Fingerprint analysis of Ligusticum chuanxiong using hydrophilic interaction chromatography and reversed-phase liquid chromatography

Journal of Chromatography A, 1216 (2009) 2136–2141

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Fingerprint analysis of Ligusticum chuanxiong using hydrophilic interaction chromatography and reversed-phase liquid chromatography Yu Jin a , Tu Liang a , Qing Fu a , Yuan-Sheng Xiao a , Jia-Tao Feng a , Yan-Xiong Ke b , Xin-Miao Liang a,b,∗ a b

Dalian Institute of Chemical Physics, Graduate School of the Chinese Academy of Sciences, Chinese Academy of Sciences, 457 Zhongshan Road, Dalian 116023, China School of Pharmacy, East China University of Science and Technology, 130 Meilong Road, Shanghai 200237, China

a r t i c l e

i n f o

Article history: Available online 11 April 2008 Keywords: Fingerprint analysis Quality control Ligusticum chuanxiong Hydrophilic interaction chromatography (HILIC) Reversed-phase liquid chromatography (RPLC) Polar compounds

a b s t r a c t Fingerprint analysis is considered one of the most powerful approaches to quality control in traditional Chinese medicines (TCMs). In this study, a binary chromatographic fingerprint analysis was developed using hydrophilic interaction chromatography (HILIC) and reversed-phase liquid chromatography (RPLC) to gain more chemical information about polar compounds and weakly polar compounds. This method was used to construct a chromatographic fingerprint of Ligusticum chuanxiong. The two chromatographic methods demonstrated good precision, reproducibility, and stability, with relative standard deviations of <2% for retention time and 7% for peak area for both HILIC and RPLC separations. Data from the analysis of 14 samples by HILIC and RPLC were processed with similarity analysis, with correlation coefficients and congruence coefficients. This binary fingerprint analysis, using two chromatographic modes, is a powerful tool for characterizing the quality of samples, and can be used for the comprehensive quality control of TCMs. © 2008 Elsevier B.V. All rights reserved.

1. Introduction Traditional Chinese medicines (TCMs) have been widely used in China for thousands of years. At present, the broad application of TCMs all over the world has allowed its incorporation into the pharmaceutical industry. Quality control, as one of the necessary components of TCMs research, is important in guaranteeing the safety, efficacy, and stability of a product. TCMs is a “black box” system, including numerous unknown compounds that vary greatly in their content and their physical and chemical properties. The complexity of TCMs presents a great challenge for quality control. Fingerprint analysis is now considered an effective method for controlling the quality of TCMs [1–4]. In contrast to other methods, it emphasizes the description of the total characteristics of TCMs, which is appropriate for the characteristics of a “black box” system. In recent years, many techniques, especially chromatographic techniques, have been widely used in fingerprint analysis. The current developments in fingerprint analysis endeavor to produce more information from complex samples. Hence, multiple detection methods have been coupled to multiple chromatographic separation techniques to develop a combined fingerprint analy-

∗ Corresponding author at: Dalian Institute of Chemical Physics, Chinese Academy of Sciences, 457 Zhongshan Road, Dalian 116023, China. Tel.: +86 411 84379519; fax: +86 411 84379539. E-mail address: [email protected] (X.-M. Liang). 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.04.010

