Quality Evaluation of Market Acacia catechu by Fingerprint- Chemical Pattern Recognition

College of Biological and Pharmaceutical Engineering, West Anhui University, Lu’an 237012, China Department of Pharmacy, e First A­liated Hospital of Anhui University of Chinese Medicine, Hefei 230031, China College of Pharmacy, Anhui University of Chinese Medicine, Hefei 230012, China National Resource Center for Chinese Materia Medica, China Academy of Chinese Medical Sciences, Beijing 100700, China


Introduction
Traditional Chinese medicine Acacia catechu is the dried decoction of peeled branches and trunks of Acacia catechu (L.f.) Willd. [1], which has antidiabetic [2], antihypertensive [3], hepatoprotective [4], antioxidant [5], antibacterial [6], anticancer [7,8], anti-in ammatory [9], and immunomodulatory e ects [10] and is mainly used to treat eczema, mouth ulcers, diarrhea, bruising, and traumatic hemorrhage [11,12]. In China, the medicinal use of Acacia catechu started from "Compendium of Herbology" in the Ming Dynasty (AD 1552-1578), known as "Wudieni" and "Haiercha." In addition, it is worth paying attention to that the origin of Acacia catechu also includes the dried extract of the decoction of the leafy shoots of Uncaria gambir (Hunter) Roxb, which was recorded in the Dictionary of Chinese Pharmacy, National Compilation of Chinese Herbal Medicine ( ird Edition), Chinese Basic Medicinal Herbs, etc. erefore, there is a need for a standardized assessment and monitoring of the quality of market Acacia catechu.
Market research by our team found that there are only domestic Acacia catechu and imported Acacia catechu on the market. Moreover, the price of imported Acacia catechu is two times higher than that of domestic products. However, the intrinsic quality of market Acacia catechu is also mixed and varies greatly. In terms of color appearance, the surface color of both imported and domestic Acacia catechu is black or reddish-brown. After crushing, the color of Acacia catechu powder varies from earthy yellow to reddish-brown. Importantly, domestic Acacia catechu is mostly reddishbrown, while imported Acacia catechu is mostly earthy yellow ( Figure 1). erefore, it is important to explore whether the appearance correlates with the herbs' intrinsic quality. Simultaneously, the research questions need to be urgently studied and explained, including whether the division between imported Acacia catechu and domestic Acacia catechu in the market can reflect the intrinsic quality of the herbs and whether the price difference has any actual scientific connotation. e total content of catechin and epicatechin is the key indicator for evaluating Acacia catechu quality in the Chinese Pharmacopoeia (2020 version) [1]. However, the complexity and diversity of the chemical components of traditional Chinese medicine determine that the detection results of only a few components as evaluation indicators cannot reflect the quality of traditional Chinese medicine comprehensively and accurately [13]. e chromatographic fingerprint in particular is significant for the evaluation of traditional Chinese medicine, which can better reflect the overall chemical composition of traditional Chinese medicine and is widely used [14][15][16][17]. Li and Chen [18] have established the chromatographic fingerprint of Acacia catechu by HPLC. Based on this, this study enriched the scientificity, representativeness, and universality of test samples and optimized the chromatographic conditions. Furthermore, HPLC fingerprints of 47 batches of Acacia catechu were established, and the contents of catechin and epicatechin in Acacia catechu were determined. e similarity evaluation combined with cluster analysis (CA), principal component analysis (PCA), and orthogonal partial least squares discriminant analysis (OPLS-DA) was used to analyze the quality grades of market Acacia catechu, in order to provide a scientific basis for developing classification standards and quality evaluation of Acacia catechu.

Preparation of Sample
Solutions. Acacia catechu was crushed and filtered through a 65-mesh sieve. An amount of 100 mg Acacia catechu powder was accurately weighed and placed in a 25 mL volumetric flask with a stopper, and 50% methanol was added to the scale. Ultrasonication was performed for 20 min, and then 50% methanol was used to compensate for the weight loss during the extraction. e extract was filtered through a 0.22 μm membrane and stored at 4°C for further experiments.

