid phase extraction method for gastrodin in CPM samples. 48 2.4.3 Results and discussion 2.4.3.1 Solid phase extraction The chemical constituents in the CPMs are very complex, and thus it is difficult to analyze them. The products used in this study consist of at least six raw herbs (Table 2.6). All possess very complex chemical compositions and without passing through SPE cartridges, quantitatively analysis of gastrodin in these products can be very difficult. Oasis HLB SPE cartridges were employed for the sample cleanup. The Oasis HLB sorbent is a macroporous copolymer made from a balanced ratio of two monomers, the lipophilic divinylbenzene and the hydrophilic N-vinylpyrrolidone (Waters, 2004). They can be dried during extraction. For normal C18 cartridges, the cartridges must always be kept wet during the cleanup except the finial elution step. Unlike traditional C18 bonded silica sorbent, the Oasis HLB copolymer reversed-phase adsorption mechanism is uncomplicated by the often irreproducible population of surface silanols or metal impurities. This kind of sorbent is usually suitable for extraction of a wide range of compounds, especially polar compounds (Waters, 2004). Under the developed SPE protocol, the recoveries of GA standard solution applied onto the SPE cartridges ranged from 92.0 to 100.4% (Table 2.7). 49 Table2.7 Recoveries of gastrodin applied onto SPE cartridges Concentration added Concentration measured Recovery Mean±SD RSD (µg/ml) (µg/ml) (%) (%) (n=3) (%) 18.66 18.73 100.4 97.2±3.8 3.91 18.32 98.2 17.36 93.0 33.99 93.6 93.2±1.2 1.24 33.48 92.0 34.30 94.2 51.90 93.3 93.3±0.4 0.39 51.78 93.0 52.17 93.7 36.38 55.68 Table2.9 Contents of gastrodin in four Chinese Proprietary Medicines Sample gastrodin content (mg/g) (Mean±SD) (n=3) CPM3 0.62±0.04 CPM4 0.40±0.02 CPM5 1.32±0.14 CPM6 1.27±0.15 50 2.4.3.2 Choice of separation conditions When the separation was carried out with mobile phase flow rat at 1.0 ml/min and 35 °C column temperature, using a mobile phase of Milli Q water-methanol-acetonitrile (90:5:5, v/v/v), good resolution between gastrodin and its adjacent peaks was achieved (Figure 2.14). 2.4.3.3 Linearity and sensitivity Good linearity was obtained in the concentration range of 10-100 µg/ml with the following regression equation. A=9.6041C+0.2625 (r2 = 0.99999; n=5, SD for the slope = 0.0861; SD for the intercept = 0.8951) where A is the peak area (mAU*S), and C is the concentration of gastrodin solution (µg/ml ). The limit of detection (signal-to-noise ratio of 3) was 0.8µg/ml. And the limit of quantification (signal-to-noise ratio of 10) was 3.0µg/ml. 2.4.3.4 Recovery The recovery of the method was determined with the standard addition method for gastrodin in the four CPMs with results ranging from 82.8 to 91.2% (Table 2.8). Compared with the recovery of GA standard solution applied onto SPE cartridges (Table 51 2.7), the recovery of gastrodin from the CPM was about 7% lower. This could be due to the complex matrix in the CPMs. 2.4.3.5 Application The contents of GA in the four CPM samples were determined. Figure 2.14 shows the typical chromatograms of those samples. Table 4 shows the measurement results. The contents of GA in the four CPM samples ranged from 0.4-1.32 mg/g. Table2.8 Recoveries of gastrodin added into CPM samples prior to extraction Samples CPM3 Recovery Mean±SD RSD (%) (%) (n=3) (%) 95.9 89.4±5.6 6.30 91.2±6.5 7.12 87.8±5.2 5.97 82.8±2.8 3.39 86.4 85.9 CPM4 96.8 92.8 84.1 CPM5 91.3 90.4 81.8 CPM6 85.5 82.9 79.9 52 (a) mAU GA 15 10 5 0 0 2 4 6 8 10 12 14 min 8 10 12 14 min 8 10 12 14 min 8 10 12 14 min (b) mAU 20 GA 15 10 5 0 0 2 4 6 (c) mAU GA 20 15 10 5 0 0 2 4 6 (d) mAU GA 20 15 10 5 0 0 2 4 6 Figure 2.14 Typical chromatograms of extracts of four Chinese herbal products: (a) Wild Gastrodia elata Tablet (CPM3), (b) Tian-Ma-Wan Pill (CPM4), (c) Capsules for the Stubborn Headache (CPM5) and (d) QiangLi-Tian-Ma-Du-Zhong Capsule (CPM6) 53 2.4.4 Conclusion A simple and rapid SPE HPLC method for determination of gastrodin in four CPMs was successfully developed. This method has been validated for linearity, precision, LOD and LOQ. The developed method could be useful and suitable for quality control of the four CPMs containing G. elata. 54 Chapter 3 HPLC Profiling of Herbal Medicines 3.1 Introduction Due to their complex constituents, quality control of herbal medicines has been a challenge. Bioactive constituents are often used as markers for quality assurance of herbal medicines (Thompson and Morris, 2001). However, in a large array of herbal medicines, the active substances are not known specifically. Chromatographic fingerprint analysis has been introduced and accepted by World Health Organization (1991) as a strategy for the assessment of herbal medicines. Chromatographic fingerprint is also required by the Drug Administration Bureau of China (2000) to standardize injections made from traditional Chinese medicines and their raw materials. Chromatographic fingerprinting is recommended by U.S. Food and Drug Administration (2002) be used to standardize the extracts or preparations of herbs. 3.1.1 Chromatographic fingerprint analysis of herbal medicines A HPLC chromatogram can be thought of as a chemical fingerprint where the pattern emerges from the relative intensities of the sequence of peaks passing by the detector. In many cases, the interpretation of chromatographic output is to hold an overlay of two chromatograms up to a window to see if the two chromatograms have “significant” difference. Chemometrics can afford us the ability to substitute an objective mechanism for the human pattern recognition step (Brereton, 1992). A peak table, retention time as 55 index and area as measurement variable, can be generated from the output of a chromatogram. Then, multivariate data analysis can be carried out. The use of fingerprinting in herbs tends to focus on identifying and assessing the source and quality of the medicinal plants. HPLC fingerprint analysis of some Chinese herbs (Zhao et al, 2004; Zhang et al, 2003; Chen et al, 2003; Zeng et al, 2002 and 2003; Cheng et al, 2003 and 2003a; Li et al, 2004 and 2004a; Hao et al, 2002;) and their products (Cao et al, 2004; Wang et al, 2002 and 2002a; Gong et al, 2003 and 2004;) have been reported. For example, Chen et al (2003) used HPLC fingerprinting for identifying the habitat of Ligusticum chuanxiong. Cao et al (2004) developed a HPLC fingerprint method for quality control of QingKaiLing injection. 3.1.2 Common problem associated with naming of Chinese herbs Common names for some Chinese herbs often overlap, causing confusion as to the proper identity. For example, Fangji and Guangfangji are traditionally called “Fangji” in China. However, they are from two different medicinal plants: Stephania tetrandra and Aristolochia fangchi (Tang et al, 1992). Their chemical components are also different (Tang et al, 1992). A more extensive discussion of the two herbs can be found in Section 4.1. Due to their different pharmacological actions, using the wrong herb can lead to undesirable consequences. In Belgium, at least 70 people required renal transplant or dialysis due to interstitial fibrosis of the kidney after taking the wrong herb from the Aristolochiaceae family (Vanherwegnem et al, 1993). Aristolochia fangchi contains the nephrotoxic aristolochic acids (Tang et al, 1992). Instead of Stephania tetrandra, the toxic A. fangchi was used. Hence, WHO recommends the use of Latin binomial names, and not 56 the common names. Chromatographic fingerprinting is a useful technique for distinguishing between herbs. 3.1.3 Cluster Analysis Cluster analysis is a popular technique whose basic objective is to discover sample groupings within data (Beebe, 1998). Clustering methods are divided into three categories, hierarchical, object-functional, and graph theoretical. Hierarchical methods are the most popular. For cluster analysis, each sample is treated as a point in an n-dimensional measurement space. The coordinate axes of this space are defined by the measurements used to characterize the samples. Cluster analysis assesses the similarity between samples by measuring the distances between the points in the measurement space. Samples that are similar will lie close to one another, whereas dissimilar samples are distant from each other. The choice of the distance metric to express similarity between samples in a data set depends on the type of measurement variables used. Three types of variables – categorical, ordinal, and continuous – are typically used to characterize chemical samples. Measurement variables are usually continuous. For continuous variables, the Euclidean distance is the best choice for the distance metric, because interpoint distances between the samples can be computed directly (Figure 3.1) (Massart and Kaufman, 1983). However, there is a problem with using the Euclidean distance, which is the so-called scaling effect. It arises from inadvertent weighting of the variables in the analysis that can 57 occur due to differences in magnitude among the measurement variables. The influence of variable scaling on the Euclidean distance can be mitigated by autoscaling the data, which involves standardizing the measurement variables, so that each variable has a mean of zero and a standard deviation of 1 (Equation 3.1). xi , tan dardized = xi ,orig − mi ,orig (3.1) si ,orig where xi,orig is the original measurement variable i, mi,orig is the mean of the original measurement variable I, and si,orig is the standard deviation of the original measurement variable i. Clearly, autoscaling ensures that each measurement variable has an equal weight in the analysis. For cluster analysis, it is best to autoscale the data, because similarity is directly determined by a majority vote of the measurement variables. x1 0 k Dkl l x2 Dkl = n ∑ ( X kj − X lj) 2 j =1 Figure 3.1 Euclidean distances between two data points in a two-dimensional measurement space defined by the measurement variables x1 and x2. 58 Clustering methods attempt to find clusters of patterns (i.e. data points) in the measurement space, hence the term cluster analysis. Although several clustering algorithms exist, e.g. K-means, K-median, Patric-Jarvis, FCV (fuzzy clustering varieties), hierarchical clustering is by far the most widely used clustering method. The starting point for a hierarchical clustering experiment is the similarity matrix which is formed by first computing the distances between all pairs of points in the data set. Each distance is then converted into a similarity value (Equation 3.2): sik = 1 − d ik d max (3.2) where sik (which varies from 0 to 1) is the similarity between samples i and k, dik is the Euclidean distance between samples i and k, and dmax is the distance between the two most dissimilar samples (i.e. the largest distance) in the data set. The similarity values are organized in the form of a table or matrix. The similarity matrix is then scanned for the largest value, which corresponds to the most similar point pair. The two samples constituting the point pair are combined to form a new point, which is located midway between the two original points. The rows and columns corresponding to the old data points are then removed from the matrix. The similarity matrix for the data set is then recomputed. In other words, the matrix is updated to include information about the similarity between the new point and every other point in the data set. The new nearest point pair is identified, and combined to form a single point. This process is repeated until all points have been linked (Brereton, 1992). 