sis. Some fingerprint analysis methods have been developed using photodiode array detection (DAD) [5,6], evaporation light-scatter detection [7,8], and mass spectrometry [9–11]. Multiple chromatographic fingerprints have also been developed that represent the total chemical characteristics of an analyte [12–14]. TCMs are always prepared with water, so many polar compounds are included in the decoction. Reversed-phase liquid chromatography (RPLC) using an ODS column, the most popular technique used in the fingerprinting analysis of TCMs, can achieve high-efficiency separation for moderately or weakly polar compounds, but it is unsuitable for the fingerprint analysis of polar compounds because of its poor retention of them. When a waterextracted fraction is analyzed on an ODS column, several peaks are often eluted close to the dead time, which attracts our interest. In recent years, hydrophilic interaction chromatography (HILIC), suggested by Alpert in 1990, has gained increasing attention because of its good retention of highly hydrophilic compounds [15]. HILIC is an alternative high-performance liquid chromatography (HPLC) mode for the separation of polar compounds with polar stationary phases, as in normal-phase liquid chromatography (NPLC), but with aqueous mobile phases, similar to those used in RPLC mode. In HILIC mode, water is the strongly eluting solvent and its volume fraction is usually 5–40%, which is just contrary to that in RPLC mode. The use of water as the strongly eluting solvent gives HILIC a number of advantages over conventional NPLC, e.g., eluent preparation is less complicated because total control over the solvent water content is unnecessary, and polar compounds are more solu-

Y. Jin et al. / J. Chromatogr. A 1216 (2009) 2136–2141

ble in the mobile phase of HILIC than in that of NPLC [16]. HILIC has been widely used in the analysis of highly polar compounds of plant origin [17] and from foods [18], in the metabonomic study of urine [19], and in the identification of pharmaceuticals and impurities [20]. In this study, HILIC was used to characterize the smallmolecule polar compounds in a TCM. Chuanxiong, the dried rhizome of Ligusticum chuanxiong Hort. (Umbelliferae), is a popular medicine for the treatment of angina pectoris, cardiac arrhythmias, hypertension, and stroke because of its reputation for facilitating blood circulation and dispersing blood stasis [21]. The main components in L. chuanxiong include alkylphthalides (e.g., Z-ligustilide, senkyunolide A, I, H), phthalide dimers (e.g., levistolide A, tokinolide B), and organic acids (e.g., ferulic acid, caffeic acid) [22,23]. The quality control of L. chuanxiong has attracted increasing attention with its increasing applications, so the fingerprint of L. chuanxiong has been studied with many chromatographic methods and most fingerprint analyses have been based on RPLC [24,25]. Fingerprint analysis should encompass a representative “fingerprint” that contains the greatest amount of information possible. In this study, the water-extracted fraction of L. chuanxiong was divided into two parts. The fingerprint of the polar compounds was analyzed with HILIC, whereas the fingerprint of the moderately and weakly polar compounds was analyzed with RPLC. The developed model using both HILIC and RPLC provided more information about the fingerprint, indicating that HILIC is a suitable method for increasing fingerprint coverage in conventional RPLC fingerprint studies. By using a combination of these two chromatographic fingerprinting methods, a more comprehensive quality control of L. chuanxiong, or TCMs in general, is possible. 2. Experimental 2.1. Chemicals, reagents, and samples HPLC-grade acetonitrile (Merck, Darmstadt, Germany), methanol (Yuwang Co., Shandong Province, China), and formic acid (Acros Organics, USA) were used for the RPLC and HILIC analyses. Deionised water was purified with a Milli-Q system (Millipore, Bedford, MA, USA). In total, 12 batches of L. chuanxiong samples were collected. Eight batches of samples were collected from Dujiangyan region in Sichuan Province, which is a genuine production area for L. chuanxiong. These samples were numbered 1–6 and samples 11 and 12. Samples 7 and 8 and samples 9 and 10 were purchased from Peng region and Guan region in Sichuan Province, respectively. Samples 13 and 14 were from the stems of L. chuanxiong (Chinese name: Chuanxiong lingzi). Samples 1–10 were used to construct the reference chromatogram (common pattern) and samples 11–14 were used to validate the fingerprint analysis method developed with the two separation modes. 2.2. Apparatus All chromatographic studies were performed with an Agilent HPLC 1100 Series instrument (Agilent Technologies, Waldbronn, Germany) consisting of a quaternary pump (model G1311A), an autosampler (model G1313A), a degasser (model G1379A), a photodiode array detector (model G1315B), an automatic thermostatic column compartment (1316A), and a computer with Chemstation software (Agilent, v. 10.02). 2.3. Preparation of sample solutions The dried samples were ground into powder. Accurately weighed samples of powder (10 g) were extracted twice for 2

