Sei sulla pagina 1di 9

Food Chemistry 197 (2016) 1112–1120

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Determination and quality evaluation of green tea extracts through


qualitative and quantitative analysis of multi-components by single
marker (QAMS)
Da-Wei Li a, Ming Zhu b,⇑, Yun-Dong Shao c, Zhe Shen a, Chen-Chen Weng c, Wei-Dong Yan a,⇑
a
Department of Chemistry, Zhejiang University, Hangzhou 310027, China
b
Zhejiang Institute for Food and Drug Control, Hangzhou 310051, China
c
Skyherb Ingredients, Anji 313300, China

a r t i c l e i n f o a b s t r a c t

Article history: The quality of tea is mainly attributed to tea polyphenols and caffeine. In this paper, a new strategy for
Received 21 May 2015 quality evaluation of green tea extracts was explored and verified through qualitative and quantitative
Received in revised form 14 October 2015 analysis of multi-components by single marker (QAMS). Taguchi Design was introduced to evaluate
Accepted 19 November 2015
the fluctuations of the relative conversion factors (fx) of tea catechins, gallic acid and caffeine to epigal-
Available online 28 November 2015
locatechin gallate. The regression model (Sig. = 0.000) and the deviations (R2 > 0.999) between QAMS and
normal external standard method proved the consistency of the two methods. Hierarchical cluster anal-
Chemical compounds studied in this article:
ysis and canonical discriminant analysis were employed to classify 26 batches of commercial Longjing
Epigallocatechin gallate (PubChem CID:
65064)
green tea extracts (LJGTEs) collected from different producers. The results showed a significant difference
Epicatechingallate (PubChem CID: 367141) in component profile between the samples from different origins. The QAMS method was verified to be
Epigallocatechin (PubChem CID: 72277) an alternative and promising method to comprehensively and effectively control the quality of LJGTEs
Epicatechin (PubChem CID: 72276) from different origins.
Gallocatechin gallate (PubChem CID: Ó 2015 Elsevier Ltd. All rights reserved.
199472)
Gallocatechin (PubChem CID: 65084)
Catechin gallate (PubChem CID: 6419835)
Catechin (PubChem CID: 73160)
Caffeine (PubChem CID: 2519)
Gallic acid (PubChem CID: 370)

Keywords:
Determination
Quality evaluation
Qualitative and quantitative analysis of
multi-components by single marker (QAMS)
Green tea extracts
Taguchi Design
HPLC

1. Introduction Wang, Lu, & Picinich, 2009; Zhao et al., 2014). They can be used
in the treatment of promoting blood circulation, improving liver
Green tea is one of the most popular and widely consumed bev- function and oral health, eliminating various toxins and inducing
erages all over the world. Phenolic compounds, the rich sources in apoptosis in human lymphoid leukemia cells (Fujiki et al., 1996;
green tea extracts, are potentially beneficial for health, as they act Liao, Umekita, Guo, Kokontis, & Hiipakka, 1995; Stoner &
not only as strong antioxidant agents but also as many drug effica- Mukhtar, 1995).
cies (Jankun, Selman, Swiercz, & Shrzypczak-Jankun, 1997; Yang, As the first rank of top ten teas in China, Longjing green tea has
been widely used in food, nutraceuticals, pharmaceutical products
and other areas. Tea polyphenols (TPs) were the most abundant
⇑ Corresponding authors.
substances in Longjing green tea extracts (LJGTEs). It has been
E-mail addresses: zhumingd@hotmail.com (M. Zhu), yanweidong@zju.edu.cn
(W.-D. Yan). reported that TPs was the primary contributor to affect the quality

http://dx.doi.org/10.1016/j.foodchem.2015.11.101
0308-8146/Ó 2015 Elsevier Ltd. All rights reserved.
D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120 1113