59 2 3 1 4 5 Single linkage 5 Complete linkage 5 Average linkage Distance = d15 2 3 1 4 Distance = d35 2 3 1 4 Distance = d15 + d 25 + d35 + d 45 4 Figure 3.2 The distance between a data cluster and a point using single linkage, complete linkage and average linkage. There are a variety of ways to compute the distances between data points and clusters in hierarchical clustering (Figure 3.2). The single-linkage method assesses similarity between a point and a cluster of points by measuring the distance to the closest point in the cluster. The complete linkage method assesses similarity by measuring the distance to the farthest point in the cluster. Average linkage assesses the similarity by computing the distances between all point pairs where a member of each pair belongs to the cluster. The average of these distances is used to compute the similarity between the data point and the cluster (Lavine, 1992). Each linkage has its advantages and disadvantages. The average method results in a better measure of the distance between clusters when the clusters are well separated. This is because it measures the distance between the centers of the clusters rather than the distances from the edges as in single link or complete link. On the other 60 hand, the average method is sensitive to outliers because these unusual points adversely influence the calculation of the average of a cluster. With well-separated classes, all methods give similar results (Beebe, 1998). In this study, the average linkage method is employed. To illustrate the method, consider the hypothetical data set and similarity matrix in Table 3.1 and 3.2. Table 3.1 presents the data for five objectives, which have two variables (v1 and v2). Table 3.2 gives their similarity matrix. Table 3.1 Observations of five objects Object Object Object Object Object v1 1 1 5 7 7 1 2 3 4 5 v2 1 2 4 5 7 Table 3.2 Similarity matrix (of observations in Table 3.1) 1 Object Object Object Object Object 1 2 3 4 5 0.88 0.41 0.15 0.00 2 0.88 0.47 0.21 0.08 3 0.41 0.47 0.74 0.58 4 0.15 0.21 0.74 5 0.00 0.08 0.58 0.76 0.76 The similarity matrix is then scanned for the largest value. This is a value of 0.88 between object 1 and 2. Hence, object 1 and 2 should be combined to form a new point (A). Table 3.3 shows the updated similarity matrix, implying highest similarity between two objects. The average linkage method was chosen for assessing the similarity between a data point and a point cluster. Table 3.3 Updated similarity matrix (from Table 3.2) A Cluster Object Object Object A 3 4 5 0.45 0.19 0.05 3 0.45 0.74 0.58 4 0.19 0.74 5 0.05 0.58 0.76 0.76 61 The updated similarity matrix is then scanned for the largest value (0.76); the nearest pair (object 4 and 5) is combined to form a single point (B). The similarity matrix is then updated (Table 3.4) to include information about the similarity between the new point (B) and every other point in the data set. This process is repeated, and the two clusters A and B (Table 3.5) are finally merged together at a similarity of 0.20. Table 3.4 Updated similarity matrix (from Table 3.3) A Cluster A Object 3 Cluster B 3 0.45 0.45 0.12 B 0.12 0.67 0.67 Table 3.5 Updated similarity matrix (from Table 3.4) A Cluster A Cluster B B 0.20 0.20 0.0 0.2 Similarity 0.4 0.6 0.8 1.0 1 2 3 4 5 Observations Figure 3.2 Average-linkage dendrogram of the hypothetical data set (Table 3.1). The results of a hierarchical clustering study are usually displayed as a dendrogram (Figure 3.2), which is a tree-shape map of intersample distances in the data set. The 62 dendrogram shows the merging of samples into clusters at various stages of the analysis and the similarities at which the clusters merge, with the clustering displayed hierarchically. Interpretation of the results is intuitive, which is the major reason for the popularity of these methods. 3.1.4 Objective The objective of this work is to develop a HPLC method for fingerprinting analysis of some commonly used Chinese herbs in Singapore. Some marker compounds are also identified in the herbal samples by HPLC-DAD. 3.2 Experimental 3.2.1 Materials and reagents Thirty-four Chinese herbs (Table 3.6) were purchased. Twenty-seven of them were from three retailers – Eu Yan Sang (EYS), Sinchong Meheco (SINC) and Wong Yiu Nam (WYN) in Chinatown, Singapore; and another 7 were purchased from China, including the reference herb (Guangfangji) purchased from the National Institute for the Control of Pharmaceutical and Biological Products (NICPBP) (Beijing, China). Aristolochic acid I (AAI), fangchinoline, tetrandrine, baicalin, indigotin and catapol were obtained from the National Institute for the Control of Pharmaceutical and Biological Products (NICPBP) (Beijing, China). 4-hydroxybenzyl alcohol, 4-hydroxybenzaldehyde and L-pyroglutamic acid of analytical grade were obtained from Sigma (St. Louis, USA). 63 Table 3.6 List of Traditional Chinese Herbal Medicines and their sources No Chinese Name Responding Latin Binomial Name Source 1 Fangji （防己） Radix Stephaniae Tetrandrae Eu Yan Sang (Singapore) 2 Fangji（防己） Radix Stephaniae Tetrandrae Wong Yiu Nam Medical Hall 3 Fangji（防己） Radix Stephaniae Tetrandrae Sinchong Meheco 4 Guangfangji（广防己） Radix Aristolochiae Fangchi Guangzhou, China 5 Guangfangji（广防己） Radix Aristolochiae Fangchi Beijing, China 6 Hanzhongfangji Radix Aristolochiae Heterophyllae Shanxi, China （汉中防己） 7 Fangji（防己） Radix Stephaniae Tetrandrae Shenyang, China 8 Fangji（防己） Radix Stephaniae Tetrandrae TongRenTang, Beijing, China 9 Fangji（防己） Radix Stephaniae Tetrandrae LeRenTang, Shijiazhuang, China 10 Fangji（防己） Radix Stephaniae Tetrandrae Shibao Co, Shijiazhuang, China 11 Tianma（天麻） Rhizoma Gastrodiae Eu Yan Sang (Singapore) 12 Tianma（天麻） Rhizoma Gastrodiae Wong Yiu Nam Medical 13 Tianma（天麻） Rhizoma Gastrodiae Sinchong Meheco 14 Huangqin（黄芩） Radix Scutellariae Eu Yan Sang (Singapore) 15 Huangqin（黄芩） Radix Scutellariae Wong Yiu Nam Medical Hall 16 Huangqin（黄芩） Radix Scutellariae Sinchong Meheco 17 Madouling（马兜玲） Fructus Aristolochiae Eu Yan Sang (Singapore) 18 Madouling（马兜玲） Fructus Aristolochiae Wong Yiu Nam Medical Hall 19 Madouling（马兜玲） Fructus Aristolochiae Sinchong Meheco 20 Duzhong（杜仲） Cortex Eucommiae Eu Yan Sang (Singapore) 21 Duzhong（杜仲） Cortex Eucommiae Wong Yiu Nam Medical Hall 22 Duzhong（杜仲） Cortex Eucommiae Sinchong Meheco 23 Banlangen（板蓝根） Radix Isatidis Eu Yan Sang (Singapore) 24 Banlangen（板蓝根） Radix Isatidis Wong Yiu Nam Medical Hall 25 Banlangen（板蓝根） Radix Isatidis Sinchong Meheco 26 Banxia（半夏） Rhizoma Pinelliae Eu Yan Sang (Singapore) 27 Banxia（半夏） Rhizoma Pinelliae Wong Yiu Nam Medical Hall 28 Banxia（半夏） Rhizoma Pinelliae Sinchong Meheco 29 Shengdi（生地） Radix Rehmanniae Eu Yan Sang (Singapore) 30 Shengdi（生地） Radix Rehmanniae Wong Yiu Nam Medical Hall 31 Shengdi（生地） Radix Rehmanniae Sinchong Meheco 32 Shudi（熟地） Radix Rehmanniae Preparate Eu Yan Sang (Singapore) 33 Shudi（熟地） Radix Rehmanniae Preparate Wong Yiu Nam Medical Hall 34 Shudi（熟地） Radix Rehmanniae Preparate Sinchong Meheco 64 Gastrodin was provided by Kunming Pharmacy Industry (Kunming, China) and purified (Section 2.2) before use. The solvents, acetonitrile and methanol, were of HPLC grade. All the chemical standards were from National Institute for Control of Pharmaceutical and Biological Products (Beijing, China). Milli Q water (>18 mΩ) (Milli Pore, France) were used. Other reagents were of analytical grade. 3.2.2 HPLC conditions The analysis was performed using an Agilent 1100 HPLC series chromatography. The column used was an Inertsil ODS-3 column (4.6 mm × 250mm, 5 µm), maintained at 35°C. The mobile phase was 0.01% H3PO4 (v/v) in Milli Q water (pH 3.0) (A) and acetonitrile (B) with a gradient program as follows: 5%B to 100% B (55 min), 100% B to 100% B (5 min), post-run (10 min) at a flow rate of 1ml/min. The injection volume for all samples was 5 µl. Detection wavelength was 210 nm. The UV spectra from 190 to 400 nm were recorded during the chromatographic run. 3.2.3 Sample preparation 2 g of powdered herbs were accurately weighed. 10 ml of 70% (v/v) of methanol were added, and then extracted under ultrasonic water bath for 20 min. The process was repeated 3 times for each herb. After filtration, the combined methanol extracts were evaporated to dryness by a rotary evaporator. The residue was dissolved in 10 ml of 70% methanol and filtered by 0.45 µm membrane before analysis. 65 For detection and identification of marker compounds, 4-hydroxybenzyl alcohol (HA) was spiked in as an internal standard in the reference standards and samples. All the chemical standards were also dissolved in 70% methanol and filtered by 0.45 µm membrane before inject into HPLC system. 3.2.4 HPLC fingerprint analysis 3.2.4.1 Chromatographic fingerprinting A chromatographic fingerprint was mathematically represented by a vector of peak areas in this study. Chromatographic peak areas for each chromatographic were obtained by computer integration and entered into a spreadsheet. The peaks were assigned their retention times manually. A peak table was developed by examining integration reports and adding features to the table which did not match the retention times of previous observed features. A preliminary data vector was produced for each chromatogram by matching the retention times of HPLC peaks with the retention times of the features in the table. A feature would be assigned a value corresponding to the area of the HPLC peak in the chromatogram. Unmatched peaks were zeroed. Finally, 81 peaks were selected in the table, although not all peaks were present in all chromatograms. Hence, for cluster analysis, each liquid chromatogram was initially represented as a 81-dimensional data vector, x = (x1, x2, x3, …, xj, …, x81), where xj is the jth peak. The data vectors were normalized to constant sum, i.e., each xj was divided by the total integrated peak area. 66 3.2.4.2 Data Analysis All data analyses were performed on a Pentium-based IBM-compatible personal computer. The MINITAB Release 13.20 for windows software (Minitab Inc., USA) was used to perform the hierarchical cluster analysis. 3.2.5 Detection and identification of marker compounds 3.2.5.1 UV library of chemical standards LC-UV Library G218x (Rev.A.01.00) was used for developing the UV spectra database of chemical reference standards. 3.2.5.2 RT and RRT of chemical standards Relative retention time (RRT) is determined by the ratio of retention time of chemical standard (RT) to the retention time of internal standard (RTIS): RRT = RT RTIS (3.3) All chemical reference standards were analyzed on three different days. The data of RT and RTIS were recorded and RRT were calculated based on the above formula. 3.2.5.3 Peak identification Of these 34 TCMs, 12 TCMs were selected based on the availability of reference standards and screened for their marker compounds by HPLC-DAD. With a DAD, the identity of the compounds in the samples was carried out by comparing the RT, RRT and 67 UV spectra with those of the reference standard. Before a library search, a peak would be checked for purity. If not, spectral contribution arising from the impurity might overlay with the main compound and give an incorrect library search result. Figure 3.4 c shows a typical chromatogram of Madouling. The peak eluting at 35.6 min was first checked by comparing the normalized spectra at the peak upslope, apex and downslope. Next, the peak upslope spectrum was subtracted from the apex spectrum to eliminate small spectral irregularities arising from the sample matrix, the so-called chemical noise. Then the corrected spectrum was compared with those in a library. When comparing standard library with unknown spectra, the computer calculated a match factor. This factor indicates how closely the unknown spectrum matches the library spectrum. At the extremes, a match factor 0 indicates no match and 1000 indicates identical spectra. Generally, values above 990 indicate that the spectra are similar. Values between 900 and 990 indicate there is some similarity but the result should be interpreted with caution. All values below 900 indicate, in effect, that the spectra are different (Huber and George, 1993). 68 (a) (b) Norm *DAD1, 35.573 (Apex) of MaDouLing Purity factor: 997.374 14 *AAI standard 12 Match factor: 983.527 10 8 6 4 2 0 200 225 250 275 300 325 350 375 nm 200 225 250 275 300 325 350 375 nm (c) mAU IS 25 20 35.6 15 10 5 0 10 20 30 40 50 min Figure 3.4 (a) Peak purity check by overlaying normalized spectra acquired in the upslope, apex, and downslope of the peak; (b) Overlaid UV spectra of peak at 35.6 min in Madouling (continuous line) and that of AAI standard (broken line); and (c) typical chromatogram of an extract of Madouling 69 3.3 Results and discussion 3.3.1 Choice of HPLC conditions A gradient HPLC method was developed to elute the complex constituents of the herbs. 210 nm was selected as detection wavelength, because most herbal constituents have good absorption at low detection wavelength. However, if the detection wavelength is lower than 210 nm, it may cause a baseline shift when gradient program is performed. To detect acidic and basic ingredients, 0.01% H3PO4 (v/v) was used. At 210 nm, the UV absorbance of 0.01% H3PO4 solution is very small, which does not lead to baseline shift (Snyder, 1997). 