2137

and 1 h with 100 mL water each time. The decoctions were combined and concentrated to 100 mL, and 5 mL sample solutions were removed for pretreatment. Acetonitrile was added to the 5 mL sample solution slowly until the acetonitrile content reached 95%. After standing for 24 h at 4 ◦ C, the supernatant was filtered and concentrated to 5 mL to produce sample solution I for RPLC analysis. This deposition was extracted with 70% methanol, and the solution was concentrated to dryness and redissolved with 1 mL of 50% acetonitrile. This was sample solution II for HILIC analysis. 2.4. Chromatographic conditions 2.4.1. RPLC analysis The analysis was carried out on a Hypersil ODS2 column (250 mm × 4.6 mm i.d., 5 ␮m). Chromatographic separation was performed with a gradient elution of (A) acetonitrile and (B) 0.1% formic acid in water, as follows: 0–30 min, A 5–30%, B 95–70%; 30–50 min, A 30–70%, B 70–30%. The flow rate was set at 1.0 mL/min, the column temperature was 30 ◦ C, injection volume was 10 ␮L, the wavelength of DAD ranged from 200 to 400 nm, and the chromatograms were recorded at 260 nm. The chromatograms of sample solutions I were used to develop Fingerprint I. 2.4.2. HILIC analysis HILIC was performed on a Venusil HILIC column (250 mm × 4.6 mm i.d., 5 ␮m; Agela Technologies). The mobile phase consisted of (A) water containing 0.1% formic acid, (B) methanol, and (C) acetonitrile. The concentration of methanol (B%) was maintained at 5% for the entire analysis. The linear gradient started from 0% A and reached 12% A at 20 min, then reached 35% A at 40 min, and was maintained at this concentration for 10 min. The flow rate was 1.0 mL/min and the column temperature was 30 ◦ C. The injection volume was 8 ␮L. The wavelength of DAD ranged from 200 to 400 nm and the chromatograms were recorded at 260 nm. The chromatograms of sample solutions II were used to develop Fingerprint II. 2.5. Data analysis of chromatograms Data were analyzed with the professional software Computeraided Similarity Evaluation (CASE), which was developed based on chemometrics by the Research Center for the Modernization of Traditional Chinese Medicines (Central South University, Changsha, China) [26]. With CASE software, the chromatogram can be normalized, and the identical peaks in each chromatogram can be matched in automatic or manual mode. The software constructs a reference chromatogram with median or average data. The total chromatograms for the samples can be included in the evaluation of similarity, and similarity can be evaluated with the correlation coefficient (r1 ) and congruence coefficient (r2 ) calculated with median or average data. The correlation coefficient and congruence coefficient are expressed by Eqs. (1) and (2), respectively: r1 =

n ¯ (x − x¯ )(yi − y) i=1 i  n 2 n (x − x¯ ) i=1 i

¯ (y − y) i=1 i

2

(i = 1, 2, 3, . . . , n)

(1)

n r2 =

 n i=1

xy i=1 i i

(xi )2

n i=1

(i = 1, 2, 3, . . . , n)

(2)

(yi )2

where xi and yi are the ith elements in the two different chromatograms (namely, x and y, respectively), and n is the number of the elements in the chromatograms. x¯ and y¯ are the mean values

2138

Y. Jin et al. / J. Chromatogr. A 1216 (2009) 2136–2141

of the n elements in chromatograms x and y, respectively, i.e.,

n

x¯ =

x i=1 i n

(3)

n y¯ =

y i=1 i n

tion II in HILIC mode provided much chemical information (Fig. 1C), which confirmed that the peaks that eluted close to the dead time in RPLC mode constituted a complex system. The fingerprint analysis developed in HILIC mode was an effective tool for the quality control of polar compounds.