of the tea in the flavor, color, aroma and healthy benefits (Zhao simultaneous determination of multi-components in Coptides
et al., 2014). The catechins accounts for about 60–90% of TPs in Rhizoma (Huanglian) by QAMS had also been included (Chinese
general. The catechins consists of epigallocatechin gallate (EGCG), Pharmacopoeia, 2010). It is yet to be studied that whether the rel-
epicatechingallate (ECG), epigallocatechin (EGC), epicatechin (EC), ative conversion factor (fx) of QAMS in simultaneous determination
gallocatechin gallate (GCG), gallocatechin (GC), catechin gallate of multi-components in LJGTEs could be used as a constant in dif-
(CG) and catechin (C) (Lee, Hwang, Lee, & Choung, 2014; vander ferent laboratories, and what influenced fx (Wang et al., 2003;
Hooft et al., 2012). Caffeine (CAF) and gallic acid (GA) in LJGTEs Zhang et al., 2014). There is also a lack of study on whether QAMS
were both somewhat harmful to human beings. Caffeine can trig- could be proposed to easily and effectively control the quality of
ger palpitations and increase blood pressure in sensitive individu- LJGTEs.
als. Gallic acid belongs to chemical raw materials. Because of the In our work, the new quality control of LJGTEs through QAMS
possible adverse health effects, the content of them in the tea was presented for the simultaneous determination of eight tea cat-
extracts should be restricted. Chemical profiles greatly affect the echins, GA and CAF in LJGTEs. The applicability and feasibility of
quality of green tea, which vary significantly among different ori- QAMS mainly rely on the consistency of the environmental and
gins (Maeda-Yamamoto, Ema, Monobe, Tokuda, & Tachibana, operational parametric variables, such as analysts, HPLC instru-
2012; Serpen et al., 2012; Tian et al., 2012). Health claims need ments, and columns. In this study, the fluctuations and stability
to be backed by scientific study of the components, it is therefore of fx with a focus on different columns, HPLC instruments and ana-
necessary to propose a method that could accurately determine lysts were evaluated and discussed (Goupy, 2005). The linear
the components in LJGTEs. regression model was built between two variables to examine
In addition, the quality and benefits of LJGTEs were not solely deviations of the content results obtained by QAMS and ESM.
determined by the single compound, but rather the combinational Hierarchical cluster analysis (HCA) and canonical discriminant
effects of multiple constituents (Yang, 1997). For the moment, a analysis (CDA), which were based on the contents of eight
large number of methods have been reported for the simultaneous catechins were employed to classify 26 batches of LJGTEs collected
determination of multi-components (SDMC) in herbal or botanical from different producers.
products. High performance liquid chromatography (HPLC) is
mainly used to determine the contents of most organic com-
2. Materials and methods
pounds. Based on conventional HPLC, there were two main meth-
ods for SDMC. In the first method, multiple reference standards
2.1. Chemicals and materials
were used for the analysis of multiple components, as the normal
external standard method (ESM). ESM has been most commonly
Eight catechins reference standards were purchased from
used for the qualitative and quantitative analysis of components
Chengdu Biopurity Phytochemicals Co., Ltd. The structures of these
in herbal or botanical products and indeed has played an important
standards were identified and confirmed through analyses with
role in the fields of research, production regulation and evaluation
UV, MS, 1H NMR and 13C NMR (see S1–S3 in Supporting Informa-
(Jin, Ma, Ma, Yao, & Chen, 2014; Samanidou, Tsagiannidis, &
tion). CAF and GA were provided by National Institutes for Food
Sarakatsianos, 2012; Wang & Helliwell, 2001; Zhang et al., 2013).
and Drug Control (NIFDC), which has Laboratory Accreditation Cer-
However, in the recent years, quality control and evaluation of Tra-
tificate. The purity of these ten compounds was verified to be
ditional Chinese Medicines (TCMs) by ESM has become challenging
greater than 98% with HPLC using the peak area normalization
in some cases due to the high cost of some reference standards, the
method. Methanol for HPLC and other related materials were
diversity and complexity of botanical products, the instability of
obtained from Sinopharm Chemical Reagent Co. Ltd (Shanghai,
detection and many other uncertainties. According to some reports
China). Voucher specimens were stored at Zhejiang Institute for
(Gao, Jiang, Lu, & Tu, 2009; He, Li, Zhao, & Dong, 2012; He, Yang, &
Food and Drug Control. All 26 batches of LJGTEs were obtained
Zhang, 2012; Hou et al., 2011), ESM may have relatively low speci-
from seven major producers (Table 1).
ficity and low efficiency, and to some extent, it was challenging to
determine the chemical profile with ESM, which would make it dif-
ficult to evaluate TCMs comprehensively. It was especially difficult 2.2. Instruments and conditions
to determine more than five components simultaneously in the
routine quality supervision and evaluation. Meanwhile, quality Analyses were primarily performed by a Wufeng HPLC-100
control of botanical products (herb drugs or TCMs) by multi- system composed of a double solvent delivery system, an on-line
components has been acknowledged by both United States Phar- degasser, a column temperature controller, and dual wavelength
macopeia (USP) and Chinese Pharmacopoeia (Ch.P.). UV detector coupled with an analytical workstation (WS-100
In comparison, the second method requires one single reference software) (Wufeng Tech. Shanghai, China). Analytes were separated
standard to simultaneously determine the contents of multi- on a reverse phase C18 column (Thermo ODS hypersil column,
components, which was abbreviated to QAMS (qualitative and 250 mm  4.6 mm  5 lm). Samples with a volume of 20 lL were
quantitative analysis of multi-components by single marker). injected into the column and the column temperature was main-
Qualitative and quantitative analysis of multi-components by sin- tained at 40 °C. The mobile phase was programmed with a linear gra-
gle marker can not only greatly reduce the detection time and the dient, 100% A (900:1:100, v/v; water, phosphoric acid, acetonitrile),
cost of the experiment, but improve the practicability of the 0% B (200:1:800, v/v; water, phosphoric acid, acetonitrile) for
method and control the quality of herbal or botanical products 10 min, from 0% B to 10% B in 20 min, from 10% B to 0% B in 20 min,
more effectively and comprehensively. It has been reported that 0% B for 5 min, at a flow rate of 1.0 mL/min. The system was
QAMS has been applied in the determination of Rhubarb Rhizome maintained at pre-equilibrium for 10 min between consecutive injec-
(Gao et al., 2009), Panax Ginseng (Li et al., 2013), Salvia Miltior- tions. The detection wavelength was set at 280 nm. Reproducibility
rhiza (Hou et al., 2011) and other botanical ingredients (Du, Li, was verified by duplicate analyses. Two additional HPLC instruments
Liu, Yin, & Bi, 2014; He et al., 2012; Wang, Gao, Fu, & Wang, (see S4 in Supporting Information) and five representative additional
2006; Wang, Provan, & Helliwell, 2003). Furthermore, quality con- columns were used (see S5 in Supporting Information).
trol of Cat’s Claw (United States Pharmacopeia, 2010a) and Bilberry UV-TU1950 was from Beijing Purkinje General Instrument
Extract (United States Pharmacopeia, 2010b) by QAMS have both Co., Ltd (Scanning range of 190–700 nm). 1H and 13C NMR spectra
been included in USP 33-NF. In Ch.P. 2010 edition (volume I), were measured on a Bruker 400 MHz spectrometers with chemical
1114 D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120

Table 1
The content of ten compounds in 26 batches of investigated commercial LJGTEs samples (mg/g).