3.3.2 Chromatographic fingerprint analysis 3.3.2.1 HPLC profiling Thirty-four herbs (Table 3.6) purchased from China and three local herbal halls, EYS, WYN and SINC, were profiled by the HPLC method. A chromatographic fingerprint is a set of peaks of a chromatogram that represents the characteristics of a sample. Figure 3.5 shows a typical chromatogram of an extract of Huangqin. In this study, the reverse phase column was used. The polar components elute more quickly than the less polar ones. The similarity between HPLC chromatographic patterns reveals the similarity between the samples. 70 mAU 120 100 80 60 40 20 0 0 10 20 30 40 50 min Figure 3.5 Typical HPLC chromatogram of an extract of Huangqin 71 3.3.2.2 Comparison of the similarity of chromatographic fingerprints Euclidean distance was used as the similarity measure in this work. As described in Section 3.2.4.1, each chromatographic fingerprint was regarded as an 81-dimentional vector. To compare the similarity of two chromatographic fingerprints, the Euclidean distance of two vectors were calculated. The smaller the Euclidean distance is, the larger the similarity. The value of Euclidean distance is a relative indicator. It should be compared under the same system. Table 3.7 lists the Euclidean distance matrix of the thirty four Chinese herbs studies. 3.3.2.3 Hierarchical Cluster Analysis Hierarchical cluster analysis (HCA) is the process of subdivision of a group of samples into clusters that exhibit a high degree of both intra-cluster similarity and inter-cluster dissimilarity. HCA was applied to the entire normalized area matrix to explore the structure of the area data. The Euclidean distance was used to measure the similarities between clusters. The average method was applied to link the clusters. Figure 3.6 depicts the HCA results presented in the form of a dendrogram. Ten distinguished clusters were classified with similarity higher than 70%. The Madouling samples (17, 18 and 19), Huangqin samples (14, 15 and 16), Tianma samples (11, 12 and 13), and Banxia samples (26, 27 and 28) were clustered respectively as group B, C, D and I. 72 Table 3.7 Euclidean distance matrix of thirty-four Chinese herbs Case Euclidean Distance 1 2 3 4 5 6 7 8 9 10 11 12 0.056 0.063 0.053 0.028 0.043 0.043 0.102 0.098 0.078 0.091 0.125 0.149 0.123 0.123 0.121 0.136 0.161 0.138 0.133 0.138 0.057 0.330 0.342 0.330 0.330 0.327 0.294 0.301 0.330 0.341 0.330 0.330 0.327 0.295 0.303 0.013 0.128 0.147 0.127 0.127 0.116 0.092 0.121 0.304 0.304 0.227 0.248 0.230 0.229 0.224 0.178 0.191 0.350 0.349 0.206 0.216 0.238 0.219 0.218 0.213 0.164 0.179 0.344 0.343 0.195 0.037 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 0.056 0.063 0.028 0.102 0.125 0.136 0.330 0.330 0.128 0.227 0.216 0.227 0.221 0.229 0.246 0.300 0.225 0.229 0.196 0.190 0.211 0.196 0.351 0.218 0.053 0.043 0.098 0.149 0.161 0.342 0.341 0.147 0.248 0.238 0.247 0.237 0.245 0.261 0.311 0.237 0.244 0.221 0.212 0.230 0.212 0.361 0.236 0.043 0.078 0.123 0.138 0.330 0.330 0.127 0.230 0.219 0.230 0.215 0.221 0.241 0.300 0.225 0.230 0.197 0.189 0.209 0.195 0.351 0.221 0.091 0.123 0.133 0.330 0.330 0.127 0.229 0.218 0.230 0.221 0.230 0.248 0.301 0.224 0.229 0.201 0.193 0.213 0.196 0.351 0.220 0.121 0.138 0.327 0.327 0.116 0.224 0.213 0.225 0.214 0.222 0.242 0.295 0.214 0.222 0.199 0.186 0.207 0.189 0.348 0.214 0.057 0.294 0.295 0.092 0.178 0.164 0.179 0.188 0.200 0.221 0.272 0.197 0.196 0.147 0.141 0.170 0.154 0.331 0.183 0.301 0.303 0.121 0.191 0.179 0.192 0.209 0.219 0.238 0.284 0.211 0.211 0.161 0.165 0.193 0.176 0.341 0.202 0.013 0.304 0.350 0.344 0.350 0.343 0.349 0.361 0.403 0.357 0.358 0.333 0.315 0.326 0.314 0.435 0.288 0.304 0.349 0.343 0.349 0.342 0.349 0.361 0.402 0.356 0.357 0.334 0.314 0.326 0.313 0.435 0.287 0.206 0.195 0.207 0.190 0.200 0.221 0.282 0.205 0.207 0.175 0.161 0.184 0.163 0.331 0.191 0.037 0.031 0.248 0.257 0.273 0.324 0.262 0.257 0.234 0.213 0.234 0.208 0.369 0.251 0.055 0.239 0.249 0.266 0.316 0.253 0.247 0.217 0.204 0.227 0.205 0.364 0.242 26 27 28 29 30 31 32 0.183 0.162 0.172 0.187 0.184 0.197 0.183 0.203 0.186 0.195 0.219 0.216 0.227 0.211 0.178 0.165 0.176 0.201 0.199 0.210 0.191 0.180 0.164 0.175 0.194 0.192 0.202 0.186 0.180 0.165 0.173 0.207 0.203 0.218 0.197 0.117 0.110 0.132 0.140 0.141 0.153 0.127 0.119 0.133 0.155 0.127 0.129 0.127 0.104 0.321 0.317 0.320 0.335 0.337 0.345 0.331 0.323 0.318 0.321 0.338 0.340 0.349 0.334 0.152 0.133 0.137 0.184 0.181 0.198 0.176 0.219 0.188 0.208 0.242 0.239 0.250 0.235 0.208 0.174 0.197 0.230 0.224 0.237 0.222 33 34 0.141 0.173 0.178 0.202 0.161 0.181 0.151 0.178 0.163 0.182 0.109 0.117 0.125 0.116 0.316 0.326 0.316 0.328 0.133 0.159 0.209 0.224 0.195 0.208 Continued 73 Table 3.7 (Continued) Case 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Euclidean Distance 13 14 15 16 17 18 19 20 21 22 23 0.227 0.247 0.230 0.230 0.225 0.179 0.192 0.350 0.349 0.207 0.031 0.055 0.221 0.237 0.215 0.221 0.214 0.188 0.209 0.343 0.342 0.190 0.248 0.239 0.248 0.229 0.245 0.221 0.230 0.222 0.200 0.219 0.349 0.349 0.200 0.257 0.249 0.257 0.049 0.246 0.261 0.241 0.248 0.242 0.221 0.238 0.361 0.361 0.221 0.273 0.266 0.274 0.056 0.052 0.300 0.311 0.300 0.301 0.295 0.272 0.284 0.403 0.402 0.282 0.324 0.316 0.324 0.318 0.326 0.338 0.225 0.237 0.225 0.224 0.214 0.197 0.211 0.357 0.356 0.205 0.262 0.253 0.265 0.258 0.267 0.283 0.095 0.229 0.244 0.230 0.229 0.222 0.196 0.211 0.358 0.357 0.207 0.257 0.247 0.259 0.255 0.265 0.281 0.099 0.052 0.196 0.221 0.197 0.201 0.199 0.147 0.161 0.333 0.334 0.175 0.234 0.217 0.226 0.230 0.238 0.257 0.309 0.257 0.253 0.190 0.212 0.189 0.193 0.186 0.141 0.165 0.315 0.314 0.161 0.213 0.204 0.209 0.213 0.222 0.242 0.302 0.242 0.237 0.148 0.211 0.230 0.209 0.213 0.207 0.170 0.193 0.326 0.326 0.184 0.234 0.227 0.232 0.230 0.239 0.257 0.316 0.256 0.252 0.199 0.064 0.196 0.212 0.195 0.196 0.189 0.154 0.176 0.314 0.313 0.163 0.208 0.205 0.204 0.216 0.228 0.246 0.299 0.236 0.234 0.212 0.188 0.207 0.248 0.257 0.274 0.324 0.265 0.259 0.226 0.209 0.232 0.204 0.369 0.251 0.220 0.191 0.208 0.239 0.234 0.247 0.233 0.206 0.219 0.049 0.056 0.318 0.258 0.255 0.230 0.213 0.230 0.216 0.363 0.240 0.205 0.200 0.140 0.241 0.241 0.254 0.238 0.203 0.224 0.052 0.326 0.267 0.265 0.238 0.222 0.239 0.228 0.370 0.249 0.216 0.211 0.147 0.249 0.249 0.262 0.247 0.214 0.233 0.338 0.283 0.281 0.257 0.242 0.257 0.246 0.381 0.266 0.234 0.230 0.162 0.265 0.265 0.277 0.263 0.232 0.251 0.095 0.099 0.309 0.302 0.316 0.299 0.421 0.313 0.303 0.293 0.296 0.254 0.316 0.330 0.276 0.277 0.298 0.052 0.257 0.242 0.256 0.234 0.377 0.255 0.240 0.224 0.229 0.207 0.262 0.275 0.224 0.212 0.240 0.253 0.237 0.252 0.234 0.376 0.250 0.235 0.216 0.225 0.206 0.259 0.274 0.222 0.208 0.239 0.148 0.199 0.212 0.359 0.237 0.180 0.175 0.187 0.173 0.136 0.177 0.166 0.133 0.103 0.064 0.188 0.349 0.216 0.287 0.166 0.173 0.197 0.190 0.211 0.195 0.159 0.167 0.207 0.359 0.231 0.204 0.187 0.194 0.228 0.228 0.241 0.225 0.193 0.209 0.212 0.110 0.175 0.164 0.171 0.218 0.217 0.229 0.213 0.180 0.195 Continued 74 Table 3.7 (Continued) Case 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Euclidean Distance 24 25 26 27 28 29 30 31 32 33 34 0.351 0.361 0.351 0.351 0.348 0.331 0.341 0.435 0.435 0.331 0.369 0.364 0.369 0.363 0.370 0.381 0.421 0.377 0.376 0.359 0.349 0.359 0.212 0.218 0.236 0.221 0.220 0.214 0.183 0.202 0.288 0.287 0.191 0.251 0.242 0.251 0.240 0.249 0.266 0.313 0.255 0.250 0.237 0.216 0.231 0.110 0.203 0.183 0.203 0.178 0.180 0.180 0.117 0.119 0.321 0.323 0.152 0.219 0.208 0.220 0.205 0.216 0.234 0.303 0.240 0.235 0.180 0.287 0.204 0.175 0.344 0.215 0.162 0.186 0.165 0.164 0.165 0.110 0.133 0.317 0.318 0.133 0.188 0.174 0.191 0.200 0.211 0.230 0.293 0.224 0.216 0.175 0.166 0.187 0.164 0.339 0.198 0.102 0.172 0.195 0.176 0.175 0.173 0.132 0.155 0.320 0.321 0.137 0.208 0.197 0.208 0.140 0.147 0.162 0.296 0.229 0.225 0.187 0.173 0.194 0.171 0.342 0.199 0.137 0.105 0.187 0.219 0.201 0.194 0.207 0.140 0.127 0.335 0.338 0.184 0.242 0.230 0.239 0.241 0.249 0.265 0.254 0.207 0.206 0.173 0.197 0.228 0.218 0.365 0.230 0.150 0.169 0.185 0.184 0.216 0.199 0.192 0.203 0.141 0.129 0.337 0.340 0.181 0.239 0.224 0.234 0.241 0.249 0.265 0.316 0.262 0.259 0.136 0.190 0.228 0.217 0.362 0.230 0.147 0.162 0.179 0.087 0.197 0.227 0.210 0.202 0.218 0.153 0.127 0.345 0.349 0.198 0.250 0.237 0.247 0.254 0.262 0.277 0.330 0.275 0.274 0.177 0.211 0.241 0.229 0.367 0.245 0.146 0.180 0.198 0.099 0.065 0.183 0.211 0.191 0.186 0.197 0.127 0.104 0.331 0.334 0.176 0.235 0.222 0.233 0.238 0.247 0.263 0.276 0.224 0.222 0.166 0.195 0.225 0.213 0.361 0.228 0.129 0.160 0.181 0.048 0.075 0.069 0.141 0.178 0.161 0.151 0.163 0.109 0.125 0.316 0.316 0.133 0.209 0.195 0.206 0.203 0.214 0.232 0.277 0.212 0.208 0.133 0.159 0.193 0.180 0.343 0.194 0.147 0.125 0.138 0.119 0.113 0.148 0.127 0.173 0.202 0.181 0.178 0.182 0.117 0.116 0.326 0.328 0.159 0.224 0.208 0.219 0.224 0.233 0.251 0.298 0.240 0.239 0.103 0.167 0.209 0.195 0.350 0.218 0.132 0.144 0.163 0.100 0.064 0.102 0.093 0.101 0.203 0.344 0.339 0.342 0.365 0.362 0.367 0.361 0.343 0.350 0.215 0.198 0.199 0.230 0.230 0.245 0.228 0.194 0.218 0.102 0.137 0.150 0.147 0.146 0.129 0.147 0.132 0.105 0.169 0.162 0.180 0.160 0.125 0.144 0.185 0.179 0.198 0.181 0.138 0.163 0.087 0.099 0.048 0.119 0.100 0.065 0.075 0.113 0.064 0.069 0.148 0.102 0.127 0.093 0.101 75 Similarity 18.63 45.76 72.88 A B C D E F G H I J 100.00 24 8 9 17 18 19 14 15 16 11 13 12 23 25 21 22 1 4 2 3 5 6 7 10 26 27 28 20 29 32 30 34 31 33 Figure 3.6 Dendrogram of thirty-four Chinese herbs using average linkage method 76 All three Fangji samples bought locally (1, 2 and 3) were clustered with Guangfangji (Guangzhou) (4) and reference Guangfangji (Beijing) (5) as a group (G); while two Fangji samples from China (8 and 9) were clustered as group A. Figure 3.7 shows the overlaid chromatograms of these five samples. Their chromatographic fingerprints have good similarity. Thus three Fangji samples from local herbal halls could be Guangfangji. The Fangji sample from Shenyang (7) was clustered with Hanzhongfangji (6) and the Fangji sample from Shijiazhuang (10) as group H. This indicates that the Fangji samples bought in Shenyang and Shijiazhuang (Sample 7 and 10) are not Stephania tetrandra. As mentioned in Section 3.1.2 and more extensively in section 4.1, many reported cases of renal failure were due to the wrong use of Guangfangji (Aristolochia fangchi) instead of the intended Fangji (Stephania tetrandrae) (Vanherwegnem et al., 1993; Cosyns et al., 2003; Zhou et al., 2004). Further studies on the local Fangji samples are needed to establish their identities. The Banlangen sample from WYN (24) was not clustered with another two Banlangen samples (23 and 25). Figure 3.8 shows the overlaid chromatograms of these three samples. Further work on Banlangen is currently in progress. The Duzhong sample from EYS (20) was also not clustered with the other two Duzhong samples (21 and 22). Figure 3.9 shows the overlaid chromatograms of these three Duzhong samples. 77 (a) (b) (c) (d) (e) 0 10 20 30 40 50 min Figure 3.7 Overlaid chromatograms of Fangji from (a) EYS, (b) WYN, (c) SINC and Guangfangji from (d) Guangzhou and (e) Beijing 78 (a) (b) (c) 0 10 20 30 40 50 min Figure 3.8 Overlaid chromatograms of extracts of Banlangen from (a) EYS, (b) WYN and (c) SINC 79 (a) (b) (c) 0 10 20 30 40 50 min Figure 3.9 Chromatograms of extracts of Duzhong from (a) EYS, (b) WYN and (c) SINC 80 The three Shengdi samples (sample 29-31) and three Shudi samples (sample 32-34) were clustered into group J and were not differentiated under this system (Figure 3.10). Shudi is the steam-processed Shengdi (Pharmacopoeia of the People’s Republic of China, 2000). Processing of Chinese herbs is very important for their effect and quality. The processing of herb can reduce toxicity or side effect or/and potentiate effects (Zhu, 1998). The chromatograms of these extracts showed largely unresolved peaks in the retention time between 2 to 10 minutes under the current chromatographic conditions. This may have contributed to the similarity of the chromatograms as determined by the hierarchical cluster analysis. Optimised separation of the components may allow greater differentiation between Shengdihuang and Shudihuang samples. 81 (a) (b) (c) (d) (e) (f) 0 10 20 30 40 50 min Figure 3.10 Overlaid chromatograms of extracts of Shudihuang from (a) EYS, (b) WYN, and (c) SINC, and extracts of Shengdihuang from (d) EYS, (e) WYN and (f) SINC 82 3.3.3 Detection of marker compounds by HPLC-DAD 3.3.3.1 RT and RRT The RT and RRT of 8 reference standards have been recorded and calculated (Table 3.8). 