Similarity Analysis.
e raw HPLC chromatographic data of 47 batch samples were exported as AIA format files. e software "Similarity Evaluation System for Chromatographic Fingerprint of Traditional Chinese Medicine" (Version 2012, Chinese Pharmacopoeia Committee) was used to analyze similarity. e reference fingerprint was obtained automatically by the median method with a time window width of 0.5 min, and the similarity values of all the samples were calculated.

Chemical Pattern Recognition Analysis.
In order to further analyze the quality grade of Acacia catechu, the possible grades of market Acacia catechu were clarified, and the differences of different grades were evaluated. e 218 peak areas with different retention times from the chromatographic peak matching data of Acacia catechu in the "Similarity Evaluation System for Chromatographic Fingerprint of Traditional Chinese Medicine" were used as variables, and 47 batches of Acacia catechu were analyzed using CA, PCA, and OLPS-DA.

Software.
e software "Similarity Evaluation System for Chromatographic Fingerprint of Traditional Chinese Grayish-brown Earthy yellow (c) Figure 1: e powder color of market Acacia catechu. e samples are S1 sample of domestic Acacia catechu (a), S19 sample of domestic Acacia catechu (b), and S38 sample of imported Acacia catechu (c).

Optimization of Sample Preparation.
Previous literature review revealed that the extraction of Acacia catechu was mostly done by ultrasonic extraction with 50% methanol, but the concentration of the test preparation varied widely from 0.4 to 12 mg/mL [19]. In this study, sample concentrations including 0.4 mg/mL, 8 mg/mL, and 12 mg/mL were prepared for the relevant experiments. e results showed that the fingerprints of most domestic samples were almost a straight line without useful peaks at the concentration of 0.4 mg/mL (Figure 2). At the concentration of 12 mg/mL, most samples had poor separation of catechins, while the test samples were able to obtain relatively reasonable peaks at the concentration of 8 mg/mL. us, a sample concentration of 8 mg/mL was used in the study (Figure 3).

Optimization of Chromatographic
Conditions. e mobile phase compositions, including water-methanol, water-acetonitrile, 0.01% aqueous phosphoric acid-methanol, and 0.05% aqueous phosphoric acid-methanol, were screened for establishing chromatographic fingerprint. e results showed that acetonitrile and methanol had similar elution effects and 0.05% aqueous phosphoric acid could greatly improve the peak shape and separation effect of the chromatographic ngerprint. Finally, 0.05% aqueous phosphoric acid (A)-methanol (B) was utilized as the optimum mobile phase ( Figure 4). Detection wavelengths of 245 nm, 265 nm, 280 nm, and 300 nm were investigated to obtain useful chemical information. It was found that using the wavelength of 265 nm could get a smoother baseline and more detectable peaks. erefore, the wavelength of 265 nm was selected ( Figure 5).

Method Validation of Fingerprint.
e analytical method was validated through precision, repeatability, and stability. One sample was prepared, and the chromatographic ngerprints were recorded in six consecutive injections. e similarity between the ngerprints was calculated and was not less than 0.999, indicating that the instrument has good precision. Six independent samples were prepared in parallel, and the chromatographic ngerprints were recorded. e similarity between the ngerprints was calculated and was not less than 0.997, indicating that the method has good repeatability. One sample was prepared and stored at room temperature for 0, 2, 4, 8, 12, and 24 h for the evaluation of stability. e results showed that the similarity between the ngerprints was calculated and was not less than 1.000, indicating that the samples were stable for 24 h. All in all, the method was suitable for ngerprint analysis.

HPLC-DAD Fingerprint Analysis.
e results of ngerprints analysis are shown in Figures 6 and 7. In the ngerprints, 14 peaks were de ned as common peaks, and peaks 5, 7, 8, and 14 were identi ed as catechin, epicatechin gallate, epicatechin, and quercetin, respectively. e similarities were calculated according to the sample ngerprints and reference ngerprints. e results of the similarity analysis are shown in Table 2. e similarities were in the range of 0.965-0.996 for imported Acacia catechu and 0.111-0.996 for domestic Acacia catechu. Among the domestic Acacia catechu samples, the similarities of nine samples (S16, S15, S20, S21, S10, S14, S7, S8, S12) were lower than 0.900. More unusually, the S9 sample had a similarity of 0.996, demonstrating that the chemical compositions of the batches of imported Acacia catechu were very similar. In contrast, the chemical compositions of domestic Acacia catechu varied signi cantly from batch to batch.