No. Producersa Batches GA GC EGC CAF C EC EGCG GCG ECG CG T1b T2c
s1 ZGP 0301a 17.1 34.5 181.7 24.6 12.9 59.2 449.4 13.2 100.0 10.2 861.1 902.8
s2 ZGP 1101a 16.4 24.5 155.6 21.7 3.2 57.8 441.1 12.8 113.9 8.4 817.3 855.4
s3 ZGP 0301b 10.6 38.5 183.5 23.8 9.6 62.6 476.0 15.6 106.8 10.5 903.1 937.5
s4 ZGP 0302 17.9 38.5 144.5 26.3 20.4 54 461.0 17.6 103.8 7.0 846.8 891
s5 ZGP 1002 10.8 38.7 170.4 26.6 20.8 55.8 505.3 20.5 125.8 7.5 944.8 982.2
s6 ZGP 1101b 13.5 31.6 174.2 24.1 8.1 61.5 475.3 15.7 107.2 7.0 880.6 918.2
s7 XGP 0801 23.0 24.2 128.5 2.8 2.3 51.2 464 16.2 106.9 7.4 800.7 826.5
s8 STP 1201 17.9 44.1 186.2 12.5 4.8 57.5 495.1 30.7 114.6 11.4 944.4 974.8
s9 STP 0401 16.4 37.7 182.4 12.9 3.9 54.6 524.6 25.5 114.1 4.6 947.4 976.7
s10 LWP 0301a 11.2 44.4 223.5 7.5 9.1 67.4 457.8 51.9 102.4 3.7 960.2 978.9
s11 HSP 1201 15.0 38.0 127.7 10.9 9.1 51.8 464.3 26.5 109.4 3.8 830.6 856.5
s12 ZGP 1227 14.3 38.9 112.5 7.1 9.3 57.3 430.1 22.7 99.9 4.4 775.1 796.5
s13 LWP 0101a 15.5 53.1 177.2 7.0 12.1 58.4 409.5 49.3 97.5 3.9 861 883.5
s14 SSP 0401 28.6 29.7 149.9 3.1 6.7 37.4 372.9 15.9 80.1 5.8 698.4 730.1
s15 SSP 0402 30.1 32.7 160.5 3.3 9.0 42.4 380.6 20.2 85.6 2.1 733.1 766.5
s16 SSP 0403 30.3 35.6 157.7 3.1 8.8 41.0 376.8 20.6 83.6 1.5 725.6 759.0
s17 SOP 0102 13.8 59.3 192.9 18.9 12.9 64.3 404.4 24.2 145.6 1.7 905.3 938.0
s18 SSP 0701 11.9 41.9 228 3.0 18.5 44.0 469.4 36.3 97.6 2.8 938.5 953.4
s19 SSP 0301 12.2 33.5 215.3 3.1 15.5 46.2 459.8 27.0 98.6 3.7 899.6 914.9
s20 SSP 1201 12.5 38.4 218.5 2.9 16.3 43.8 445.3 28.9 97 3.1 891.3 906.7
s21 SOP 0904 12.5 60.7 206.5 3.2 19.3 72.6 465.8 29.8 74.8 1.2 930.7 946.4
s22 LWP 0401b 11.0 40.3 226.8 7.4 11 62.3 450.7 51.5 98.5 1.2 942.3 960.7
s23 LWP 0401a 10.6 43.3 229.4 6.9 10.4 65.2 456.3 53.7 104.3 1.2 963.8 981.3
s24 LWP 0301b 16.2 55.8 225.6 7.5 6.7 62.8 418 58.6 103.8 4.9 936.2 959.9
s25 LWP 1101 14.2 56.5 194.3 6.2 5.9 63.9 412 50.2 99.3 4.3 886.4 906.8
s26 LWP 0101b 18.3 55.7 228.8 7.8 7.6 77.3 421.5 57.5 100.7 5.6 954.7 980.8
Minimum 10.6 24.2 112.5 2.8 2.3 37.4 372.9 12.8 74.8 1.2 698.4 730.1
Maximum 30.3 60.7 229.4 26.6 20.8 77.3 524.6 58.6 145.6 11.4 963.8 982.2
Means 16.2 41.2 183.9 10.9 10.5 56.6 445.9 30.5 102.8 5.0 876.1 903.3
Standard deviation 5.8 10.2 34.9 8.5 5.3 9.9 38.2 15.3 14.1 3.0 77.5 74.8
a
LJGTEs were collected from 7 major producers: HSP; SOP; SSP; LWP; ZGP; STP; XGP.
b
The total content of 8 catechins.
c
The total content of 10 compounds.

shifts reported as parts per million (in CD3OD at 20 °C). Mass be abundant in each sample. It should be stable, easily obtainable
spectroscopy data of the compounds was collected on MS-ESI and have a clear pharmacological effect or related to clinical
(Esquire 3000plus) from Bruker Daltonics Inc. efficacy. Moreover, it should be easily separated under the
conventional chromatographic conditions. Epigallocatechin gallate
2.3. Preparation of LJGTEs and sample solutions (EGCG) was the most abundant and active component in LJGTEs
and it was used as a quality indicator in this paper for its low cost,
The preparation of LJGTEs has been included in the national high stability, high content, easy to obtain and significant
drug standard (Prescription of Chinese Medicinal, 1998). Green pharmacological activities in LJGTEs.
tea extracts were prepared by placing 10.0 g powder of Longjing The standard solutions were analyzed under the conditions
tea leaves into 50 mL boiled water and heated (100 °C) for detailed above. Using EGCG as a reference standard, fx could be
50 min. Then the solutions were first filtered, and another 50 mL calculated by the ratio of the peak areas and the ratio of the
boiled water was added for second extraction. Then the saturated concentration between the analyte and EGCG (Eq. (1)). With
limewater was gradually added in the combined filtrate until the results of fx, the quantification of bioactive components (Cx)
pH = 9. Being equilibrated for 10 min, the solid precipitate was fil- in the samples could be carried out according to the following
trated and then the hydrochloric acid solution (37%) was added to equations (Eqs. (2) and (3)):
dissolve the precipitate until pH = 4. Ethyl acetate was added to
extract the tea bioactive components from the solutions. The solu- AEGCG =C EGCG
fx ¼ ð1Þ
tions were evaporated under reduced pressure. Approximately 10– Ax =C x
15 mg tea extracts were dissolved with water/methanol (3:1, v/v)
10 mL, centrifuged at 10,000 rpm for 10 min prior to HPLC analysis. Ax  C EGCG
Cx ¼  fx ð2Þ
AEGCG
2.4. Preparation of standard solutions
Cx  V
Stock solutions were prepared by dissolving approximately xx ¼ ð3Þ
m
10 mg of each reference standard respectively into a 10 mL volu-
metric flask with water/methanol (3:1, v/v) to volume, mixed, where Ax is the peak area of the bioactive component in the single
and then diluted (dilution factor = 5) with water/methanol (3:1, standard solution or in the samples; fx is the average relative
v/v) prior to HPLC analysis. conversion factor of each bioactive component to EGCG; Cx is the
concentration of analyte in the single standard solution or in the
2.5. Calculation of relative conversion factors samples (mg/mL); V is the volume of the green tea extracts
solutions (mL); m is the mass of the green tea extracts (mg); xx is
In order to obtain fx for each analyte, it was essential to select a the mass concentration of the bioactive components in the green
suitable reference analyte. The selected reference analyte should tea extracts (mg/g)).
D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120 1115