4-hydroxybenzyl alcohol (HA) was spiked in as an internal standard. The variation of the RT and RRT were also investigated. RSD of RT ranged between 0.347% and 1.546%, while RSD of RRT was from 0.031% to 0.971%. RSDs of RTs of the reference standards were generally smaller than that of RRT. This indicates that the variation of relative retention time is smaller than that of absolute retention time. Table 3.8 Retention time (RT) and relative retention time (RRT) of standards Chemical standards RT(min) (n=3) RRT (n=3) mean SD RSD mean SD RSD Aristolochic acid I 35.653 0.135 0.377 3.521 0.032 0.908 Fangchinoline 14.255 0.223 1.566 1.397 0.014 0.971 Tetrandrine 15.326 0.203 1.325 1.502 0.013 0.861 Gastrodin 7.033 0.026 0.373 0.698 0.001 0.031 4-Hydroxybenzaldehyde 16.147 0.056 0.347 1.601 0.001 0.067 Baicalin 21.229 0.204 0.961 2.088 0.018 0.870 Indigotin 18.521 0.072 0.389 1.838 0.008 0.464 Catapol 5.260 0.027 0.509 0.524 0.002 0.315 83 3.3.3.2 Peak identification A combination of retention time and spectral match was used for the identification of the peaks of interest. Table 3.9 shows the RTs and RRTs and the UV match factors of the peaks of interest. The percentage difference of the RTs of the suspected marker compound peaks compared with those of the reference standards were less than 6%. The percentage difference of the RRT of the peaks of interest and those of the reference standards were less than 3%. Most of the marker compounds were found to be present in the samples (Table 3.9). In the samples of Fangji from local medical halls, fangchinoline and tetrandrine were not found, but AA I was instead found present. However the content of AA I in these three samples was much lower than that in the reference Guangfangji. The purity factors of the peaks were generally above 950, except for the AAI peak in Fangji samples from EYS and WYN. The UV match factors were generally high (≥900) except for the AAI peaks in the local Fangji samples. From the results, it could be concluded that fangchinoline and tetrandrine were detected in the Fangji samples from Tongrentang and Lerentang, while gastrodin was detected in all three local samples of Tianma and indigotin was detected in all three local samples of Banlangen. Other results have to be treated with caution, in particular, the presence of AAI in local Fangji samples. Further work to confirm the presence of AAI in the local Fangji samples will be presented in Chapter 4. 84 Table 3.9 RT and RRT of target compound peak in the samples and their differences compared with the standard values Sample ID Marker Compound for detection RT (min) RRT Madouling_EYS Madouling_SINC Madolling_WYN Fangji_EYS Fangji_SINC Fangji_WYN Guangfangji_Ref Fangji_TRT Fangji_LRT Fangji_TRT Fangji_LRT Tianma_EYS Tianma_SINC Tianma_WYN Tianma_EYS Tianma_SINC Tianma_WYN Huangqin_EYS Huangqin_SINC Huangqin_WYN Banlangen_EYS Banlangen_SINC Banlangen_WYN Shengdi_EYS Shengdi_SINC Shengdi_WYN Shudi_EYS Shudi_SINC Shudi_WYN AAI AAI AAI AAI AAI AAI AAI fangchinoline fangchinoline tetrandrine tetrandrine GA GA GA HD HD HD baicalin baicalin baicalin indigotin indigotin indigotin catapol catapol catapol catapol catapol catapol 35.571 35.552 35.549 35.486 35.561 35.585 35.708 14.325 14.375 15.365 15.406 7.001 7.004 7.003 16.189 16.188 16.182 21.303 21.369 21.409 18.472 18.466 18.465 5.114 5.218 5.267 5.431 5.523 5.564 3.535 3.534 3.537 3.530 3.538 3.539 3.496 1.385 1.383 1.485 1.482 0.698 0.697 0.698 1.613 1.612 1.612 2.084 2.078 2.071 1.843 1.841 1.840 0.511 0.514 0.514 0.519 0.525 0.530 Difference of RT (%) with standard value Difference of RRT (%) with standard value -0.230 -0.283 -0.292 -0.468 -0.258 -0.191 0.154 0.491 0.842 0.254 0.522 -0.005 -0.004 -0.004 0.260 0.254 0.217 0.349 0.659 0.848 -0.265 -0.297 -0.302 -2.776 -0.798 0.133 3.251 5.000 5.779 0.403 0.379 0.450 0.262 0.484 0.522 -0.710 -0.869 -1.011 -1.105 -1.328 -0.069 -0.076 -0.040 0.745 0.689 0.702 -0.170 -0.465 -0.809 0.260 0.157 0.132 -2.453 -1.843 -1.879 -1.036 0.172 1.146 Purity factor 997.374 979.349 994.462 821.232 964.640 825.039 999.629 999.822 999.918 930.491 992.955 999.969 999.953 999.954 989.868 997.286 999.140 998.602 997.870 998.966 999.809 999.798 999.799 972.379 969.964 952.968 960.460 985.526 983.352 UV match factor 983.527 899.119 960.153 695.430 743.039 773.034 999.849 999.709 999.898 999.737 999.866 999.904 999.979 999.917 896.541 881.236 899.594 983.306 974.514 993.455 999.978 999.831 999.880 901.756 891.891 938.491 842.947 957.366 881.254 85 3.4 Conclusion The present study demonstrated a simple HPLC method for fingerprinting analysis of some Chinese Herbal Medicines. Extracts of thirty-four herbal samples, constituting 11 different types of herbs were analyzed by HPLC. The chromatograms were further subject to hierarchical cluster analysis. Results showed that three samples bought as “Fangji” (Stephania tetrandra) in Singapore may be actually Guangfangji (Aristolochia fangji) instead. Further studies (Chapter 4) need to be carried out to confirm this finding. The HPLC fingerprint method developed can be potentially used for the differentiation of some Chinese Herbal Medicines. 86 Chapter 4. Differentiation and Authentication of Stephania tetrandra and Aristolochia fangchi by HPLC-DAD and LC-MS/MS 4.1 Introduction Radix Stephaniae Tetrandrae, Fangji 防己（also called Fenfangji 粉防己, or Hanfangji 汉防己）, is the dry roots of Stephania tetrandra S. Moore (Menispermaceae). Radix Aristolochiae Fangchi, or also known as Guangfangji 广防己 , is the dry root of Aristolochia fangchi Y.C. Wu ex L. D. Chou et S. M. Hwang. Both herbs have been traditionally used as diuretic in China (Tang et al, 1992). However, their chemical constituents are quite different. The main chemical constituents in the roots of S. tetrandra are the alkaloids, tetrandrine and fangchinoline; while aristolochic acid I (AAI) is the major bioactive constituent of the root of A. fangchi (Figure 4.1) (Tang et al, 1992). Tetrandrine and fangchinoline exhibit the characteristics of Ca2+ channel blocker (King, et al, 1988, Liu et al, 1991, 1992; Weinsberg et al, 1994; Kim et al, 1997). They have inhibitory activities on human platelet aggregation (Kim et al, 1999, 1999a), anti-inflammatory effects (Shen et al, 2001), hypotensive effects (Kim et al, 1997) and cardiopretective effects (Yu et al, 2001), neuroprotective effects (Kim et al, 2001; Koh et al, 2003). Tsutsumi et al (2003) reported that fangchinoline has antihyperglycemic effect; while tetrandrine has no such effect. Tetrandrine was also reported to be potentially useful as a chemotherapeutic/chemopreventive agent in hepatocellular carcinoma (Yoo et al, 2002; Oh and Lee, 2003). 87 OMe MeO Me N H H OR O N Me O OMe R= Me tetrandrine R= H O O fangchinoline CO2H NO2 OMe aristolochic acid I Figure 4.1 Chemical structures of tetrandrine, fangchinoline and aristolochic acid I. 88 Aristolochic acids (AA), a mixture of structurally related nitrophenanthrene carboxylic acids, with AAI being the major component, were reported to be nephrotoxin and carcinogen. In February 1993, Vanherweghem reported that many young women patients, after taking a slimming regimen that included Chinese herbs, experienced renal failure (Vanherwegnem et al, 1993). The nephropathy that was characterized by a rapid deterioration in renal function and biopsy displayed extensive interstitial fibrosis without glomerular lesions, as well as atrophy and loss of tubules. The nephritic symptoms are initially called Chinese Herbs Nephropathy (CHN) (Cosyns, et al, 1994). Suspicion that the disease was due to the introduction of Chinese herbs in the slimming regimen was reinforced by identification in the slimming pills of the nephrotoxic and carcinogenic aristolochic acid extracted from species of Aristolochia (Vanherwegnem et al, 1994). This hypothesis was substantiated by the identification of pre-mutagenic AA-DNA adducts in the kidney and ureteric tissues of CHN patients. Finally, induction of clinical features typical of CHN in rodents given AA alone removed any doubt on the causal role of this phytotoxin in CHN, now better called aristolochic acid nephropathy (AAN) (Cosyns et al, 2003). Nortier et al (2000) also found cancer in 18 of 39 Belgians who had been diagnosed with end-stage kidney failure caused by Aristolochia fangchi. Outside the Belgian epidemic, at least 65 cases of CHN have been reported worldwide and most of these cases could be classified as AAN patients (Cosyns et al, 2003). Several countries, including Canada, Australia, Germany and UK, banned the use of herbs containing AA (Kessler, 2000). The roots of Stephaniae tetrandra and Aristolochia fangchi are similar in appearance and their products in the markets are usually sold in the form of slices. The physical differentiation between the two herbs is difficult. The quantification of fangchinoline 89 and tetrandrine in Stephaniae tetrandra (Yang et al, 1998 and Li et al, 2004) and aristolochic acids in Aristolochia fangchi (Ong et al, 2000; Hashimoto et al, 1999; Ioset et al, 2003; and Hanna, 2004) have been extensively studied. However, there are no comparative studies on analysis of both herbs. The purpose of this work is to develop HPLC-DAD and LC/MS methods for the differentiation of Stephania tetrandra and Aristolochia fangchi. 4.2 Experimental 4.2.1 Chemicals and materials Aristolochic acid I, fangchinoline and tetrandrine were obtained from the National Institute for the Control of Pharmaceutical and Biological Products (NICPBP) (Beijing, China). Methanol and acetonitrile were of HPLC grade. Milli Q water (Millipore, France) was used. Stephania tetrandrae (Fangji) was bought from Tongrentang (Beijing, China), and the reference Aristolochia fangchi (Guangfangji) was from NICPBP (Beijing, China) Ten dried and sliced Stephania tetrandrae (Fangji) (as claimed) were purchased from local medical halls (Table 4.1). 4.2.2 Sample preparation 2 grams of powdered herbs were accurately weighed and were ultrasonicated with 20 ml methanol for 20 min. The process was repeated 3 times for each herb. After filtration, the combined methanol extracts were evaporated to dryness by a rotary evaporator. The residue was dissolved in 10 ml methanol and filtered by 0.45 µm membrane before analysis. 90 Table 4.1 Local sources of herbs bought as Fangji (Stephania tetrandra) No Source Address Price (SIN$) 1 Eu Yan Sang (Singapore) PTE LTD 269, South Bridge Road 1.00 2 Wong Yiu Nam Medical Hall PTE LTD 51 Temple Street 1.60 3 Sinchong Meheco LTD #03-19 Pearl’s Center 2.20 100 EU Tong Seng Street 4 Thye Shan Medical Hall PTE Ltd 201 New Bridge Road 2.00 5 Teck Soon Medical Hall PTE Ltd 231 South Bridge Road 1.60 6 TCM Chinese Medicines PTE Ltd #02-32 Pearl’s Center 1.00 100 EU Tong Seng Street 7 Teck Soon Medical Hall No. 22 Smith Street 1.00 8 Heng Say Tong Medical Hall PTE Ltd 101, Upper Cross Street 1.50 #01-29 People’s Park Center 9 Teck Yin Soon Chinese Medical Hall 71 Temple Street 1.20 10 Ban Tai Loy Medical Hall 276 South Bridge Road 1.60 91 4.2.3 HPLC-DAD conditions The analysis was performed using an Agilent 1100 HPLC series. The column used was an Inertsil ODS-3 column (4.6 mm × 250mm, 5 µm), maintained at 35°C. The mobile phase was 0.01% H3PO4 (v/v) in Milli Q water (pH 3.0) (A) and acetonitrile (B) with a gradient program as follows: 5%B to 100% B (55 min), 100% B to 100% B (5 min), post-run (10 min) at a flow rate of 1ml/min. The injection volume for all samples was 5 µl. Detection wavelength was 210 nm. UV scan was set from 200 to 400 nm. 4.2.4 LC/MS conditions A Spectra Series HPLC (Finnigan, San Jose, USA ) equipped with an autosampler was used. Separation were performed using on a Luna 5u C18(2) (2.0 × 150 mm, 5 µ), at a flow rate of 0.3 ml/min. The mobile phase was 0.1% acetic acid (v/v) in Milli Q water (A) and methanol (B) with a gradient program: 5%-100% (B) in 20 minutes. Mass spectrometric analysis was performed on a Finnigan LCQ ion trap mass spectrometer (San Jose, USA) equipped with an air-pressured chemical ionization (APCI) interface. Positive ionization mode was applied. The following APCI parameters was optimized for detection of AAI, tetrandrine and fangchinoline: vaporizer temperature (ramped from 150°C to 500°C in steps of 50°C), capillary temperature (increased from 125°C to 275°C in steps of 25°C) and tube lens offset (increased from -50V to 50V in steps of 10V). The HPLC fluid was nebulized using N2 as both the sheath gas at a flow rate of 60 au (arbitrary units), and auxiliary gas at a flow rate of 5 au. Collision induced dissociation (CID) experiments, MS-MS, were conducted using helium as the collision gas, and the relative collision energy was set at 35-60%. 