Quality Analysis of Long dan Xie gan Pill by a Combination of Fingerprint and Multicomponent Quantification with Chemometrics Analysis

Long dan Xie gan pill is a traditional complex compound preparation with a long history for treatment of diseases, including hepatocolic hygropyrexia, dizziness, tinnitus, and deafness. Quality of products from different manufacturers may be varied. Since the current standard could not control the quality of products in a comprehensive and effective way, this study aimed at establishing a practical and convenient approach for holistic quality control of the preparation. This study included both qualitative and quantitative works to get information on the overall composition and main components, respectively. As a result, HPLC fingerprint (UV 240 nm) similarities of all fifty samples were in the range of 0.65∼0.99. Results indicated that there was a difference among products from different manufacturers. Additionally, ten characteristic peaks of the fingerprint were tentatively identified by LC-MS. Further chemometrics analysis was utilized to evaluate the products from different manufacturers. At the same time, the HPLC (UV 285 nm) multicomponent quantification result showed that contents of gentiopicrin, baicalin, baicalein, and wogonin were in the range of 0.61–5.40, 1.96–5.33, 0.10–3.40, and 0.046–1.16 mg·g−1, respectively. Data analysis verified the main different component of baicalein from the fingerprint statistical analysis. It is worth mentioning that the qualitative fingerprint and quantitative multicomponent determination were simultaneously accomplished by HPLC-DAD with dual channels. The study provided sound basis for improving quality control standards. This study also provided practical strategy for overall quality control of traditional Chinese medicines.


Introduction
Long dan Xie gan pill is prepared from ten species of crude drugs including Gentianae radix et rhizoma, Scutellariae radix, and Akebiae caulis (Mutong) in Chinese Pharmacopoeia (2015 Edition, Volume I) [1]. It is widely used for the treatment of diseases, including hepatocolic hygropyrexia, dizziness, tinnitus and deafness, hypochondriac pain [1]. e preparation has attracted widespread attention, since it caused aristolochic acid nephropathy (AAN) [2,3]. At that time, the prescription collected the crude drug of Caulis aristolochiae manshuriensis (Guanmutong) instead of Akebiae caulis [4][5][6]. Because of the serious adverse effect of aristolochic acids, the medicinal standard of Caulis aristolochiae manshuriensis was abolished and replaced by Akebiae caulis without containing such toxic constituents since 2003.
As one commonly used Chinese patent medicine (CPM) with a long history, Long dan Xie gan pill has about 200 manufacturers.
erefore, the quality of products from different manufacturers may be varied. Since quality is directly related to drug safety and efficacy, it is very important to evaluate the holistic quality of the products. Researchers have been working on the essential quality control and evaluation methods for years [2]. For most CPMs, the effective components are not clear and the consistency of product quality is a key indicator of quality product evaluation. It is worth mentioning that fingerprint is an internationally recognized effective method because it could reflect the overall quality information [2][3][4][5][6][7][8][9][10]. However, fingerprint is usually used for qualitative consistency evaluation. And the quantification could be achieved by applying multicomponent determination [10][11][12][13]. In recent years, more and more chromatographic and spectroscopic methods including LC, LC-MS n , and quantitative nuclear magnetic resonance (QNMR) are applied for the aforementioned qualitative and quantitative work [14][15][16][17][18][19][20][21]. Among these methods, high-performance liquid chromatography (HPLC) is still the main method deployed in quality control of traditional Chinese medicines (TCMs) because of its advantages, including good repeatability, wide application, and high efficiency.
In this study, the qualitative and quantitative consistency information of Long dan Xie gan pill samples was achieved at the same time by a combination of fingerprint with multicomponent quantification by HPLC-diode array detector (DAD) with dual channels (UV 240 nm and 285 nm). Additionally, further deep mining of the data by chemometrics analysis helped to evaluate the differences of products in a more comprehensive and effective way. e results indicated that the established method could comprehensively analyze the product quality.
is strategy could provide a practical approach for the holistic quality control of TCM.

Preparation of Standard Solutions.
Standard stock solutions of baicalin (0.1 mg·mL −1 ) were prepared by dissolving suitable amounts of reference substance in methanol for fingerprint establishment.