The ESM method required that all the reference standards solu- curve, precision, accuracy and specificity were investigated accord-
tions corresponding to analytes would be prepared first. Con- ing to the USP32 <1225 VALIDATION OF COMPENDIAL PROCEDURES>
versely, the QAMS method did not require preparation for all the (United States Pharmacopeia, 2009).
reference standards solutions instead only one standard solution The evaluation results of the calibration model were shown in
was needed. S7 (in Supporting Information). Using Student’s t-test, the intercept
and the regression coefficient were investigated. Both the calibra-
3. Results and discussion tion line and the residuals were graphically inspected and evalu-
ated. The correlation coefficient for all trials was at least 0.999.
3.1. Representative HPLC chromatograms The graphical inspection showed a linear correlation and the high-
est residual of 4% was still lower than the limit of 5%. It could be
In this study, a gradient elution method has been developed for concluded that a linear model was applicable in the concentration
the separation of LJGTEs. In such systems, chromatograms of range as described in S7 (in Supporting Information).
reference standards and the typical sample of LJGTEs were shown The precision experiment mainly included the investigation of
in Fig. 1. A good separation was achieved within 40 min using the the repeatability and the precision. Six independent samples on
chromatographic conditions described. The most intense band was three different days were analyzed by the same analyst. The results
the one for EGCG. In addition, no other chromatographic peaks in S7 obtained on different days were not statistically different.
were observed around the 180 min elution time (0–50 min using Since the RSD values for intra-day precision and inter-day precision
the gradient conditions above, 50–180 min using 100% A) at the were of the same order of magnitude and less than 3%, the method
given wavelength. A detection wavelength of 280 nm was selected could be considered precise.
due to its stability compared to detection at 210 nm. The baseline The accuracy was investigated by performing a recovery exper-
obtained at 210 nm was very unstable and difficult to balance, thus iment. The results of the recovery experiment were shown in S7
it was difficult to achieve baseline separation (see S6 in Supporting (see Supporting Information). The average of the recovery for all
Information). ten components was (97.6–101.5)% with an RSD of below 1.9%. It
was shown that the recovery was not statistically different from
3.2. Validation of HPLC method 100%, which indicated that the method was accurate.
The specificity of the HPLC method was investigated by check-
Validation of the method was required to assure the suitability ing mean retention time and the peak purity of green tea extracts.
for its intended use. Different characteristics such as calibration This was performed by comparing the UV spectrum of the peak in
the reference standard and in the green tea extract. No deviations
were observed during the comparison of the UV spectra of the
peaks in the reference standard and in green tea extracts. The SD
of mean retention time was all less than 3%. The results indicated
that no interference occurred by other constituents in the solution.
The limit of detection (LOD) was determined at a signal-to-
noise ratio of 3:1. The limit of quantification (LOQ) was determined
at a signal-to-noise ratio of 10:1. The data of LOD and LOQ were
listed in S7.

3.3. Robustness test

To apply the determination method in different laboratories,


the analysts, instruments and columns were inevitable to change.
The different types of columns with different carbon load, surface
area and end capping may affect the results of the tailing factor
or symmetry factor, resolution and theoretical plate number. The
different HPLC instruments and different detectors affected the cal-
culation of peak area which mainly related with slit of detector, the
bandwidth value, the slope sensitivity and peak width and ways of
integration. The accidental errors caused by different analysts
acted sometimes in one and sometimes in another direction and
by their very nature can’t be predicted. Thus the potential influ-
ences should be fully considered in order to better control the
related parameters.
Taguchi Design: Taguchi Design also called Robust Design, is a
scientific and effective approach for the robust design of multi-
factor experiment (Peng, 2013; Wang, Sun, & Liu, 2014). In this
test, to evaluate the fluctuations and stability of QAMS, Taguchi
Design – Static design (Nominal is the best) was performed
(Table 2). Given the range of available reversed-phase stationary
phases and HPLC instruments, the representative columns (S5)
and HPLC instruments systems (S4) were selected. Signal-to-
noise (S/N) were introduced as evaluation parameters to consider
Fig. 1. Representative HPLC chromatograms of reference standards (A) and green the influence of the mean and variance together. The target value
tea extract (B). Peak identification: 1. GA. 2. GC. 3. EGC. 4. CAF. 5. C. 6. EC. 7. EGCG. 8.
GCG. 9. ECG. 10. CG. It was obtained using the Thermo ODS hypersil column and
was the true fx of other nine components, the signal factor was
Wufeng HPLC instruments, the similar chromatogram could be obtained using the measured fx of other nine components, six columns, three LC
other columns and instruments. instruments and three analysts were set as the controlling factors,
1116 D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120