92 4.3 Results and discussion 4.3.1 HPLC-DAD analysis 4.3.1.1 HPLC profiling of roots of Aristolochia fangchi and Stephania tetrandra Figure 4.2 shows the HPLC chromatograms of aristolochic acid I and an extract of reference Aristolochia fangchi. The peak eluting at 35.7 min (peak 1) was first checked by comparing the normalized spectra at the peak upslope, apex and downslope. The peak upslope spectrum was subtracted from the apex spectrum to eliminate small spectral irregularities arising from the sample matrix, the so-called chemical noise. Then the corrected spectrum was compared with the reference AAI standard. The match factor was more than 999, which indicates that the spectra are similar. The HPLC profiles of roots of Aristolochia fangchi (NICPBP) (Figure 4.2b) and Stephania tetrandra (Tongrentang) (Figure 3c) were clearly different. The marker compounds, tetrandrine, fangchinoline and AAI, were also successfully identified by comparison with the reference chemical standards. In Figure 4.3, the retention times of peak1 and peak2 (14.1 and 15.1 min respectively) were comparable to those of the chemical standards, fangchinoline and tetrandrine. Their UV match factors were also better than 950, indicating that there is high similarity. The UV absorbance of a compound might depend on the physical and chemical characteristics of the solvent in which the compound is dissolved. For HPLC, this means that the spectrum might be dependent on the mobile phase. Therefore, it is strongly recommended that the spectrum of reference standard should be obtained under the same condition as the sample (Huber L and George S A, 1993). 93 (a) mAU AAI 5 0 -5 10 10 0 mAU 80 20 30 40 50 min 40 50 min (b) 1 60 40 20 0 10 0 20 30 (c) Norm * Peak 1 in GuangFJ_Ref * AAI 50 Match factor: 999.849 40 30 20 10 0 200 225 250 275 300 325 350 375 nm Figure 4.2 HPLC chromatograms of (a) reference standard of AAI and (b) reference herb of Guangfangji, and (c) overlaid UV spectra of peak 1 (continuous line) and that of AAI (broken line). The RT and UV spectrum of peak 1 in Guangfangji match those of AAI. 94 (a) mAU Tetrandrine 100 50 0 0 10 20 30 40 50 min 20 30 40 50 min 20 30 40 50 min (b) mAU Fangchinoline 100 50 0 0 10 (c) mAU 1 2 800 400 0 0 10 (d) (e) Norm *DAD1, 14.077 (787 mAU,Apx) *Fangchinoline 600 Norm *DAD1, 15.090 (1090 mAU,Apx) *tetrandrine 1000 800 Match factor: 983.609 Match factor: 968.041 600 400 400 200 200 0 0 200 250 300 350 nm 200 250 300 350 nm Figure 4.3 HPLC chromatograms of (a) reference standard of tetrandrine, (b) reference standard of fangchinoline, and (c) Fangji from Tongrentang, and UV spectra match of (d) peak 1 with reference standard of fangchinoline, (e) peak 2 with reference standard of tetrandrine. The RT of peaks 1 and 2 in Fangji from Tongrentang math those of fangchinoline and tetrandrine respectively. 95 4.3.1.2 HPLC analysis of local Fangji samples Using the same chromatographic conditions, the HPLC profiles of the extracts of 10 local samples of Fangji (or Stephania tetrandra) were compared to those of the extracts of the reference herb Aristolochia fangchi (from NICPBP) and that of the extract of Stephania tetrandra (from Tongrentang, Beijing). The results were intriguing. The HPLC profiles of extracts of 9 local Stephania tetrandra samples were found to be similar to that of the reference fangchi. They were clearly different from the HPLC profiles of the extract of Stephania tetrandra (Figure 4.4). The HPLC profile of the extract of sample 10 was found to be different from those of Aristolochia fangchi (Figure 4.4) and Stephania tetrandra (Figure 4.3c). The marker compounds of Stephania tetrandra, namely, tetrandrine and fangchinoline, could not be found in all the ten herbal samples. Instead, AAI was found to be present in all of the local Fangji samples except sample 10. The retention time and UV spectra of the suspected peaks were compared with those of the reference standards. Retention time of suspected AAI peak was also checked by spiking AAI standard into the herbal sample. Figure 4.5 shows the overlaid chromatograms of the sample from SINC and the same sample spiked with AAI standard. The suspected AAI peak (peak 1) has the same retention time with AAI standard. Its UV spectrum also has some similarity with that of AA I standard (Figure 4.6). However, the match factor is about 743, not as high as that of the reference herb, whose match factor is 999.8. This could be due to the much lower concentration of AAI in the sample from SINC than that in the reference Fangji (Figure 4.6). The on-line MS detector offers superiority over DAD detection in term of specificity and sensitivity. Thus it is necessary to confirm the presence of AAI in local Fangji 96 samples using MS detection, to provide further evidence that the local Fangji samples may be indeed Aristolochia fangchi instead of Stephania tetrandra (i.e. what they should be). This is important as Aristolochia fangchi is known to contain toxic aristolochic acids. 4.3.2 LC-MS/MS analysis 4.3.2.1 Optimization of APCI parameters The LC/MS interface parameters were optimized in order to improve sensitivity and selectivity. This was done in the positive mode by taking into consideration of the vaporizer and the capillary temperature. The influence of the studied parameters on the response of the target analytes and the interaction of these parameters were evaluated based on the ammonium adduct molecular ion [M+NH4]+ m/z 359 for AAI, protonated molecular ion [M+H]+ m/z 623 for tetrandrine and [M+H]+ m/z 609 for fangchinoline in the positive mode. The heated nebulizer of the APCI interface generates a rapid evaporating mist of droplets prior to corona discharge ionization. The spray performance is controlled by the vaporizer temperature. The results show that AAI signal is highest when a vaporizer temperature is 300°C (Figure 4.7a), while fangchinoline and tetrandrine signals are highest when vaporizer temperature is 350°C (Figure 4.7b). However, AAI has a much (around 10000 folds) lower signal than tetrandrine and fangchinoline at the same concentration (Figure 4.7). Consequently, optimized parameters were based on AAI detection. Thus a vaporizer temperature of 300°C was chosen. 97 Gurangfangji_reference #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 Guangfangji_Guangzhou 0 10 20 30 40 50 min Figure 4.4 Overlaid chromatograms of reference Guangfangji, Guangfangji from Guangzhou and 10 local Fangji samples (Table 4.1). 98 mAU 6 4 (a) Peak 1 2 0 -2 -4 -6 -8 -10 32 34 36 38 40 min 38 40 min mAU 4 (b) Spiked with AAI 2 0 -2 -4 -6 -8 -10 32 Peak 1 34 36 Fangji_SINC Spiked with AAI 0 10 20 30 40 50 min Figure 4.5 Overlaid chromatograms of Fangji_SINC and that spiked with AAI standard. Peak 1 is the suspected AAI peak. (a) and (b) are the enlarged parts of the chromatograms from RT 30.5 to 41 min. 99 Norm 3 Norm (b) (c) *DAD1, 35.6 of FJ_SINC *AA *DAD1, 35.7 of GuangFJ_Ref *AAI 50 Match factor: 743.039 2.5 Match factor: 999.849 40 2 30 1.5 20 1 10 0.5 0 0 200 225 250 275 300 325 350 375 nm 200 225 250 275 300 325 350 375 nm (a) Peak 1 Fangji_SINC Guangfangji_Ref 0 10 20 30 40 50 min Figure 4.6 (a) Overlaid chromatograms of the extracts of local Fangji sample from SINC and that of the reference Guangfangji. (b) Overlaid UV spectra of peak at 35.7 min of reference Guangfangji (continuous line) and that of AAI (broken line). (c) Overlaid UV spectra of peak 1 in Fangji sample from SINC (continuous line) and that of AAI (broken line). 100 (a) 4.50E+06 Ion abundance (count) 4.00E+06 3.50E+06 3.00E+06 2.50E+06 2.00E+06 1.50E+06 1.00E+06 5.00E+05 0.00E+00 100 200 300 400 500 600 Vaporizer temperature (°C) (b) Tetrandrine Fangchinoline 4.00E+10 Ion abundance (count) 3.50E+10 3.00E+10 2.50E+10 2.00E+10 1.50E+10 1.00E+10 5.00E+09 0.00E+00 100 200 300 400 500 600 Vaporizer temperature Figure 4.7 Effect of vaporizer temperature on abundance of (a) [M+NH4]+ m/z 359 for AAI, and (b) [M+H]+ m/z 623 for tetrandrine and [M+H]+ m/z 609 for fangchinoline 101 After evaporating the solvent in the ionization chamber the analytes reach the first vacuum stage via the capillary. While the analytes never reach the set vaporizer temperature (due to the evaporation process), the analytes are fully exposed to the capillary temperature. Consequently, the thermally labile analytes will be more affected by the capillary temperature than by the vaporizer temperature. Figure 4.8a shows that AAI signal is best when a capillary temperature is applied at 150°C. In Figure 4.8b, higher fangchinoline and tetrandrine signals are achieved when applying lower capillary temperature in the range of 125 to 275°C. No lower capillary temperature was investigated due to the fact that too low a temperature in some cases might be insufficient for keeping the analytes in vaporized state. Thus 150°C was chosen as the optimized capillary temperature. In order to further improve target ion signals in positive mode, the effect of the tube lens offset was also evaluated. From figure 4.9a, it is evident that the value of zero V can provide the highest AAI signal. However, as Figure 4.9b shows, there is no much influence of tube lens offset on the abundance of fangchinoline and tetrandrine signals. Thus, zero V was selected as tube lens offset. The achieved optimized instrument APCI parameters were hence vaporizer temperature 300°C, capillary temperature 150°C and tube lens offset 0V. 102 (a) 3.00E+06 Ion abundance (count) 2.50E+06 2.00E+06 1.50E+06 1.00E+06 5.00E+05 0.00E+00 100 150 200 250 300 Capillary temperature (°C) (b) Tetrandrine Fangchinoline 3.50E+10 3.00E+10 Ion abundance 2.50E+10 2.00E+10 1.50E+10 1.00E+10 5.00E+09 0.00E+00 100 150 200 250 300 Capillary temperature Figure 4.8 Effect of capillary temperature on abundance of (a) [M+NH4]+ m/z 359 for AAI, and (b) [M+H]+ m/z 623 for tetrandrine and [M+H]+ m/z 609 for fangchinoline 103 (a) 2.60E+06 Ion abundance (count) 2.40E+06 2.20E+06 2.00E+06 1.80E+06 1.60E+06 1.40E+06 1.20E+06 1.00E+06 -60 -40 -20 0 20 40 60 40 60 Tube lens offset (V) (b) Tetrandrine Fangchinoline 3.60E+10 Ion abundance (count) 3.10E+10 2.60E+10 2.10E+10 1.60E+10 1.10E+10 6.00E+09 1.00E+09 -60 -40 -20 0 20 Tube lens offset (V) Figure 4.9 Effect of tube lens offset on abundance of (a) [M+NH4]+ for AAI, and (b) [M+H]+ for tetrandrine and fangchinoline 104 4.3.2.2 Mass spectra of the reference standards In an attempt to obtain mass spectrometric-based information for confirming the marker compounds present in the roots of Aristolochia fangchi and Stephania tetrandra, direct flow injection experiments of reference standards of AAI, fangchinoline and tetrandrine were performed under the optimized APCI conditions. In the full scan mode MS spectrum (Figure 4.10a) of AAI, both protonated ([M+H]+, m/z 342) and ammoniated ([M+NH4]+, m/z 359) molecules were generated together with a fragment ion, [(M+H)-18]+ (m/z 324). The ammoniated molecule ion was the base peak. Full product ion spectrum of precursor ion at m/z 359 (the ammoniated molecular ion of AAI) was recorded (Figure 4.10b). Protonated molecular ion, [(M+NH4)-NH3]+ m/z 342, was generated by loss of ammonia. Further loss of water from the precursor ion was very common, so the [(M+NH4)-NH3-H2O]+ (m/z 324) ion was found. Another two specific product ions (m/z at 298 and 296) were also generated. The [(M+NH4)-44]+ ion due to the loss of carbon dioxide, while the [(M+NH4)-46]+ was due to loss of HCOOH. The fragment patterns are in good agreement with those previously reported using LC/ESI-MS (Kite et al, 2002). These product ions provide useful confirmatory information for AAI. In the full scan mode MS spectra of fangchinoline and tetrandrine (Figure 4.11), only protonated ions [M+H]+ (m/z 609 for fangchinoline and m/z 623 for tetrandrine respectively) were generated. With these protonated molecular ions as precursor ions, the CID tandem spectra (MS/MS) (Figure 4.12) showed more confirmatory product ions. 105 [M+NH4]+ 358.9 100 (a) 90 Relative Abundance 80 70 60 50 40 30 [M+H-18]+ [M+H]+ 324.2 341.8 20 10 0 60 80 100 120 140 160 180 200 220 m/z 240 260 280 298.0 100 300 320 340 360 380 400 [(M+NH4)-NH3 -CO2]+ (b) 90 80 Relative Abundance 70 60 50 [(M+NH4)-NH3-H2O]+ 324.1 40 [(M+NH4)-NH3 -HCOOH]+ 30 [(M+NH4)-NH3]+ 296.0 341.8 20 10 0 100 120 140 160 180 200 220 240 260 m/z 280 300 320 340 360 380 400 Figure 4.10 (a) Full scan mode MS spectrum of AAI; (b) MS/MS spectrum of AAI, with precursor ion, [M+NH4]+ m/z 359. 