(4)

When r1 and r2 are calculated with median values, x¯ and y¯ are substituted with the median values for n elements in chromatograms x and y, respectively. As r approaches 1, the two chromatograms become more similar. 3. Results and discussion 3.1. Optimization of the pretreatment method In this study, a simple method was developed to isolate the polar compounds from the water-extracted fraction of L. chuanxiong. The sample was divided into two parts, sample solution I containing the moderately and weakly polar compounds, and sample solution II, containing the polar compounds. Comparing the chromatograms of sample solutions I and II of sample 6 in RPLC mode (Fig. 1A and B, respectively), it is clear that polar compounds were isolated into solution II. The peaks eluted close to the dead time in RPLC mode are the main peaks in Fig. 1B. The macromolecular polar compounds, such as polysaccharides and polypeptides, were not the analytes of interest, and would affect the separation on HILIC. So, the deposition was redissolved in 70% methanol, which has low solubility for macromolecular polar compounds. The analysis of sample solu-

3.2. Optimization of the separation conditions of HILIC The complexity of polar compounds is a great challenge in their separation. The composition of the mobile phase in HILIC is usually fixed as water and acetonitrile, so the optimization of the mobile phase mainly involves the selection of the pH and the buffer. The pH greatly affects the separation of such “blind” samples, and better retention and separation were achieved under acidic conditions. The buffers commonly used in HILIC are ammonium salts, such as ammonium acetate and ammonium bicarbonate. The separation of sample solution II with three kinds of mobile phase (10 mmol/L ammonium bicarbonate containing 0.1% formic acid; 10 mmol/L ammonium acetate containing 0.1% formic acid; 0.1% formic acid) was compared on a Venusil HILIC column. Sample solution II could be separated with all three mobile phases. The mobile phase added with buffer achieved better retention of the polar compounds, but also led to serious relative drift in the baseline. It is possible that the impurities in the buffer were enriched and were eluted with the change from a low percentage of strongly eluting solvent to a very high percentage during gradient elution over a wide range, which is necessary for a complex sample. Therefore, only formic acid was added to the mobile phase in these experiments. More and more special separation materials, designed specifically for HILIC, have appeared in recent years, including underivatized silica, amido silica, poly(succinimide)-bonded silica, and polyhydroxyl silica [16]. The separations on an HILIC silica column (underivatized silica) and a Venusil HILIC column (amide-bonded silica) were compared, and the latter displayed the better retention of polar compounds. Using these optimal stationary and mobile phases, a fingerprint analysis method for polar compounds was developed. The selection of a detection wavelength is one of the key factors in a reliable and reproducible fingerprint analysis. This wavelength in the two separation modes ranged from 200 to 400 nm, so that both chromatograms could be extracted at a reasonable wavelength in this range and used for the evaluation of similarity. Analyzing the three-dimensional plots of the chromatograms acquired with the HPLC-DAD system revealed that chromatograms at 260 nm contained the maximum number of peaks detectable in both HILIC and RPLC modes. So chromatograms at 260 nm were finally selected to construct the reference chromatograms and calculate similarities. Using the same wavelength in the two separation modes also had the advantage of allowing the separation of polar compounds in the HILIC and RPLC modes to be compared. 3.3. Validation of the method

Fig. 1. The analysis of sample solutions I and II in RPLC and HILIC modes, respectively. Sample solution I, containing the moderately and weakly polar compounds, was analyzed with RPLC (A), and sample solution II, containing the polar compounds, was analyzed with RPLC (B) and HILIC (C). The chromatogram of sample solution I in RPLC mode (A) was designated Fingerprint I, representing the information about the moderately and weakly polar compounds, and the chromatogram of sample solution II in HILIC mode (C) was designated Fingerprint II, representing the information about the polar compounds.