Cluster Analysis.
Cluster analysis is an unsupervised pattern recognition method, which is widely used to classify samples with similar properties into one class, and samples with large di erences in properties into di erent classes [20]. In this study, centroid based clustering was used to perform a systematic cluster analysis with squared Euclidean distance as the measure, and the results are     Journal of Chemistry shown in Figure 8.

Principal Component Analysis.
PCA is one of the most widely used dimensionality reduction algorithms, which is able to extract the main characteristics of things through data processing, greatly reducing the di culty of processing problems [21]. e peak areas of each batch of Acacia catechu at di erent retention times were used as variables, and the data were imported into Unscrambler X 10.4 software. After preprocessing by SNV, PCA was utilized to observe the natural aggregation of the samples, and the score and loading plots of each peak area were obtained. e results showed that the contribution rate of the rst 4 principal components to the original data is 53.579%, 17.479%, 14.764%, and 3.737%, respectively, with a cumulative contribution of 89.558%, and the model had good prediction ability. e PCA score plot of 47 batches of Acacia catechu is shown in Figure 9, which indicates that the discrete degrees of samples S12 and S7 were greater. Combining this with the results of the similarity analysis, we could judge S12 and S7 to be abnormal samples.
Furthermore, the rst-class and second-class samples in CA could be well aggregated, while the ve samples in class 3 (S3, S13, S16, S17, S18) were closer to the second-class samples, and the rest of the samples in class 3 were more discrete and could not be well aggregated into one category. Combining the determination results of principal component groups and index components in the ngerprint of the samples, we could infer that the principal component groups of these 5 samples were consistent with the second-class sample, but the content of chemical components was very low, which could be treated as defective Acacia catechu samples. However, the principal component groups of the other 10 samples in the third-class of samples were significantly di erent from those of the reference herbs, and the content determination results of each component were all low, suggesting that the 10 samples were fakes. e loading plot is based on the distance of the variable from the origin to judge the in uence of the variable on the weight of the principal component. e farther away the variables are from the center, the more they contribute to the weight of the principal component [22]. e PCA loading plot in this study is depicted in Figure 10, indicating that the six variables with greater in uence on the weight of principal components 1 and 2 were peaks 1 (17.402), 2 (15.775, catechin), 3 (22.247, epicatechin), 4 (58.354), 5 (19.572, epicatechin gallate), and 6 (57.622). All the ndings suggested that the PCA model had a good ability to identify class 1 Acacia catechu, while the two chemical components, including catechin and epicatechin, are important indicators for responding to the intrinsic composition of Acacia catechu.

Orthogonal Partial Least Squares Discriminant Analysis.
OPLS-DA is a supervised pattern recognition method based on partial least squares, which is suitable for the case of few numbers of sample observations, many explanatory variables, and presence of multicollinearity and has some advantages in quality control in combination with ngerprinting [23]. In this study, the peak areas of each batch of Acacia catechu at di erent retention times were introduced into SIMCA 13.0 software, where the OPLS-DA model was developed better to analyze the rank classication of market Acacia catechu. e explanatory rate parameter R2X (cum), the model di erentiation parameter R2Y (cum), and the model prediction parameter Q2 (cum) of the data matrix were 0.490, 0.971, and 0.897, respectively. With three new principal component variables, the explanatory capacities of the model to variable X and variable Y were 49% and 97.1%. Furthermore, the fraction of the variation of variable Y predicted by the model based on cross validation was 89.7%, indicating that the established model was stable and had good prediction ability.
e OPLS-DA score is presented in Figure 11, revealing that the sample S38 in class 1 had a large degree of dispersion and the other two classes were better clustered. Variable importance in projection (VIP) values was used as indicator to screen the main contributing chemical components. e higher the VIP value, the greater the contribution of the variable. With VIP >1 as the threshold, the common peaks with great S21 S8 S14 S15 S20 S1 S12 S10 S6 S3 S13 S18 S16 S17 influence on the classification of three classes of market Acacia catechu were screened out. e result of VIP values is given in Figure 12, indicating that there were 98 components with VIP >1 and six compounds, which were the main markers of quality difference between 47 batches of market Acacia catechu, were successively represented by peak 1 (22.247, epicatechin), peak 2 (15.775, catechin), peak 3 (12.247), peak 4 (33.923), peak 5 (14.079), and peak 6 (43.159). e results indicated that catechin and epicatechin were the main differential components in the three classes of market Acacia catechu and had a significant effect on the quality of each batch.