106 609.2 100 90 (a) 80 Relative Abundance 70 60 50 40 30 20 10 0 200 250 300 350 400 450 m /z 500 550 600 623.2 100 90 650 (b) Relative Abundance 80 70 60 50 40 30 20 10 0 200 250 300 350 400 m/z 450 500 550 600 650 Figure 4.11 Full scan mode MS spectra of (a) fangchinoline and (b) tetrandrine 107 (a) [(M+H)-31]+ 578.3 100 90 80 Relative Abundance 367.1 [(M+H)-43]+ 70 566.2 60 50 40 [(M+H]-63]+ 30 [M+H]+ 546.3 609.3 20 382.0 10 0 250 200 300 350 400 450 500 550 600 650 m/z [(M+H)-31]+ (b) 592.3 100 90 80 Relative Abundance 70 60 [(M+H)-43]+ 580.2 50 40 [M+H]+ 30 623.3 20 [(M+H)-63]+ 561.3 10 0 200 250 300 350 400 450 500 550 600 650 m/z Figure 4.12 MS/MS spectra of (a) fangchinoline with precursor ion m/z 609 and (b) tetrandrine with precursor ion m/z 623. 108 4.3.2.3 LC-MS/MS Analysis of Aristolochia fangchi (NICPBP) and Stephania tetandra (Tongrentang) Using the optimized APCI method, reference Aristolochia fangchi and Stephania tetrandra (Tongrentang) were analyzed. Figure 4.13 shows a typical total ion chromatogram (TIC) of reference Aristolochia fangchi (NICPBP, Beijing) (Figure 4.13a) and related on-line mass spectrum of the peak of AAI with retention time at 15.21min, obtained in positive ion mode (Figure 4.13b). The major ions in the full scan mode are the ammoniated ([M+NH4]+, m/z 359) and protonated ([M+H]+, m/z 342) molecules, and a fragment ion, [(M+H)-18]+ (m/z 324). The ammoniated molecule was further selected. MS/MS experiment was performed. Figure 4.13c shows the chromatogram of product ions of the ammoniated molecule ion. The secondary spectrum (Figure 4.13d) shows all the characteristic product ions: m/z 342 for protonated molecule ion, m/z 324 for the ion due to further loss of water, m/z 298 for the ion further loss of CO2 and m/z 296 for the ion further loss of HCOOH. The full scan mode mass spectrum and MS/MS spectrum are consistent with those of AAI standard (Figure 4.10). 109 15.21 100 (a) Relative Abundance 80 60 40 20 0 0 2 4 6 8 10 12 14 16 18 Time (min) Relative Abundance 359.0 (b) 100 80 324.5 60 342.2 40 20 0 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 m/z 100 15.24 (c) Relative Abundance 80 60 40 20 0 0 2 4 6 8 10 12 14 16 18 Time (min) Relative Abundance 100 298.0 (d) 80 60 296.1 40 324.1 20 341.7 0 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 m/z Figure 4.13 LC/MS/MS analysis of reference Guangfangji: (a) Total ion chromatogram (TIC), (b) Full scan mode mass spectrum, (c) Ion chromatogram of reference Guangfangji, [M+NH4]+ m/z 359, and (d) MS/MS spectrum. The data agrees with MS data from AAI (Figure 4.10), hence confirming the presence of AAI in the reference Guangfangji. 110 Figure 4.14 shows a typical TIC of Stephania tetrandra (Tongrentang, Beijing) (Figure 4.14a) and related full scan mass spectra of the peaks of tetrandrine and fangchinoline with retention time at 4.9 min and 4.4 min respectively, both obtained in positive ion mode (Figure 4.14 b and c). Only protonated molecular ions were detected, m/z 623.3 for tetrandrine and m/z 609.4 for fangchinoline. The protonated molecules were further selected. MS/MS experiments were performed. Figure 4.15 shows the chromatograms of product ions of the protonated molecular ions (Figure 4.15a and 4.15c). The secondary spectra (Figure 4.15b and 4.15d) show all the characteristic product ions. The full scan mode mass spectra and MS/MS spectra are consistent with those of tetrandrine and fangchinoline standards (Figure 4.11 and 4.12). 111 4.9 100 Relative Abundance 80 (a) 60 4.4 40 20 0 0 2 4 6 8 10 Time (min) 12 14 16 18 20 623.3 Relative Abundance 100 80 60 (b) 40 20 0 100 200 300 400 500 600 m/z 609.4 Relative Abundance 100 80 (c) 60 40 20 0 100 200 300 400 500 600 m/z Figure 4.14 LC-MS analysis of Fangji (Tongrentang): (a) Total ion chromatogram, (b) Mass spectrum of peak at 4.9 min, and (c) mass spectrum of peak at 4.4 min. The MS data agrees with those of the standards fangchinoline and tetrandrine (Figure 4.11), hence confirming the presence of these two standards in the extract of Fangji (Stephania tetrandra) from Tongrentang. 112 (a) 4.4 Relative Abundance 100 80 60 40 20 0 0 2 4 6 10 8 Relative Abundance (b) 16 18 578.2 80 367.1 60 566.3 40 20 200 250 300 350 400 4.9 609.4 546.3 381.9 100 Relative Abundance 14 100 0 450 500 550 600 650 m/z (c) 80 60 40 20 0 100 Relative Abundance 12 Time (min) 0 2 4 6 8 10 Time (min) 12 14 16 18 592.3 (d) 80 60 580.2 40 20 0 623.3 561.2 200 250 300 350 400 m/z 450 500 550 600 650 Figure 4.15 LC-MS/MS analysis of Fangji (Tongrentang): (a) ion chromatogram, m/z 609; (b) MS/MS spectrum in full scan mode, precursor ion m/z 609; (c) ion chromatogram, m/z 623; and (d) MS/MS spectrum in full scan mode, precursor ion m/z 623. The data agrees with those obtained from the fangchinoline and tetrandrine standards (Figure 4.11 and 4.12), hence confirming the presence of these 2 standards in the extract of Fangji (Tongrentang). 113 4.3.2.4 LC-MS/MS analysis of local samples In the previous section (4.3.1.2), HPLC analysis indicated that 9 of the 10 local Fangji samples may be Aristolochia fangchi instead of the expected Stephania tetrandra. AAI was found to be present in all but 1 of the local Fangji samples; while tetrandrine and fangchinoline were not found in any of them. Figure 4.16a shows a typical total ion chromatogram of local Fangji sample. The fall scan mode mass spectrum of peak at 15.14min (Figure 4.16b) exhibited the characteristic ammoniated molecular ion, m/z 359 ([M+NH4]+), protonated ion, m/z 342 ([M+H]+) and a fragment ion at m/z 324 ([(M+H)-H2O]+). Figure 4-16c shows the ion chromatogram with precursor ion m/z 359. The secondary spectrum of peak at 15.07 min shows the characteristic product ions, m/z 342 ([(M+NH4)-NH3]+), m/z324 ([(M+NH4)-NH3-H2O]+), m/z 198 ([(M+NH4)-NH3-CO2]+), and m/z 296 ([(M+NH4)-NH3-HCOOH]+). The content of AAI in these samples are at trace level, not as much as that in the reference Aristolochia fangchi (NICPBP, Beiing) (Figure 4.13). These results are consistent with those obtained from HPLC DAD analysis (Figure 4.4). 114 Relative Abundance 80 (a) 60 15.14 40 20 0 4 6 8 10 12 14 16 Time (min) 358.9 Relative Abundance 100 (b) 80 60 341.6 40 324.1 20 0 220 240 260 280 300 320 340 360 380 m/z 100 Relative Abundance 80 (c) 60 40 15.07 20 0 0 2 4 6 8 10 Time (min) 12 14 18 298.0 100 Relative Abundance 16 (d) 80 296.2 60 40 324.0 20 0 100 120 140 160 180 200 220 240 m/z 260 280 300 320 341.7 340 359.1 360 380 400 Figure 4.16 LC-MS/MS analysis of a typical local Fangji sample (no 3). (a) Total ion chromatogram; (b) full scan mode mass spectrum of the peak marked as arrow in (a) ; (c) ion chromatogram, m/z 359; and (d) MS/MS spectrum of the peak marked as arrow in (c), precursor ion m/z 359. 115 4.4 Conclusion The HPLC-DAD and LC/MS methods developed are suitable for the differentiation of Stephania tetrandra (Fangji) and Aristolochia fangchi (Guangfangji). Ten herbal samples bought as Fangji (or Stephania tetrandra) were analyzed by these methods. In nine of the ten herbal samples, AAI was detected by HPLC-DAD and confirmed by LC/MS/MS. In addition, the marker compounds of Stephania tetrandra (fangchinoline and tetrandrine) were not detected in any of these 10 samples. The evidences indicate that the nine samples bought locally are Aristolochia fangchi instead of Stephania tetrandra. This case has been reported to Health Sciences Authority. Further work need to be done to further elucidate the identity of the sample (sample 8) that did not appear to resemble Aristolochia fangchi or Stephania tetrandra. Most of the herbs bought in Singapore under the name of Fangji (Stephania tetrandra) were found to contain the toxic aristolochic acid I, and not the expected tetrandrine and fangchinoline. This study underscores, once again, the necessity of the introduction of measures to control the quality and safety of herbs. 116 Chapter 5 Conclusion Analysis of complex mixtures of botanical origins is very challenging. The work presented in this thesis has clearly demonstrated that methods for the analysis of some commonly used herbs in Singapore and around the region, have been developed and have been shown to be useful for the quality control of such botanicals and their products. The methods developed can be extended to other complex mixtures of botanical origins with appropriate optimization, when necessary. Chapter 1 gives a general introduction to the importance of herbal medicine, the increasing need for quality control and highlights some common analytical methods used for quality control of herbal medicine. The specific objectives are also stated, namely, to develop methods for the analysis of G. elata and its products; for chromatographic fingerprinting of 11 different commonly used herbs and finally, for differentiating between 2 plants with the same common name, both available locally but one of which is toxic. In Chapter 2, a rapid and sensitive HPLC method for the detection of 4 biologically active chemical markers in G. elata, namely, gastrodin, 4-hydroxbenzyl alcohol and 4hydroxybenzaldehyde and L-pyroglutamic acid has been successfully developed. To the author’s knowledge, this is the first report of simultaneous detection of 4 active constituents in this medicinal plant. The contents of gastrodin, which is the main active ingredient, hasve also been determined in the plant and in CPM products. This method has been validated for linearity, precision, accuracy, LOD and LOQ. For the CPM, the 117 presence of other plants resulted in the gastrodin peak being co-eluted with other herbal ingredients. Hence a SPE sample cleanup is successfully developed and employed. Chapter 3 addresses the usefulness of chromatographic fingerprinting in the differentiation of commonly used herbs. The HPLC chromatographic fingerprint can be used for quality control of herbal medicines, especially when the marker compounds are not available. In the present work, a HPLC chromatographic fingerprinting method is successfully developed and applied to 34 botanical samples, constituting 11 different types of herbal medicine. Hierachical analysis yields interesting results. 4 herbs (namely, Madouling 马兜玲, Huangqin 黄芩, Tianma天麻 and Banxia 半夏) give distinctive and consistent chemical fingerprints. Differences between the same herbs from different sources have been detected for botanicals such as Duzhong (杜仲) and Banlangen (板蓝 根). The results for the “Fangji”, Guangfangji (广防己) and Hanzhongfangji (汉中防己) samples indicate possible misidentification / substitution and deserve further investigations. For complex mixtures that are not well resolved in this general protocol, clearly, further optimisation of the method is necessary. This is evident from the similar chemical fingerprints of the extracts of Shengdi (生地) and Shudi (熟地). The latter is essentially the steamed form of the former. HPLC-DAD has been proven to be a useful tool for identification of marker compounds in herbal medicines, based on its ability of providing three-dimension information: retention time and UV spectrum. Eight biologically active compounds are employed as biomarkers and are screened in these herbal samples. Aristolochic acid I (AAI), a nephrotoxic component of Guangfangji, is found in the local Fangji samples. 