Sample 6 was analyzed with the established separation method to validate the method. The injection precision in both the RPLC and HILIC separations was evaluated by successive analysis of the same sample solution five times, and its reproducibility in both RPLC and HILIC modes was evaluated with five independently prepared sample solutions. The analysis of the same sample solution at different times (0, 2, 4, 6, 8, 12, and 24 h) was used to evaluate the stability of sample solutions I and II within 24 h. The relative standard deviations (RSDs) of the retention times and peak areas of typical peaks in the two separation modes were used to reflect the precision, reproducibility, and sample stability of the methods. In the RPLC chromatograms, 11 peaks were used to validate the RPLC method (Fig. 1A). The injection precision, represented by the

Y. Jin et al. / J. Chromatogr. A 1216 (2009) 2136–2141

2139

Table 1 Analytical method validation results for Fingerprint I (moderately and weakly polar compounds were analyzed by RPLC) Peak no.

1 2 3 4 5 6 7 8 9 10 11

RSD of retention time (%)

RSD of peak area (%)

Precision (n = 5)

Reproducibility (n = 5)

Stability (n = 7)

Precision (n = 5)

Reproducibility (n = 5)

Stability (n = 7)

0.26 0.22 0.11 0.11 0.13 0.09 0.11 0.09 0.06 0.05 0.05

0.11 0.07 0.07 0.05 0.05 0.05 0.04 0.04 0.02 0.02 0.02

0.37 0.35 0.33 0.30 0.27 0.25 0.20 0.16 0.10 0.10 0.08

1.71 0.76 0.20 2.81 0.20 0.40 0.84 1.10 0.33 0.20 0.14

5.41 5.69 5.69 4.05 3.50 3.67 3.77 5.91 3.29 3.91 5.28

4.39 0.82 0.20 3.19 0.35 0.62 0.48 1.06 0.92 0.61 0.50

RSD, was below 0.30% (n = 5) for the retention times and 0.14–2.81% (n = 5) for the peak areas. The reproducibility (RSD) was 0.02–0.11% (n = 5) for the retention times and 3.29–5.91% (n = 5) for the peak areas. Stability was evaluated from 0 to 24 h. The stability (RSD) was 0.08–0.37% (n = 7) for the retention times and 0.20–4.39% (n = 7) for the peak areas, which confirmed that sample solution I was stable for 24 h (Table 1). Ten peaks were used to validate the method in HILIC mode (Fig. 1C). Before analysis by HILIC, sample solution II was analyzed by RPLC to ensure that the peaks predominantly eluted close to the dead time in RPLC mode were isolated to sample solution II (Fig. 1B). The injection precision (RSD) of the HILIC method was below 1.40% (n = 5) for the retention times and below 5.4% (n = 5) for the peak areas. The reproducibility (RSD) was 0.13–1.07% (n = 5) for the retention times and 2.08–6.73% (n = 5) for the peak areas. Sample solution II was stable for 24 h according to the stability experiment (Table 2). 3.4. Fingerprint analysis of L. chuanxiong by RPLC and HILIC The chromatograms of samples 1–10 in RPLC and HILIC modes were used to construct the reference chromatograms for Fingerprints I and II, respectively, with CASE software. The compounds in the peaks that eluted before 3.7 min on the RPLC chromatograms were processed during the pretreatment. Therefore, these peaks were excluded because this part of the sample did not reflect the information in the sample. Moreover, these peaks, which included much co-eluting information, made no effective contribution to the evaluation of similarity. When the chromatograms of the samples were first input to the software, the peak areas of the chromatograms were normalized to eliminate injection error. The identical peaks in every chromatogram were matched after correction for the drift in retention times. This software can output the reference chromatogram, including peaks common to every chromatogram, based on median or average data. The reference chromatograms of Fingerprints I and II derived with CASE software are shown in Fig. 2. The chromatograms of 14 samples in HILIC and RPLC modes were compared with the corresponding reference chromatogram, and their similarity was evaluated with correlation coefficients and congruence coefficients, which were calculated from median and average data. The similarity values for 14 samples in Fingerprints I and II are listed in Table 3. To compare the quality of the 14 samples, the trend in the change in similarity among the 14 samples, calculated with four different methods (correlation coefficients and congruence coefficients with median and average data) is described in Fig. 3. The four evaluation methods showed identical trends in the change in the respective fingerprints of the 14 samples. Evaluation of the similarity of Fingerprints I and II can produce inconsistent results, e.g., the similarity value for sample 3 was relatively lower