and the noise was the fluctuations of the results changing with the Variance analysis: Based on the Intuitive analysis results, the
circumstances which were difficult to control. Whether the results significance level of variance analysis for the column, instrument
were around the target value was the major consideration. From and analyst were given. For the stationary phase factors, the results
the results of residual plots in Fig. 2, it can be seen that all the in S12 (in Supporting Information) clearly showed that the factor
numerical points were presented as linear distribution (Fig. 2A), of column had more important influence than the factor of analyst
which means that the results could be judged as normal distribu- and instrument and the column was the most significant factor
tions. And also its characteristic in Fig. 2B was similar to ‘‘bell” (P < 0.05) especially for the component GA, GCG, ECG and CG.
shaped curve, which means that data collections were reasonable The column had the most significant influence on the results and
and adequate and the process design was stable. The results were should be well controlled.
reliable. From Fig. 2C and D, it can be seen that all the results were Correlation analysis: In order to describe the possible relation-
distributed randomly, which showed that there were no abnormal ship of the fx results of nine components, the correlation analysis
values in the text. was made based on the results in Table 2. From the results in
The results in S8 (in Supporting Information) clearly showed S13 (in Supporting Information), the Pearson Correlation between
that the factor of column (P < 0.05) had more important influence C and EC was 0.870, the Pearson Correlation between ECG and CG
to the mean value. The factor of column and the instrument had was 0.851. This indicated that there is a strong correlation of fx
more important influence (P < 0.05) to S/N. The results were consis- between C and EC, ECG and CG. The results were very consistent
tent with the response results in S9 and the main effects plot in with the very similar compound structure of C and EC, ECG and
S10. For mean value, the column was the most significant factor CG. The Pearson Correlation between GC and EGC was 0.152, but
(Delta = 0.084), the analyst was intermediate (Delta = 0.029), the the Sig. (2-tailed) was 0.546, which indicated that GC and EGC
instrument was least affected (Delta = 0.026). For S/N, the column had some similar relevance. What’s more, the Pearson Correlation
was also the most significant factor (Delta = 0.421), the instrument or the Sig. (2-tailed) was large among different compounds such
was intermediate (Delta = 0.177), the analyst was least affected as GA and C (0.675), GC and EC (0.624), EC and ECG (0.771). The
(Delta = 0.096). reason may be that the molecular structure of these compounds
Intuitive analysis: The results in Table 2 showed that the mean had some correlations during HPLC detection because of the influ-
fx for all the nine components on all columns and instruments by encing factors such as solvent effect, conjugate effect, chro-
different analysts were 0.54, 6.79, 7.37, 0.55, 1.81, 1.87, 0.88, mophore (auxochrome), absorption band or electronic transition.
0.69, and 0.69 respectively. And the RSDs for all were 1.75%, Such factors act together to produce the similar fx.
3.18%, 4.47%, 2.44%, 2.93%, 2.86%, 3.67%, 1.15%, and 1.12% respec- Based on the above analysis results, it can be seen that the con-
tively. It can be seen that the deviations of fx values for all the nine version factors were more prone to be affected by the column. The
components were 65%. The influences of stationary phase, LC column had the most significant influence on the results. Still, the
instrument and analyst on the results of fx for nine compounds errors caused by different instrument, analyst and column were
were further studied by Intuitive analysis. Through Intuitive anal- small. Maybe it was because different instruments such as sensitiv-
ysis, the main factors that affect the fx may be found. The analysis ity, integration or other parameters and different columns, analysts
of range (R) in S11 (in Supporting Information) showed the impor- were different, but if the different compounds were detected in the
tance of factors effected on fx of GC were column > instru- same instrument and column, the influence might be more or less
ment > analyst. While on fx of EGC, the sequence was identical, the results of fx would not differ a lot. Although fx was
column > analyst > instrument. The effect of above elements on inevitable to change under different analysts, instruments, or col-
the other 7 components was small. Altogether, the range results umns, these changes did not affect the fx results significantly and
showed that the errors caused by different instrument, analyst the range of its fluctuations could be acceptable. QAMS could be
and column were small. This provides the foundation for further applied in different laboratories with carefully controlling the
examining the stability of fx. parameter of column.

Table 2
The results of fx for nine compounds by 18-run Taguchi Design – Static design⁄.

Run order Stationary phase LC instrument Analyst EGCGa GA GC EGC CAF C EC GCG ECG CG
1 GRACE Agilent 1 1 0.53 6.52 7.51 0.56 1.74 1.87 0.86 0.68 0.67
2 GRACE Wufeng 2 1 0.53 6.56 7.40 0.56 1.75 1.85 0.87 0.68 0.67
3 GRACE Shimadzu 3 1 0.53 6.68 6.92 0.54 1.81 1.87 0.89 0.69 0.68
4 APROMA Agilent 1 1 0.55 6.83 7.61 0.57 1.81 1.86 0.88 0.69 0.68
5 APROMA Wufeng 2 1 0.56 7.02 7.79 0.57 1.82 1.86 0.88 0.69 0.68
6 APROMA Shimadzu 3 1 0.55 6.77 7.23 0.53 1.79 1.82 0.86 0.70 0.69
7 Hypersil Agilent 2 1 0.55 6.98 7.45 0.56 1.83 1.87 0.89 0.71 0.70
8 Hypersil Wufeng 3 1 0.54 6.79 7.29 0.57 1.85 1.88 0.90 0.71 0.70
9 Hypersil Shimadzu 1 1 0.55 6.47 7.59 0.57 1.92 1.96 0.91 0.69 0.69
10 Agilent Agilent 3 1 0.53 6.96 6.88 0.54 1.79 1.86 0.89 0.69 0.69
11 Agilent Wufeng 1 1 0.54 6.72 6.74 0.55 1.85 1.89 0.93 0.69 0.68
12 Agilent Shimadzu 2 1 0.56 6.31 7.21 0.56 1.90 1.99 0.93 0.69 0.69
13 Phenomenex Agilent 2 1 0.53 7.08 7.81 0.54 1.70 1.76 0.84 0.69 0.69
14 Phenomenex Wufeng 3 1 0.53 6.89 7.94 0.54 1.75 1.78 0.84 0.69 0.69
15 Phenomenex Shimadzu 1 1 0.54 7.00 7.12 0.55 1.83 1.91 0.92 0.69 0.68
16 Kromasil Agilent 3 1 0.54 7.03 7.21 0.54 1.80 1.86 0.84 0.70 0.69
17 Kromasil Wufeng 1 1 0.54 6.84 7.43 0.54 1.79 1.86 0.84 0.70 0.69
18 Kromasil Shimadzu 2 1 0.54 6.74 7.54 0.54 1.79 1.87 0.84 0.70 0.69
Means 1 0.54 6.79 7.37 0.55 1.81 1.87 0.88 0.69 0.69
SD 0 0.01 0.22 0.33 0.01 0.05 0.05 0.03 0.01 0.01
RSD (%) 0 1.75 3.18 4.47 2.44 2.93 2.86 3.67 1.15 1.12

Obtained by Minitab 16, DOE, Taguchi Design, L18 (61  32).
a
EGCG was used as a quality indicator.
D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120 1117

Fig. 2. Residual plots for mean value.