118 Liquid chromatography with mass spectrometry/mass spectrometry detection (LCMS/MS) offers superiority over HPLC-DAD in terms of specificity. The secondary MS spectrum can provide confirmatory information. HPLC-DAD and LC-MS/MS methods have been further developed for differentiation of 2 medicinal plants, namely Fangji 防己 (Stephania tetrandra) and Guangfangji 广 防 己 (Aristolochia fangchi) in Chapter 4. Tetrandrine and fangchinoline, bioactive components of Stephania tetrandra, and aristolochic acid I (AAI), a nephrotoxic component of Aristolochia fangchi, are successfully used for differentiation of these two herbs. Using these techniques, 9 out of 10 local herbal samples bought as Fangji (Stephania tetrandra) are found to be the toxic Aristolochia fangchi instead. Fangchinoline and tetrandrine, the marker compounds of Stephania tetrandra, are not detected in them, while AAI is found to be present in most of these samples. The Health Sciences Authority has been informed of this finding. This study also indicates the urgent need for greater quality control of natural products. It is hoped that through the cooperation of academia, researchers, regulators, practitioners, patients, and the industry, the benefits of botanical products can be exploited while reducing risks of toxicity or adverse reactions. The bioactive components can be potentially used as markers for quality control of herbal medicines. However, in most cases, the bioactive components of a herb used as markers may not reflect the efficacy of the whole herb. As new bioactive constituents are isolated, the analytical methods need to be updated too. The components of the herbs from the same species can usually have some similarity. For example, aristolochic acids are the common components of many herbs from the family of “Aristolochia”, such as Aristolochia fangchi (Guangfangji 广防己), Aristolochia recurvilabra (Qingmuxiang 清 119 木香) and Aristolochia manshuriensis (Guanmutong 关木通). In such a case, aristolochic acid alone as marker is not enough to indicate the authenticity of the herb. Other methods, such as chromatographic fingerprinting, DNA fingerprinting, need to be further developed. Future work can involve a combination of these techniques and obtaining samples from the sources of origin (i.e. from the plantations). Herbal medicines continue to be widely used by people throughout the world. The usage is increasing rapidly and this has important medical and socio-economic implications. Assessment of the safety, quality and efficacy of herbal medicines is an important issue for health professionals and regulators. Academics/researchers, health professionals, consumers, regulators and people from industries must work together to ensure the safety and quality of herbal medicines. The entire process of quality control of medicinal plants and their products should involve good agricultural practice, good storage /supply practices and good manufacturing practice etc. More collaborative research work needs to be done to establish a good and feasible quality control framework. This framework should include guidelines endorsed by international organisations such as WHO; appropriate policies adapted to the specific requirements of herbal medicines and regulated by the regulatory agency in each country and be followed by the farmers/suppliers/manufacturers, to achieve good quality, safety and efficacy in complex mixtures of botanical origins. 120 BIBLIOGRAPHY Altria KD, Elder David. Overview of the status and application of capillary electrophoresis to the analysis of small molecules. J Chromat A. 2004; 1023: 1-14. Andersson M, Bergendorff O, Nielsen M, Sterner O, Witt R, Ai JL, Lu AP, Wang AM. Inhibition of kainic acid binding to glutamate receptors by extracts of Gastrodia. Phytochem. 1995; 38(4): 835-836. Bannerman R, Burton J, Chen WC. Traditional medicine and healthcare coverage, World Health Organization, Geneva, Switzerland, 1983. Beebe KR (ed.). Chemometrics : a practical guide. New York: Wiley, 1998. Brereton RG (ed.). Multivariate Pattern Recognition in Chemometrics. Elsevier, Amsterdam, 1992. Cao J, Xu Y, Zhang YZ, Wang YM and Luo GA. Fingerprint chromatogram by high performance liquid chromatography with evaporative light scattering detection of Qingkailing injection and its relationship with quality control. Fenxi Huaxue. 2004; 32(4): 469-473. Chen AJ, Zhang JY, Li CH, Chen XF, Zhi DH and Chen XG. Separation and determination of active components in Radix Salviae miltiorrhizae and its medicinal preparations by nonaqueous capillary electrophoresis. Journal of Separation Science. 2004, 27(7-8): 569-575. 121 Chen MJ, Wu YJ, Fan XH and Chen YY. Identification of the habitat of Ligusticum chuanxiong with chromatographic fingerprint. Zhongguo Zhongyao Zazi. 2003; 28(7): 606-610. Cheng S, Liang C, Ran M, Ye Y, Hu X, Wang F and Xiong H. Studies on quality control of the Chinese crude drug Tian Ma (Gastrodia elata BL.) I. HPLC analysis of gastrodin as a quality indicator. Yaowu Fenxi Zazhi. 1984; 4(6): 344-347. Cheng YY, Chen MJ and Welsh WJ. Fractional fingerprint of chromatographic profiles based on wavelet analysis and its application to characterize the quality grade of medicinal herbs. Journal of chemical information and computer sciences. 2003; 43(6): 1959-1965. Cheng YY, Chen MJ and Tong WD. An approach to comparative analysis of chromatographic fingerprints for assuring the quality of botanical drugs. Journal of chemical information and computer sciences. 2003a; 43(3): 1068-1076. Choi HS, Kim HS, Min KR, Kim Y, Lim HK, Chang YK and Chung MW. Antiinflammatory effects of fangchinoline and tetrandrine. J Ethnopharmacol. 2000; 69(2):173-179. Cosyns JP, Jadoul M, Squifflet JP, et al. Chinese herbs nephropathy: a clue to Balkan endemic nephropathy? Kidney Int. 1994; 45: 1680-1688. Cosyns JP. Aristolochic acid and ‘Chinese herbs nephrophathy’ A review of the evidence to date. Drug safety. 2003; 26(1): 33-46. Cui J, Garle M, Eneroth P, Bjorkhem I. What Do Commercial Ginseng Preparations Contain? Lancet 1994; 344:134. 122 Drug Administration Bureau of China. Requirements for studying fingerprint of traditional Chinese medicine injections. Beijing, 2000. Eisenberg DM, Davis RB, Ettner SL. Trend in alternative medicine uses in the United States, 1990-1997. Results of a follow-up National Survey. JAMA. 1998 (20): 1569-1575. Eskinazi D (ed.). Botanical Medicine – Efficacy, Quality Assurance and Regulation. Larchmont, NY: Mary Ann Liebert, 1999. Fang DC and Jiang MX. A new calcium antagonist of Chinese medicinal origin: tetrandrine. J Hypertens. 1986; 4(Suppl. 6): S150-S152. Farnsworth NR and Morris RW. Higher plants: the sleeping giant of drug discovery. Amer J Pharm 1976 (40): 115-121. Feng HT, Yuan LL and Li SFY. Analysis of Chinese medicine preparations by capillary electrophoresis-mass spectrometry. Journal of Chromatography, A. 2003; 1014(1-2): 8391. Feng XZ, Chen YW, Yang JS. Studies on constituents of Tian-Ma (Gastrodia elata Bl.). Huaxue Xuebao. 1979; 37(3): 175-182. Food and Drug Administration. FDA Announces Plans to Prohibit Sales of Dietary Supplements Containing Ephedra. FDA, USA, 2003. Food and Drug Administration. Guidance for Industry Botanical Drug Products. Center for Drug Evaluation and Research, FDA, USA, 2002. Gail BM, Harry HSF, Norman RF. Botanical dietary supplements: quality, safety and efficacy. Netherland: Swets & Zeitlinger, 2001. 123 Gong F, Liang YZ, Xie PS, Chau FT. Information theory applied to chromatographic fingerprint of herbal medicine for quality control. Journal of Chromatography, A. 2003; 1002(1-2), 25-40. Gong F, Ling YZ, Fung YS, Chau FT. Correction of retention time shifts for chromatographic fingerprints of herbal medicines. Journal of Chromatography, A. 2004; 1029(1-2): 173-183. Ha JH, Shin SM, Lee SK, Kim JS, Shin US, Huh K, Kim JA, Yong CS, Lee NJ and Lee DU. Planta Medica. 2001; 67(9): 877-880. Hanna GM. NMR regulatory analysis: determination and characterization of chinese-herb aristolochic acids. Pharmazie. 2004; 59(3): 170-174. Hao Q, Li Y and Li Q. Effects and quality-control of different varieties of Bupleurum. Journal of Chinese Clinical Medicine. 2002; 22(3): 1-3. Hayen H and Karst U. Strategies for the liquid chromatographic-mass spectrometric analysis of non-polar compounds. Journal of Chromatography A. 2003; 1000(1-2): 549565. Hennion MC. Solid-phase extraction: method development, sorbents, and coupling with liquid chromatography. Journal of Chromatography A. 1999; 856(1): 3-54. Hu Y, Liang YZ, Li BY, Xu CJ and Zeng ZD. Multicomponent spectral correlative chromatography applied to analysis of chemical fingerprints of traditional Chinese medicines. Huaxue Xuebao. 2003; 61(9): 1466-1470. Huang Yancai. Identification and Application of the Common-used Chinese Herbs. Chemical Industry Press, Wuhan, 2000. 124 Huber L and George A. Diode array detection in HPLC. New York: Marcel Dekker Inc., 1993. Ioset JR, Raolison GE, and Hostettmann. Detection of aristolochic acid in Chinese phytomedicines and dietary supplements used as slimming regimens. Food and Chemical Toxicology. 2003; 41: 29-36. Jiang TF, Li Y and Shi YP. Determination of six major flavonoid glycosides in Saussurea mongolico by capillary electrophoresis. Planta Medica. 2004; 70(3): 284-287. Kessler DA. Cancer and herbs. N Engl J Med. 2000; 342: 1742-1743. Kim SD, Oh SK, Kim HS and Seong YH. Inhibitory effect of fangchinoline on excitatory amino acids-induced neurotoxicity in cultured rat cerebellar granule cells. Arch Pharm Res. 2001; 24(2):164-170. Kim HS, Zhang YH, Fang LH, Yun YP and Lee HK. Effects of tetrandrine and fangchinoline on human platelet aggregation and thromboxane B2 formation. J Ethnopharmacol. 1999; 66(2):241-246. Kim HS, Zhang YH and Yun YP. Effects of tetrandrine and fangchinoline on experimental thrombosis in mice and human platelet aggregation. Planta Med. 1999a; 65(2):135-138. Kim HS, Zhang YH, Oh KW and Ahn HY. Vasodilating and hypotensive effects of fangchinoline and tetrandrine on the rat aorta and the stroke-prone spontaneously hypertensive rat. J Ethnopharmacol. 1997; 58(2): 117-123. King VF, Garcia ML, Himmel D, Reuben JP, Lam YT, Pan JX, Han GQ, and Korowski GJ. Interaction of tetrandrine with slowly inactivating calcium channels. 125 Characterization of calcium channel modulation by an alkaloid of Chinese medicinal herb origin. J Biol Chem. 1988; 263(5): 2238-2244. Kite GC, Yule MA, Leon C and Simmonds MSJ. Detecting aristolochic acid in herbal remedies by liquid chromatography/serial mass spectrometry. Rapid Communication in Mass Spectrometry. 2002; 16: 585-590. Koh HL, Teo HH and Ng HL. Pharmacists’ Patterns of uses knowledge and attitudes towards Complementary and Alternative Medicine. J Altern Complement Med. 2003; 5163. Koh SB, Ban JY, Lee BY and Seong YH. Protective effects of fangchinoline and tetrandrine on hydrogen peroxide-induced oxidative neuronal cell damage in cultured rat cerebellar granule cells. Planta Medica. 2003; 69(6): 506-512. Ku Y R, Lin Y T, Lin J H, Wen K C, Liao C H. Determination of parishin, parishin B and parishin C in traditional Chinese medicinal formulas by micellar electrokinetic capillary chromatography. J. Chromatogr., A. 1998; 805(1, 2): 301-308. Ku Y R, Lin Y T, Wen K C, Lin J H, Liao C H. Analysis of parishin, parishin B and parishin C in Gastrodiae Rhizoma by micellar electrokinetic capillary chromatography. J. Chromatogr., A. 1998a; 805(1, 2): 330-336. Lau AJ, Woo SO and Koh HL. Analysis of saponins in raw and steamed Panax notogenseng using high-performance liquid chromatography with diode array detection. Journal of Chromatography A. 2003; 1011: 77-87. Lavine BK. “Signal processing and data analysis” in Practical guide to Chemometrics, ed. Haswell SJ. Marcel Dekker, New York, 1992. 126 Li BY, Hu Y, Liang YZ, Huang LF, Xu CJ and Xie PS. Spectral correlative chromatography and its application to analysis of chromatographic fingerprints of herbal medicines. Journal of Separation Science. 2004; 27(7-8): 581-588. Li BY, Hu Y, Ling YZ, Xie PS and Du YP. Quality evaluation of fingerprints of herbal medicine with chromatographic data. Analytica Chimica Acta. 2004a; 514(1): 69-77. Li H X, Ding M Y, Lu K, Wei Y, Yu J Y. Identification and determination of the active compounds in Gastrodia elata Blume by HPLC. J. Liq. Chromatogr. Relat. Technol. 2001; 24(4): 579-588. Li H X, Ding M Y, Yu J Y. Simultaneous determination of p-hydroxybenzaldehyde, phydroxybenzyl alcohol, 4-(β -D-glucopyranosyloxy)-benzyl alcohol, and sugars in Gastrodia elata Blume measured as their acetylated derivatives by GC-MS. J. Chromatogr. Sci. 2001a; 39(6): 251-254. Li YQ, Cui SY, Cheng YQ, Chen XG, and Hu ZD. Application of nonaqueous capillary electrophoresis for quantitative analysis of quinolizidine alkaloids in Chinese herbs. Analytica Chimica Acta. 2004b; 508: 17-22. Lin JH, Liu YC, H JP, Wen KC. Parishins B and C from rhizomes of Gastrodia elata. Phytochemistry, 1996; 42(2): 549-551. Liu QY, Karpinski E and Pang PKT. Tetrandrine inhibits both T and L calcium channel currents in ventricular cells. J Cardiovasc Pharmacol. 1992; 20(4): 513-519. Liu QY, Karpinski E, Rao MR, Pang PKT. Tetrandrine: a novel calcium channel antagonist inhibits type I calcium channels in neuroblastoma cells. Neuropharmacology. 1991; 30(12A): 1325-1331. 127 Liu SY, Woo SO, Holmes MJ and Koh HL. LC and LC-MS-MS analysis of undeclared codeine in antiasthmatic Chinese proprietary medicine. J Pharm Biomed Anal. 2000; 22(3): 481-486. Ludwig H and Stephan AG. Diode array detection in HPLC. New York: M. Dekker, 1993. Maclennan, et al. Prevalence and cost of alternative medicine in Australia. Lancet, 1996; 347:569-573. Massart DL and Kaufman L. The interpretation of analytical chemical data by the use of cluster analysis. John Wiley & Sons, New York, 1983. Niessen WMA. Progress in liquid chromatography-mass spectrometry instrumentation and its impact on high-throughput screening. Journal of Chromatography A. 2003; 1000(1-2): 413-436. Nortier JL, Muniz Martinez MC, Schmeiser HH. New Eng. J. Med. 342: 1686, 1742, 2000. Nyquist RA. Interpreting infrared, Raman, and nuclear magnetic resonance spectra. San Diego: Academic Press, 2001. Oh SH and Lee BH. Induction of apoptosis in human hepatoblastoma cells by tetrandrine via caspase-dependent Bid cleavage and cytochrome c release. Biochem Pharmacol. 