Fig. 2. The similarity of the fingerprints of 14 samples derived with Computer-aided Similarity Evaluation (CASE) software. Fingerprints I and II of Ligusticum chuanxiong were developed using the RPLC (A) and HILIC (B) modes, respectively. The samples are numbered 1–14 from the bottom to the top. Samples 1–10 were used to construct Fingerprints I and II, and samples 11–14 were used to validate the fingerprint analysis method developed using RPLC and HILIC. Samples 13 and 14 were taken from the stem of L. chuanxiong (Chinese name: Chuanxiong lingzi).

2140

Y. Jin et al. / J. Chromatogr. A 1216 (2009) 2136–2141

Table 2 Analytical method validation results for Fingerprint II (polar compounds were analyzed by HILIC) Peak no.

1 2 3 4 5 6 7 8 9 10

RSD of retention time (%)

RSD of peak area (%)

Precision (n = 5)

Reproducibility (n = 5)

Stability (n = 7)

Precision (n = 5)

Reproducibility (n = 5)

Stability (n = 7)

0.69 1.33 1.35 0.49 0.54 0.40 0.59 0.47 0.25 0.09

0.24 0.97 1.07 0.34 0.72 0.44 0.49 0.45 0.23 0.13

0.60 1.38 1.33 0.56 0.47 0.42 0.60 0.54 0.29 0.14

0.93 4.99 4.15 3.88 3.90 3.05 2.40 3.22 5.37 4.45

3.97 2.73 5.45 2.08 6.73 5.85 2.41 3.64 4.83 6.18

1.33 6.46 2.72 3.78 5.06 5.40 3.79 4.50 4.18 6.01

Table 3 Evaluation of the similarity of 14 samples using Fingerprints I and II Sample no.

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Fingerprint I

Fingerprint II

Correlation coefficient

Congruence coefficient

Correlation coefficient

Congruence coefficient

Median

Average

Median

Average

Median

Average

Median

Average

0.9345 0.9892 0.9529 0.9784 0.9810 0.9952 0.9924 0.9915 0.9912 0.9752 0.8978 0.9555 0.7865 0.7329

0.9420 0.9895 0.9582 0.9787 0.9849 0.9936 0.9938 0.9937 0.9929 0.9749 0.9060 0.9551 0.7985 0.7462

0.9445 0.9891 0.8843 0.9782 0.9825 0.9959 0.9915 0.9928 0.9918 0.9790 0.9128 0.9590 0.8165 0.7782

0.9513 0.9887 0.8444 0.9777 0.9864 0.9944 0.9933 0.9945 0.9936 0.9788 0.9201 0.9589 0.8285 0.7911

0.9234 0.9551 0.9190 0.9087 0.9271 0.8914 0.9689 0.9752 0.9733 0.9280 0.8034 0.8665 0.8833 0.8747

0.9411 0.9425 0.9387 0.9304 0.9338 0.9155 0.9569 0.9654 0.9616 0.9307 0.7862 0.8522 0.9003 0.8866

0.9622 0.9764 0.9604 0.9393 0.9589 0.9436 0.9843 0.9879 0.9855 0.9650 0.8889 0.9350 0.9356 0.9352

0.9709 0.9700 0.9703 0.9514 0.9624 0.9560 0.9784 0.9831 0.9795 0.9664 0.8796 0.9280 0.9446 0.9412