Mass Spectrometry Condition. MS analysis was per-
formed on an Agilent1260-6410B LC-MS couplet system equipped with Agilent Mass Hunter ChemStation (Agilent, Santa Clara, USA). e mass spectrometry settings were as follows: split ratio � 1 : 9; desolvation temperature: 350°C; desolvation air flow N 2 : 540 L·h −1 ; nebulizer pressure: 30 psi; and capillary: 4000 V. Both positive and negative modes were performed with a scan range of m/z 50-1200.

Results of Fingerprint.
For fingerprint study, all the samples were prepared and analyzed according to conditions under 2.5 and 2.6. Baicalin (t R � 34.02 min) was taken as the reference peak. Relative retention times (RRTs) and relative peak areas (RPAs) of the characteristic peaks were calculated for method validation.

Instrument Precision.
e same sample solution (no. 21) was injected for six consecutive times. e result showed that the RSDs of RRTs and RPAs were in the range of 0.073%-2.0% and 0.26-1.74%, respectively. It showed that the precision of the instrument was good.

Repeatability.
e same batch sample (no. 21) was taken and prepared for six independent sample solutions for analysis. e result showed that the RSDs of RRT and RPA were in the range of 0.010%-0.44% and 0.51-2.24%, respectively. It indicated that method repeatability was good.

Stability.
e same sample solution (no. 21) was injected at 0, 4, 8, 12, 16, 20, and 24 h at room temperature. e result showed that the RSDs of RRT and RPA were in the range of 0.039%-0.81% and 0.71-4.58%, respectively. It demonstrated the sample solution was stable within 24 h.

Establishment of Fingerprint.
After fifty batches of the sample solutions were analyzed, their chromatograms (UV 240 nm) were recorded ( Figure 3) and imported to ChemPattern software. All variables were used as the common peak screening condition. And Gauss curve simulation method was applied to generate the common mode with 16 characteristic peaks ( Figure 4). All sample chromatograms were analyzed by comparison with the common mode.

Identification and Attribution of Characteristic Peaks.
e sample solution was analyzed according to the conditions under Sections 2.6 and 2.7. By combination with the chromatographic behavior of the components, ten characteristic peaks were identified by comparing with reference standards. Also the main origins of the peaks were attributed (Table 2), and they were mainly from five species of crude drugs in the prescription.

Statistical Analysis
(1) Similarity Analysis. e above HPLC (UV 240 nm) fingerprint common mode was taken as a reference. During the analysis, the included angle cosine method was used to calculate the similarity of each sample ( Figure 5). Finally, the similarities of all samples were in the range of 0.65∼0.99. Among them, similarities of seven batches of samples were lower than 0.8, including all samples from enterprise B. e result indicated that there was a certain difference among the overall product quality of these samples from others. Also it was clear that the uniformity of most BCP samples (no. 6∼10 and 26∼30) was not good as the WBP ones.
(2) Principle Component Analysis. Principle component analysis (PCA) was carried out after standardization of all sample data (Figures 6 and 7). e contribution rates of the first and second principle component (PC1) were 45.92% and 39.22%, respectively. And the total contribution rate of 85.14% showed that it could reflect the differences between samples in a more comprehensive way. PCA scatter plot ( Figure 6) displayed that samples from each enterprise basically could be grouped into a class. It showed that samples from enterprise B deviated far away from others. e principle component load diagram (Figure 7) gave the proportion of each chromatographic peak in the principal component. And the greater the distance from X � 0 longitudinal axis, the greater the contribution to PC1, such as gentiopicrin and baicalein. Likewise, the greater the distance from Y � 0 transverse axis, the greater the contribution to PC2, such as baicalein and geniposide. e result displayed that samples from enterprise B separated with others along both PC1 and PC2. erefore, the major  contributions to PC1 and PC2 were their main differential components from others. Because of the location at the position of PC1 > 0 and PC2 < 0, contents of such compounds were in positive correlation with PC1 and in negative correlation with PC2. erefore, the contents of baicalein and wogonin were higher in these samples. Along the PC1, the concentration of baicalein distinguished the samples of enterprise B, whose values were higher than those presented for the others (Figure 6). On the contrary, along with the PC2, the levels of geniposide and gentiopicrin showed that there was a tendency for the separation of the products of enterprises D and G from others. One sample of enterprise A was grouped with samples of enterprises D and G, which presented higher levels for these compounds.