3.4. The system suitability and chromatographic peaks identification (S15 in Supporting Information) were 0.000, and the model yielded
an R2 value of 0.994, 0.990, 0.983, 0.991, 0.997, 0.987, 0.996, 0.976
The system suitability was evaluated via tailing factor, theoret- and 0.986, respectively. The results implied that there was good
ical plate number and resolution (S14 in Supporting Information). statistically significant correlation between the two variables,
The theoretical plate number ranged from 2667 to 94,488, respec- and there were no significant differences in the content results
tively. The resolutions of the ten compounds were >1.5, which obtained by QAMS and ESM.
demonstrated that the compound peaks were not disturbed by The results could be predictable from Eqs. (1) and (2). This is
adjacent peaks in the quantization of their amounts. The tailing because under the reliable chromatographic conditions, the nature
factor of the compounds peaks ranged from 0.74 to 1.3, indicating of the calculation of fx was the intrinsic function and proportion
the absence of severe peak fronting or tailing. relationship between the selected reference analyte and other ana-
For better authentication as well as convenience to quality con- lytes. The various errors related with instruments, columns, ana-
trol of the commercial LJGTEs, the relative retention time (RRT) of lysts, reagents, experimental methods or environment conditions
the other nine components (relative to EGCG) by different had been considered in the process of calculation of fx. Thus if
columns, instruments and analysts (Table 2) was used to identify the deviations of fx in different columns, instruments and analysts
its chromatographic peak position. EGCG was explicitly identified were small, the differences between QAMS and ESM would be
and designated as the reference peak in the LJGTEs samples. The small as well. On the other hand, large variations in fx would result
mean value of RRT relative to EGCG was 0.20 ± 0.02 (GA), in large differences in the content results. The deviations also came
0.26 ± 0.03 (GC), 0.42 ± 0.03 (EGC), 0.53 ± 0.02 (CAF), 0.60 ± 0.03 from the differences between the average fx and the single fx
(C), 0.91 ± 0.03 (EC), 1.24 ± 0.05 (GCG), 1.63 ± 0.05 (ECG), obtained from the Thermo ODS hypersil column and Wufeng
1.74 ± 0.05 (CG), respectively (mean ± SD). The results showed that instrument. Although using the same fx may obtain a smaller dif-
RRT was stable and could be used to identify the other nine com- ference, to check the differences in different laboratory between
pounds referring to EGCG in despite of the differences in columns QAMS and ESM, the average fx was used for the calculation of the
or HPLC instruments, or other factors in different laboratories. content of nine components. With the comparisons of the content
Besides, the chromatographic peak area relative to EGCG or ECG of all the nine components in LJGTEs by ESM and QAMS (see S16 in
which was obviously different in the chromatograms of LJGTEs, Supporting Information), it was shown that all the contents were
could also be an efficient way for identification and location. located within the vicinity of linearity y = x (Fitting degree:
R2 = 0.9991). The data showed that using EGCG as reference for cal-
3.5. The assessment of QAMS method culation by QAMS, the contents variations were much lower than
the quantification error. Their contents variations within the range
In order to evaluate and verify QAMS feasibility for the determi- of 5% which was the requirement of Pharmacopoeia indicated that
nation of multi-compounds in LJGTEs, the contents of eight tea cat- QAMS can be applied in the determination of eight tea catechins,
echins, GA and CAF were determined by ESM and QAMS in 26 GA and CAF in different labs.
batches, respectively. The linear regression model was built
between the two variables, to examine the deviations between 3.6. Quality evaluation of LJGTEs by QAMS
QAMS and ESM. The independent variable was determined by
QAMS and the dependent variable was determined by ESM. All The contents of eight tea catechins, GA and CAF in 26 batches of
the statistically significant coefficients in the regression model LJGTEs obtained by QAMS were listed in Table 1. It was observed
1118 D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120

Fig. 3. Dendrograms of hierarchical cluster analysis for the 26 tested samples of LJGTEs. The abscissa indicates the Euclidean distances, whereas the ordinate expresses the
sample numbers.

that among the ten components, the mean content of GA concentration was EGCG (524.6 mg/g) in producers No. s9 and the
(16.2 mg/g), CAF (10.9 mg/g), C (10.5 mg/g) and CG (5.0 mg/g) analyte of lowest concentration was CG (1.2 mg/g) in No. s21 and
accounted for lower content in LJGTEs. Conversely, the mean con- No. s22 and No. s23. The maximum total content of 10 compounds
tent of EGC (183.9 mg/g), EGCG (445.9 mg/g) and ECG (102.8 mg/g) was 982.2 mg/g, while the minimum total content of 10
was in higher content. For all the 26 batches, the analyte of highest compounds was 730.1 mg/g.
The data in Table 1 as discussed above revealed differences
among various sample pairs. To show the clear classification of
the samples and the relationships among the samples and evaluate
the variation of LJGTEs, HCA and CDA were performed using the
IBM SPSS Statistics 19.0 software (IBM, USA) (Alaerts et al., 2014;
An, Wang, Lan, Hashi, & Chen, 2013; He, Zhu, Goda, Cai, &
Komatsu, 2014; He et al., 2009; Kim, Shin, & Seo, 2014; Kong
et al., 2014). HCA can efficiently produce a qualitative and quanti-
tative representation of the original experimental results through
statistical organization and graphical display distinguishing differ-
ent herbal species in TCMs. The dendrograms (Fig. 3) and icicles
(Fig. 4) were constructed according to HCA based on the squared
Euclidean distance. It showed the relationships and distribution
patterns among the LJGTEs samples collected from seven produc-
ers of different batches from different origins. The contents of eight
principal bioactive compounds in 26 batches of commercial LJGTEs
samples were selected as the clustering variable.
The dendrograms clearly showed that the 26 tested samples of
LJGTEs were divided into three main clusters. Cluster one consisted
of nine samples collected from producers LWP, SOP. The icicles
Fig. 4. Icicles for the 26 tested samples of LJGTEs. clearly showed that these samples could be further divided into
D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120 1119