Note: Samples 1–10 were used to construct Fingerprints I and II, and samples 11–14 were used to validate the fingerprint analysis method developed using HILIC and RPLC. Samples 13 and 14 were from the stem of Ligusticum chuanxiong (Chinese name: Chuanxiong lingzi). The correlation coefficients and congruence coefficients were calculated by comparing the original chromatogram with the median and average reference chromatograms.

on Fingerprint I and higher on Fingerprint II, whereas sample 6 had a higher similarity value on Fingerprint I but a lower similarity value on Fingerprint II. The uncertainty of the concentration distribution in individual samples is a normal phenomenon in the analysis of natural materials. To address this problem, fingerprint analysis using HILIC was developed to obtain more information about a sample, and the two separation modes used together contribute to the characterization of the sample and the quality control of L. chuanxiong.

Samples 11–14 were not included in the construction of the reference chromatograms, but were used to validate the fingerprint analysis method developed with the two separation modes. Samples 11 and 12 came from the same production area as samples 1–6. Evaluation of their similarity with Fingerprints I and II showed that the quality of sample 11 was lower than that of the other samples. The quality evaluation of sample 12 resulted in a higher similarity value for Fingerprint I in RPLC mode, but a lower similarity value for Fingerprint II with HILIC separation. Hence, the fingerprint analysis of polar compounds using HILIC coupled to the fingerprint analysis by RPLC achieved strict quality control of L. chuanxiong. Chuanxiong lingzi (samples 13 and 14) and L. chuanxiong are different parts of the same herbal medicine. The similarity value for Chuanxiong lingzi was low in both fingerprints, especially in Fingerprint I. So fingerprint analysis of L. chuanxiong using RPLC and HILIC could prevent the arbitrary abuse of herbal material. The area of production of L. chuanxiong is relatively localized, mainly in Sichuan Province, and there was no obvious difference in the quality of samples from several areas in Sichuan Province, e.g., samples from Peng region (samples 7 and 8) and Guan region (samples 9 and 10) showed quality consistent with that of samples from a genuine production area (Dujiangyan region). The stability of sample quality must be strictly controlled, and the fingerprint analysis of polar compounds must be considered to this end. 4. Conclusions

Fig. 3. The trend in the change in similarity in Fingerprints I and II for 14 samples.

Polar compounds eluting close to the dead time in the RPLC mode often exhibit better retention and separation in the HILIC

Y. Jin et al. / J. Chromatogr. A 1216 (2009) 2136–2141

mode. So a fingerprint analysis of these polar compounds was developed using HILIC, to produce chemical information that could be overlooked in conventional RPLC analysis. Combining the two separation modes allows the production of a representative “fingerprint” that contains the greatest amount of information possible. With the analysis of the polar compounds of L. chuanxiong in HILIC mode, more chemical information has become available. The quality control of L. chuanxiong using these two chromatographic fingerprinting methods illustrates the fact that some samples that display good quality when evaluated by conventional RPLC separation contain polar compounds that differ from those in normal samples. So the two chromatographic fingerprints can better present the chemical characteristics of L. chuanxiong and offer more effective quality control of this TCM. We have made a significant attempt to develop a more comprehensive fingerprinting method by combining HILIC and RPLC. However, the application of HILIC to the analysis of TCMs is relatively rare. The use of HILIC for more TCMs will require the optimization of the HILIC separation conditions, especially in the sample pretreatment and the selection of the hydrophilic stationary phase. More effort must also be directed to the development of reasonable chemometric methods to evaluate the two fingerprint analyses. Polar compounds exist widely in TCMs, and their quality control by HILIC analysis, an effective method of separating them, can be applied to other TCMs. With further research into the polar compounds of TCMs, including the preparation of some polar compounds and the screening of their pharmaceutical activities, the quality control of polar compounds will require greater attention and will become an important part of the quality control of TCMs. Acknowledgments This work was supported by the Knowledge Innovation Program of DICP, CAS (K2006A3 and K2006A4).