2003; 66(5): 725-731. Ong ES, Woo SO, and Yong YL. Pressurized liquid extraction of berberine and aristolochic acids in medicinal plants. Journal of Chromatography A. 2000; 313: 57-64. Pharmacopoeia of the People’s Republic of China, Vol One. Beijing: Chemical Industry Press; 2000. 128 Poole CF. New trends in solid-phase extraction. Trends in Analytical Chemistry. 2003; 22(6): 362-373. Pyo MK, Park KM and Yun-choi HS. Isolation and anti-thrombotic activity of citric acid 1,5-dimethyl ester from Gastrodia elata. Natural Product Science. 2000. 6(2): 53-55. Shen YC, Chou CJ, Chiou WF and Chen CF. Anti-inflammatory effects of the partially purified extract of radix Stephaniae tetrandrae: comparative studies of its active principles tetrandrine and fangchinoline on human polymorphonuclear leukocyte functions. Molecular Pharmacology. 2001; 60(5): 1083-1090. Simpson NJK. Solid-phase extraction: principles, techniques, and applications. New York: Marcel Dekker, 2000. Singapore: Traditional Chinese Medicine – A Report by the Committee on Traditional Chinese Medicine. Singapore: Ministry of Health, 1995. Snyder LR, Kirkland JJ, Glach JL. Practical HPLC method development. New York: John Wiley & Sons, 1997. Taguchi H, Yosioka I, Yamasaki K, Kim I H. Studies on the constituents of Gastrodia elata Blume. Chem. Pharm. Bull. 1981; 29(1): 55-62. Tang W and Eisenbrand G. Chinese drugs of plant origin: chemistry, pharmacology, and use in traditional and modern medicine. New York: Springer-Verlag, 1992. Tsutsumi T, Kobayashi S, Liu YY and Kontani H. Anti-hyperglycemic effect of fangchinoline isolated from Stephania tetrandra Radix in streptozotocin-diabetic mice. Biol Pharm Bull. 2003; 26(3): 313-317. 129 Thompson RC and Morris S. Striving to validate nutraceuticals. LCGC. 2001; 19(11): 1142-1149 Tyler V. The new age of herbals: a pharmacognosy renaissance J APhA. 1999 (39): 1112. Vanherwegnem JL, Depierreux M, Tielemans C, Abramowicz D, Dratwa W. Rapidly progressive interstitial renal fibrosis in young women: association with slimming regimen including Chinese herbs. Lancet. 1993; 341: 387-391. Wagner H and Bladt S. Plant Drug Analysis: a Thin Layer Chromatography Atlas. Berlin: Springer-Verlag, 1996. Wang D, Yang G, Li B, Li Z, Chen Y. Investigation of the chemico-physical characteristics of the active components in the Chinese herb Gastrodia elata Bl. by capillary zone electrophoresis. Anal. Sci. 2002; 18(4): 409-412. Wang XH, An R, Zou Y, Wang ZH, Hong XK and Li GW. Application of chromatograph-digitized fingerprint spectrum in identification of medicinal materials and qualitative control of SHUANGHUANGLIAN PREPARATION. Chinese Traditional and Herbal Drug. 2002; 33(2): 115-118. Wang XL, Xiao HB, Liang XM and Bi KS. Vectorial angle method for evaluating the similarity between two chromatographic fingerprints of Chinese herb. Acta Pharmaceutica Sinica. 2002(a); 37(9): 713-717. Waters Corporation. Sample preparation methods development: solid phase extraction. 2004. 130 Wei Y, Ding M Y, Li H X. Determination of gastrodin in Gastrodia elata Blume by HPLC with ELSD and DAD. Gaodeng Xuexiao Huaxue Xuebao. 2001; 22(4): 563-565. Weinsberg F, Bickmeyer U and Wiegand H. Effects of tetrandrine on calcium channel currents of bovine chromaffin cells. Neuropharmacology. 1994; 33(7): 885-890. Wolfender JL, Ndjoko K and Hostettmann K. Liquid chromatography with ultraviolet absorbance-mass spectrometric detection and with nuclear magnetic resonance spectroscopy: a powerful combination for the on-line structural investigation of plant metabolites. Journal of Chromatography A. 2003; 1000 (1-2): 437-455. World Health Organization. Guidelines for the assessment of herbal medicines. Munich, 28.6.1991, World Health Organization, Geneva, 1991. World Health Organization. Good manufacturing practice (GMP): supplementary guidelines for the manufacture of herbal medicinal products, WHO technical report series. No. 863, 1996. World Health Organization, Geneva, 1996. World Health Organization. Guidelines for the appropriate use of herbal medicines. World Health Organization, Regional Publications, Western Pacific Series No.23. World Health Organization Regional Office for the Western Pacific, Manila, 1998. World Health Organization. Regulatory situation of herbal medicines. A worldwide review. World Health Organization, 1998a. World Health Organization. Traditional and Modern Medicine – Harmonizing the Two Approaches. World Health Organization, Western Pacific Region, Manila, 2000. World Health Organization. Regional Strategy for Traditional Medicine in the Western Pacific. World Health Organization, Western Pacific Region, Manila, 2002 131 World Health Organization. Traditional Medicine Strategy 2002-2005. World Health Organization, Geneva, 2002a. World Health Organization. WHO guidelines on good agricultural and collection practices [GACP] for medicinal plants. World Health Organization, Geneva, 2003. Yang JJ, Long H, Liu HW, Huang AJ, and Sun YL. Analysis of tetrandrine and fangchinoline in traditional Chinese medicines by capillary electrophoresis. Journal of Chromatography A. 1998; 811: 274-279. Yoo SM, Oh SH, Lee SJ, Lee BW, Ko WG, Moon CK and Lee BH. Inhibition of proliferation and induction of apoptosis by tetrandrine in HepG2 cells. J Ethnopharmacol. 2002; 81(2): 225-229. Yu XC, Wu S, Wang GY, Shan J, Wong TM, Chen CF and Pang KT. Cardiac effects of the extract and active components of radix stephaniae tetrandrae. II. Myocardial infarct, arrhythmias, coronary arterial flow and heart rate in the isolated perfused rat heart. Life Sci. 2001; 68(25):2863-2872. Zeng Z, Yang DH, Song LF, Yang T, Liu XX, Yuan M and Zeng HP. Studies on the application of high performance liquid chromatographic fingerprint to Rehmannia glutinosa Libosch. Fenxi Huaxue. 2003; 31(12): 1485-1488. Zeng Z, Yang DH, Song LF, Yang T, Liu XX, Yuan M and Zeng HP. Identification of Isatis indigotica Fort. by high performance liquid chromatographic fingerprint. Fenxi Huaxue. 2002; 30(7): 849-852. Zhang G and Liu H. Assay of gastrodin in Gastrodia elata. Zhongcaoyao. 1983; 14(8): 353-355. 132 Zhang X, Cui Z, Wang D and Zhou HY. High performance of liquid chromatographic fingerprint evaluation of the quinolizidine alkaloids from commercial Radix Sophorae Flavescentis. Journal of Asian natural products research. 2003; 5(3): 171-182. Zhao MB, Deng XL, Wang YL, Lu M and Tu PF. Establishment of chromatographic fingerprint and quality assessment of Carthamus tinctorius L. by high performance liquid chromatography. Acta pharmaceutica Sinica. 2004; 39(3): 212-216. Zhao Y, Cao Q, Xiang Y, Hu Z. Identification and determination of active components in Gastrodia elata Bl. by capillary electrophoresis. J. Chromatogr., A. 1999; 849(1): 277- 283. Zhou J, Yang YB, Yang CR. A new phenolic glucoside of Gastrodia elata Blume gastrodin. Kexue Tongbao. 1979; 24(7): 335-336. Zhou J, Yang YB, Yang TR. Chemistry of Gastrodia elata Bl. I. Isolation and identification of chemical constituents of Gastrodia elata Bl. Huaxue Xuebao; 1979a; 37(3): 183-189. Zhou J, Yang YB, Pu XY. Phenolic constituents of fresh Gastrodia elata Blume. Yunan Zhiwu Yanjiu 1980; 2(3): 370. Zhou SF, Koh HL, Gao YH, Gong ZY, Lee EJD. Herbal bioactivation: The good, the bad and the ugly. Life Science. 2004; 74: 935-968. Zhu YP. Chinese materia medica: chemistry, pharmacology and applications. Australia; Amsterdam: Harwood Academic, 1998. 133 [...]... chromatograms of (a) reference standard of AAI and (b) reference herb of Guangfangji, and (c) Overlaid UV spectra of peak 1 (continuous line) and that of AAI (broken line) The RT and UV spectrum of peak 1 in Guangfangji match those of AAI 94 Figure 4.3 HPLC chromatograms of (a) reference standard of tetrandrine, (b) reference standard of fangchinoline, and (c) Fangji from Tongrentang, and UV spectra match of. .. amounts of ephedrine To guarantee batch-to-batch reproducibility of plant material and herbal products, standardization of herbal medicine is necessary Active components of herbal medicines are often used as markers for quality control of herbal medicine (Thompson and Morris, 2001) 1.3 Analytical methods for quality control of herbal medicines Quality control directly impacts the safety and efficacy of herbal. .. herbs which are commonly used in Singapore; and 3) develop HPLC-DAD and LC-MS/MS methods for differentiation of some Chinese herbs 4) apply the developed methods to the analysis of selected herbal medicines and their products 9 Chapter 2 Analysis of Gastrodia elata by HPLCDAD 2.1 Introduction Rhizoma Gastrodiae, Tianma （天麻）, is the dried tuber of Gastrodia elata Blume (Orchidaceae) The tuber is collected... Overlaid chromatograms of extracts of Banlangen from (a) EYS, (b) WYN and (c) SINC 79 Figure 3.9 Chromatograms of extracts of DuZhong from (a) EYS, (b) WYN and (c) SINC 80 Figure 3.10 Overlaid chromatograms of extracts of Shudihuang from (a) EYS, 82 (b) WYN, and (c) SINC, and Shengdihuang from (d) EYS, (e) WYN and (f) SINC Figure 4.1 Chemical structures of tetrandrine, fangchinoline and aristolochic acid... 4hydroxybenzaldehyde and (d) L-pyroglutamic acid Chromatograms of (a) gastrodin (GA), (b) 4-hydroxybenzyl alcohol (HA), (c) 4-hydroxybenzaldehyde (HD) and (d) extract of Gastrodia elata at 270nm Figure 2.9 H-NMR spectrum of gastrodin C-NMR spectrum of gastrodin 28 36 37 Figure 2.10 Chromatograms of (a) L-pyroglutamic acid (PGA) and (b) extract of Gastrodia elata at 195 nm Figure 2.11 Chromatograms of (a) gastrodin and. .. samples resemble like G elata, their ingredients are quite different from those of G elata Therefore chemical profiling of G elata would be useful for discerning it from its fakes and for quality control of this herb 19 2.2 Preparation of gastrodin by HPLC 2.2.1 Introduction The reference standards of herbal ingredients are often unavailable commercially Even when available, the cost is often prohibitive,... and (b) [M+H]+ m/z 623 for tetrandrine and [M+H]+ m/z 609 for fangchinoline 103 Figure 4.9 Effect of tuber lens offset on abundance of (a) [M+NH4]+ for AAI, and (b) [M+H]+ for tetrandrine and fangchinoline 104 Figure 4.10 (a) Full scan mode MS spectrum of AAI; (b) MS/MS spectrum of AAI, with precursor ion, [M+NH4]+ m/z 359 106 Figure 4.11 Full scan mode MS spectra of (a) fangchinoline and (b) tetrandrine... reference standard of fangchinoline, (e) peak 2 with reference standard of tetrandrine 95 Figure 4.4 Overlaid chromatograms of reference Guangfangji, Guangfangji from Guangzhou and 10 local Fangji samples (Table 4.1) 98 XI Figure 4.6 (a) Overlaid chromatograms of the extracts of local Fangji sample from SINC and that of the reference Guangfangji (b) Overlaid UV spectra of peak at 35.7 min of reference... impacts the safety and efficacy of herbal medicines (WHO, 2003) The entire process of product of herbal medicines, from raw materials to finished herbal products, need to be controlled World Health Organization has developed a series of technical guidelines relating to the quality control of herbal medicines These include: Guidelines for the assessment of herbal medicines (WHO, 1991), Good manufacturing... differentiation of botanicals (Chapter 3) For Stephania tetrandra and Aristolochia fangchi herbs, a combination of HPLC-DAD and LC-MS/MS methods will be developed in this study (Chapter 4) 8 Thence, the specific objectives of this study are to 1) develop HPLC methods for the chemical analysis, and identification of G elata; 2) develop HPLC methods for chromatographic fingerprinting analysis of some Chinese ... identification of marker compounds 67 3.2.5.1 UV library of chemical standards 67 3.2.5.2 RT and RRT of chemical standards 67 3.2.5.3 Peak identification 67 3.3 Results and discussion 70 3.3.1 Choice of. .. structures of tetrandrine, fangchinoline and aristolochic acid I 88 Figure 4.2 HPLC chromatograms of (a) reference standard of AAI and (b) reference herb of Guangfangji, and (c) Overlaid UV spectra of. .. line) and that of AAI (broken line) The RT and UV spectrum of peak in Guangfangji match those of AAI 94 Figure 4.3 HPLC chromatograms of (a) reference standard of tetrandrine, (b) reference standard