three groups (S17; S21; the rest, respectively). Cluster two consisted References
of 14 samples collected from producers ZGP, XGP, STP, HSP, and SSP.
Cluster three consisted of three samples collected from producers Alaerts, G., Pieters, S., Logie, H., Van Erps, J., Merino-Arevalo, M., Dejaegher, B., ...
Vander Heyden, Y. (2014). Exploration and classification of chromatographic
SSP. The cluster results were consistent with the icicles. The classi- fingerprints as additional tool for identification and quality control of several
fication of results coincided well with the geographical origin of the Artemisia species. Journal of Pharmaceutical and Biomedical Analysis, 95, 34–46.
samples. Longjing green tea was mainly divided into two groups An, H., Wang, H., Lan, Y., Hashi, Y., & Chen, S. (2013). Simultaneous qualitative and
quantitative analysis of phenolic acids and flavonoids for the quality control of
based on origin: West-lake Longjing and Zhejiang Longjing. West- Apocynum venetum L. leaves by HPLC-DAD-ESI-IT-TOF-MS and HPLC-DAD.
lake Longjing was distributed in the West Lake District in Hangzhou, Journal of Pharmaceutical and Biomedical Analysis, 85, 295–304.
while Zhejiang Longjing was mainly distributed in Qiantang and Du, Y., Li, Q., Liu, J., Yin, Y., & Bi, K. (2014). Combinative method using multi-
components quantitation by single reference standard and HPLC fingerprint for
Shaoxing District. Group one belonged to West-lake Longjing. Group
comprehensive evaluation of Rhodiola crenulata H.Ohba. Analytical Methods, 6,
two belonged to Shaoxing Longjing. Lastly, group three belonged to 5891.
Qiantang Longjing. The samples produced by SSP were from two dif- Fujiki, H., Suganuma, M., Okabe, S., Komori, A., Sueoka, E., Sueoka, N., ... Sakai, Y.
ferent origins: Qiantang and Shaoxing District. The results indicated (1996). Japanese green tea as a cancer preventive in humans. Nutrition Reviews,
54, 67–70.
variations in component compositions and quantities were the Gao, X. Y., Jiang, Y., Lu, J. Q., & Tu, P. F. (2009). One single standard substance for the
result of different origins. The main difference between the different determination of multiple anthraquinone derivatives in rhubarb using high-
clusters was the source of Longjing green tea. A possible explanation performance liquid chromatography-diode array detection. Journal of
Chromatography A, 1216, 2118–2123.
was that the different origins of Longjing green tea had been culti- Goupy, J. (2005). What kind of experimental design for finding and checking
vated in different environments or in a different manner. These fac- robustness of analytical methods? Analytica Chimica Acta, 544, 184–190.
tors include soil type, climate, harvesting time, harvesting process, He, W., Hu, X., Zhao, L., Liao, X., Zhang, Y., Zhang, M., & Wu, J. (2009). Evaluation of
Chinese tea by the electronic tongue: correlation with sensory properties and
storage condition and cultural practices. The harvesting time of classification according to geographical origin and grade level. Food Research
Longjing green tea was mainly concentrated in spring and summer. International, 42, 1462–1467.
The differences and similarity between different samples were also He, F. C., Li, S. X., Zhao, Z. Q., & Dong, J. P. (2012b). Simultaneous quantitative
analysis of four lignanoids in Schisandra chinensis by quantitative analysis of
shown in approximate matrix table (squared Euclidean distance) multi-components by single marker. Acta Pharmaceutica Sinica, 47, 930–933.
(see S17 in Supporting Information). The values in the table repre- He, B., Yang, S.-Y., & Zhang, Y. (2012a). A new method of calibration and positioning
sented the similarity coefficient between each sample, in which in quantitative analysis of multi-components by single marker. Acta
Pharmaceutica Sinica, 47, 1653–1659.
the larger value represented a greater difference and the similarity
He, J. Y., Zhu, S., Goda, Y., Cai, S. Q., & Komatsu, K. (2014). Quality evaluation of
was relatively low. medicinally-used Codonopsis species and Codonopsis Radix based on the
Furthermore, CDA was introduced to create variables by taking a contents of pyrrolidine alkaloids, phenylpropanoid and polyacetylenes. Journal
special linear combination of the original variables. Based on the of Natural Medicines, 68, 326–339.
Hou, J. J., Wu, W. Y., Da, J., Yao, S., Long, H. L., Yang, Z., ... Guo, D. A. (2011).
results of HCA, Longjing green tea was divided into three categories Ruggedness and robustness of conversion factors in method of simultaneous
based on origin: West-lake Longjing, Qiantang Longjing and Shaox- determination of multi-components with single reference standard. Journal of
ing Longjing. The canonical variables of eight catechins were cre- Chromatography A, 1218, 5618–5627.
Jankun, J., Selman, S. H., Swiercz, R., & Shrzypczak-Jankun, E. (1997). Why drinking
ated so that they contained the most useful information in a set green tea could prevent cancer. Nature, 387, 561.
of original variables. To a certain extent, CDA may illustrate the dis- Jin, J. Q., Ma, J. Q., Ma, C. L., Yao, M. Z., & Chen, L. (2014). Determination of catechin
tances between the classifications under investigation in a reduced content in representative Chinese tea germplasms. Journal of Agricultural and
Food Chemistry, 62, 9436–9441.
dimensional space. This is shown in S18 of the Supporting Informa- Kim, J.-H., Shin, H.-K., & Seo, C.-S. (2014). Development of a quantitative analysis
tion where the three categories were significantly different from method for the 12 marker compounds in Palmijihwang-hwan, a herbal formula,
each other. For new samples of LJGTEs, with the results obtained using a reversed-phase C18 column and an amino column by HPLC. Analytical
Methods, 6, 3763.
by QAMS and HCA, it only needs to input the corresponding posi- Kong, M., Liu, H. H., Xu, J., Wang, C. R., Lu, M., Wang, X. N., ... Li, S. L. (2014).
tion in the joint distribution map; the map would show the new Quantitative evaluation of Radix Paeoniae Alba sulfur-fumigated with different
location of data. Through the position, which was a classification, durations and purchased from herbal markets: simultaneous determination of
twelve components belonging to three chemical types by improved high
it can be seen which classification it belonged to and the differences
performance liquid chromatography-diode array detector. Journal of
between each. The presented method suggested that the quality of Pharmaceutical and Biomedical Analysis, 98, 424–433.
different LJGTEs samples were not consistent, and it may be helpful Lee, M. S., Hwang, Y. S., Lee, J., & Choung, M. G. (2014). The characterization of
to group and classify different Longjing green tea according to their caffeine and nine individual catechins in the leaves of green tea (Camellia
sinensis L.) by near-infrared reflectance spectroscopy. Food Chemistry, 158,
cultivation and origin. It was observed that the QAMS strategy was 351–357.
accurate, stable, feasible and credible. Based on the reliable chro- Li, S. P., Qiao, C. F., Chen, Y. W., Zhao, J., Cui, X. M., Zhang, Q. W., ... Hu, D. J. (2013). A
matographic conditions, by the determination of EGCG which was novel strategy with standardized reference extract qualification and single
compound quantitative evaluation for quality control of Panax notoginseng
easy to be obtained, QAMS results can be applied in the content cal- used as a functional food. Journal of Chromatography A, 1313, 302–307.
culation of other nine components and in the quality control and Liao, S., Umekita, Y., Guo, J., Kokontis, J.-M., & Hiipakka, R.-A. (1995). Growth
evaluation of LJGTEs in different laboratories. inhibition and regression of human prostate and breast tumors in athymic mice
by tea epigallocatechin gallate. Cancer Letters, 25, 239–243.
Maeda-Yamamoto, M., Ema, K., Monobe, M., Tokuda, Y., & Tachibana, H. (2012).
Acknowledgments Epicatechin-3-O-(3’’-O-methyl)-gallate content in various tea cultivars
(Camellia sinensis L.) and its in vitro inhibitory effect on histamine release.
Journal of Agricultural and Food Chemistry, 60, 2165–2170.
The authors would like to thank Skyherb Ingredients for the col- National Commission of Chinese Pharmacopoeia, Pharmacopoeia of People’s
lections of different batches of Longjing green tea extracts for this Republic of China, vol. I, (2010). China Medical Science Press, Beijing, 285.
study. The authors are grateful to analysis and measurement of Zhe- Peng, A. H. (2013). Application of minitab software in orthogonal experimental
design with duplicate test. Journal of Jimei University, 1, 111–114.
jiang University for providing MS and NMR data. This project is Prescription of Chinese Medicinal, Drug Standards Issued by Ministry of Public
financially supported by the Chinese Pharmacopoeia Commission. Health. (1998). Vol. 17. WS3-B-3151-98.
Samanidou, V., Tsagiannidis, A., & Sarakatsianos, I. (2012). Simultaneous
determination of polyphenols and major purine alkaloids in Greek Sideritis
Appendix A. Supplementary data species, herbal extracts, green tea, black tea, and coffee by high-performance
liquid chromatography-diode array detection. Journal of Separation Science, 35,
608–615.
Supplementary data associated with this article can be found, in Serpen, A., Pelvan, E., Alasalvar, C., Mogol, B. A., Yavuz, H. T., Gokmen, V., ... Ozcelik,
the online version, at http://dx.doi.org/10.1016/j.foodchem.2015. B. (2012). Nutritional and functional characteristics of seven grades of black tea
11.101. produced in Turkey. Journal of Agricultural and Food Chemistry, 60, 7682–7689.
1120 D.-W. Li et al. / Food Chemistry 197 (2016) 1112–1120