2141

The authors gratefully acknowledge the software support of Professor Yi-Zeng Liang in Central South University, Changsha, China. References [1] P.S. Xie, S.B. Chen, Y.Z. Liang, X.H. Wang, R.T. Tian, R. Upton, J. Chromatogr. A 1112 (2006) 171. [2] D.K.W. Mok, F.T. Chau, Chemometr. Intell. Lab. Syst. 82 (2006) 210. [3] Y.Z. Liang, P.S. Xie, K. Chan, J. Chromatogr. B 812 (2004) 53. [4] P. Drasar, J. Moravcova, J. Chromatogr. B 812 (2004) 3. [5] S.K. Yan, W.F. Xin, G.A. Luo, Y.M. Wang, Y.Y. Cheng, J. Chromatogr. A 1090 (2005) 90. [6] L.W. Yang, D.H. Wu, X. Tang, W. Peng, X.R. Wang, Y. Ma, W.W. Su, J. Chromatogr. A 1070 (2005) 35. [7] J.B. Wan, S.P. Li, J.M. Chen, Y.T. Wang, J. Sep. Sci. 30 (2007) 825. [8] S.K. Yan, W.D. Zhang, R.H. Liu, Y.C. Zhan, Chem. Pharm. Bull. 54 (2006) 1058. [9] M. Cai, Y. Zhou, S.L. Gesang, C. Bianba, L.S. Ding, J. Chromatogr. B 844 (2006) 301. [10] V. Kumar, N. Mehrotra, J. Lal, R.C. Gupta, J. Chromatogr. A 1045 (2004) 145. [11] M. Liu, Y.G. Li, F. Zhang, L. Yang, G.X. Chou, Z.T. Wang, Z.B. Hu, J. Sep. Sci. 30 (2007) 2256. [12] Y.P. Li, Z. Hu, L.C. He, J. Pharm. Biomed. Anal. 43 (2007) 1667. [13] X.H. Fan, Y.Y. Cheng, Z.L. Ye, R.C. Lin, Z.Z. Qian, Anal. Chim. Acta 555 (2006) 217. [14] F.M. Li, B. Yuan, Z.L. Xiong, X.M. Lu, F. Qin, H.S. Chen, Z.G. Liu, Biomed. Chromatogr. 20 (2006) 634. [15] A.J. Alpert, J. Chromatogr. 499 (1990) 177. [16] P. Hemstrom, K. Irgum, J. Sep. Sci. 29 (2006) 1784. [17] V.V. Tolstikov, O. Fiehn, Anal. Biochem. 301 (2002) 298. [18] H. Schlichtherle-Cerny, M. Affolter, C. Cerny, Anal. Chem. 75 (2003) 2349. [19] S. Cubbon, T. Bradbury, J. Wilson, J. Thomas-Oates, Anal. Chem. 79 (2007) 8911. [20] X.D. Wang, W.Y. Li, H.T. Rasmussen, J. Chromatogr. A 1803 (2005) 58. [21] The State Pharmacopoeia Commission of People’s Republic of China, Pharmacopoeia of the People’s Republic of China, vol. 1, Chemical Industry Press, Beijing, 2005, p. 28. [22] T. Yi, K.S.Y. Leung, G.H. Lu, H. Zhang, K. Chan, Chem. Pharm. Bull. 53 (2005) 1480. [23] W.Y. Huang, S.J. Sheu, J. Sep. Sci. 29 (2006) 2616. [24] G.H. Lu, K. Chan, Y.Z. Liang, K. Leung, C.L. Chan, Z.H. Jiang, Z.Z. Zhao, J. Chromatogr. A 1073 (2005) 383. [25] G.X. Xie, A.H. Zhao, P. Li, L. Li, W. Jia, Biomed. Chromatogr. 21 (2007) 867. [26] P.S. Xie, Chromatographic Fingerprint for Traditional Chinese Medicine (TCM), People’s Medical Publishing House, 2005, p. 469.