Journal of Analytical Methods in Chemistry
(3) Cluster Analysis. Hierarchical cluster analysis (HCA) is a conventional cluster analysis method. It is a detection tool that clearly reveals the natural grouping of data. e block distance was selected for distance calculation and HCA (Figure 8) was performed by error square sum method. Similar to the result of PCA, except that some samples from manufacturer D are not distinguished from those of A, other samples from different enterprises could basically distinguish. Additionally, samples from B were relatively most far away from others. It indicated there existed some differences of these samples.

Linearity, LOD, and LOQ.
Working standard solutions containing gentiopicrin, baicalin, baicalein, and wogonin were prepared by diluting the stock mixed solution with methanol to a series of proper concentrations. en, they were injected and analyzed. e results of regression equations, linearity, determination coefficient, and limits of detection and quantification of the method are presented in Table 3. e linear range varied from 3.25 to 492.56 μg mL −1 , in accordance with the analyte. All analytes presented a determination coefficient (R 2 ) of the 0.9999, which allows the method to be considered linear. e limits of detection (LOD) and quantification (LOQ) were calculated according to guidelines for validation of analytical methods for pharmaceutical quality standards [22].

Instrument Precision.
e same sample solution (no. 21) was injected for six consecutive times and analyzed. e RSDs of peak areas for gentiopicrin, baicalin, baicalein, and wogonin were 0.63%, 0.29%, 0.41%, and 0.15%, respectively. It indicated that the precision of the instrument was in accordance with the requirement in guidelines for validation of analytical methods for pharmaceutical quality standards [22].

Repeatability.
e same batch of sample (no. 21) was taken and prepared for six independent sample solutions. en, they were analyzed according to conditions under 2.6.  e average contents of gentiopicrin, baicalin, baicalein, and wogonin were 4.77, 3.84, 0.62, and 0.40 mg·g −1 , respectively. And the RSDs were 0.49%, 1.11%, 0.40%, and 0.56%, respectively. It indicated that method repeatability was in accordance with the requirement in guidelines for validation of analytical methods for pharmaceutical quality standards [22].

Recovery.
e recovery experiment was performed by adding a known amount of individual reference standards into a certain amount of sample (no. 21).
3.2.6. Sample Analysis. Fifty batches of sample solutions were prepared and analyzed. e results (Table 5) displayed that the contents of gentiopicrin, baicalin, baicalein, and wogonin were in the range of 0.61-5.40, 1.96-5.33, 0.10-3.40, and 0.046-1.16 mg·g −1 , respectively. It was easily to find the differences among samples from different enterprises by the scatter diagram ( Figure 9). It showed that the general content trends of baicalin, baicalein, and wogonin were basically similar. Among them, the contents of baicalein and wogonin in samples from B were apparently higher than others; especially, the content of baicalein was much higher. e determination result was in accordance with the abovementioned PCA analysis result.

Investigation of Extraction Methods.
e extraction method was optimized in order to make the fingerprint reflect the chemical composition information as much as possible. For both dosage forms of samples, different extraction solvent (80% methanol and 50% methanol-water), and extraction mode and time (ultrasonic extraction for 30 min, 45 min, and 60 min) were investigated. e result showed that extraction time had little effect on both dosage forms. For WBP samples, the chromatogram could reflect rich chemical information with good separation of peaks with methanol extraction for 30 min. While for BCP     Meanwhile, UV 285 nm was determined as detection wavelength for simultaneous determination of main compounds due to the good separation.

Conclusions
Quality control is the key issue in modernization and internationalization of TCM. Qualitative fingerprint and quantitative multicomponent determination have been demonstrated as the comprehensive and effective way to accomplish the holistic quality analysis. In this study, both qualitative and quantitative works to get the overall composition and main components information were accomplished simultaneously by HPLC with dual-channel detection. Moreover, further deep mining of the data by chemometrics analysis helped to evaluate the quality of the 0  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  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49   Journal of Analytical Methods in Chemistry 9 preparation from different manufacturers.
e result indicated that this approach is a powerful tool for quality control of TCM.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.