Stoner, G.-D., & Mukhtar, H. (1995). Polyphenols as cancer chemopreventive agents. Wang, H., Provan, G. J., & Helliwell, K. (2003). HPLC determination of catechins in tea
Journal of Cellular Biochemistry, 22, 169–180. leaves and tea extracts using relative response factors. Food Chemistry, 81,
Tian, L., Shi, X., Yu, L., Zhu, J., Ma, R., & Yang, X. (2012). Chemical composition and 307–312.
hepatoprotective effects of polyphenol-rich extract from Houttuynia cordata Wang, Q. F., Sun, P. P., & Liu, Z. J. (2014). Design of automotive pressed parts
tea. Journal of Agricultural and Food Chemistry, 60, 4641–4648. process parameters in orthogonal test based on autoform. Auto Engineer, 2,
United States Pharmacopeia, 32-NF 27, (2009). US Pharmacopoeial Convention, 34, 33–35.
794. Yang, C. S. (1997). Inhibition of carcinogenesis by tea. Nature, 389, 134–135.
United States Pharmacopeia, 33-NF 28. (2010a). Vol. I. Rockville, MD 1046: US Yang, C. S., Wang, X., Lu, G., & Picinich, S. C. (2009). Cancer prevention by tea: animal
Pharmacopoeial Convention. studies, molecular mechanisms and human relevance. Nature Reviews Cancer, 9,
United States Pharmacopeia, 33-NF 28. (2010b). Vol. I. Rockville, MD 1037: US 429–439.
Pharmacopoeial Convention. Zhang, R., Chen, J., Shi, Q., Li, Z., Peng, Z., Zheng, L., & Wang, X. (2014). Quality control
vander Hooft, J. J., Akermi, M., Unlu, F. Y., Mihaleva, V., Roldan, V. G., Bino, R. J., ... method for commercially available wild Jujube leaf tea based on HPLC
Vervoort, J. (2012). Structural annotation and elucidation of conjugated characteristic fingerprint analysis of flavonoid compounds. Journal of
phenolic compounds in black, green, and white tea extracts. Journal of Separation Science, 37, 45–52.
Agricultural and Food Chemistry, 60, 8841–8850. Zhang, Y., Li, Q., Xing, H., Lu, X., Zhao, L., Qu, K., & Bi, K. (2013). Evaluation of
Wang, Z.-M., Gao, H.-M., Fu, X.-T., & Wang, W.-H. (2006). Multi-components antioxidant activity of ten compounds in different tea samples by means of an
quantitation by one marker new method for quality evaluation of Chinese on-line HPLC–DPPH assay. Food Research International, 53, 847–856.
herbal medicine. Zhong guo Zhong Yao Za Zhi, 31, 1925–1928. Zhao, F., Lin, H. T., Zhang, S., Lin, Y. F., Yang, J. F., & Ye, N. X. (2014). Simultaneous
Wang, H., & Helliwell, K. (2001). Determination of flavonols in green and black tea determination of caffeine and some selected polyphenols in Wuyi Rock tea by
leaves and green tea infusions by high-performance liquid chromatography. high-performance liquid chromatography. Journal of Agricultural and Food
Food Research International, 34, 223–227. Chemistry, 62, 2772–2781.

Potrebbero piacerti anche