Quantitative Analysis
3.7.1. Method Validation of Quantitative Analysis. 50% methanol stock solutions containing catechins and epicatechins were prepared. By analyzing different concentrations of the two analytes, the calibration curves were generated to determine the contents of catechins and epicatechins. e calibration curves for the two components were y � 6861.3x and y � 7047.2x, and the measurement ranges were 0-1684.00 μg/mL and 0-1404.00 μg/mL, respectively, with good linearity (r 2 > 0.999). e apparatus precision of the two analytes was determined, and their relative standard deviation (RSD) values were less than 2.0% (n � 6), indicating the instrument had good precision. e method's repeatability was assessed, proving the method had good reproducibility with RSD less than 2% (n � 6) for the two analytes. Stability tests were performed by analyzing the same samples stored for 0, 2, 4, 8, 12, and 24 h. e results showed that the RSD values of both components were less than 2.0% (n � 6), indicating that the samples were stable for 24 h. e spiked recovery test was performed to assess the accuracy of the content determination method. e average spiked recoveries of the two analytes were 96.73% and 97.95% with RSD � 1.5% and 1.2%, indicating that the method was accurate. erefore, the above results demonstrated that the method was considered to be accurate for quantitation analysis of market Acacia catechu.

Quantitative Analysis of Market Acacia catechu.
e developed HPLC method was used to determine two compounds, catechin and epicatechin, in the market Acacia catechu from different batches. e contents of catechin and epicatechin, the sum of catechin and epicatechin contents, and the ratio of catechin and epicatechin contents were calculated as four factors for the quality evaluation of market Acacia catechu, and the results of the determination are shown in Table 1. rough the analysis of the origin of the herbs and the content of the two compounds, it was found that the content of the main components of the imported Acacia catechu was higher (only the sample S9 in the domestic Acacia catechu had high content, and the sample S9 could be considered as an outlier), which indicated that the price of imported Acacia catechu in the market was significantly higher than that of domestic Acacia catechu with broad scientific interest and significance. In addition, in terms of analysis of sample powder color, class 1 of samples was earthy yellow, while class 3 of samples was more reddish-brown to red-black, presuming that the earthy yellow samples were high-quality products. However, this was a deviation from the records of "reddish-brown or brown-black" in various ancient books. erefore, the study of sample powder color and intrinsic quality correlation was worth further in-depth exploration. , and the ratio of the contents of the two compounds (D) were from high to low: 1 > 2 > 3, 2 > 1 > 3, 1 > 2 > 3, and 1 > 3 > 2, respectively. To further examine the scienti city of grade classi cation of market Acacia catechu, the OPLS-DA classi cation model was established, using the above four factors as independent variables, and the results are shown in Figure 13.  Figure 14). Among them, both VIPC and VIPA were greater than 1, indicating that these two in uencing factors needed to be given great attention in the quality control of Acacia catechu, which was of great signi cance to the development of Acacia catechu quality standards and could be used as the preferred parameters for the development of Acacia catechu grade standards.

Comprehensive Analysis of the Intrinsic Quality of ree
Classes of Market Acacia catechu. e PCA of the ngerprint found that the rst-and second-class samples could be well aggregated, and the two control samples (S46 and S47) were all aggregated in the second class. Based on the main component groups of the ngerprint, it could be judged that the rst-class samples were consistent with the control samples and the content of the index components was high, indicating that the rst-class samples could be judged to be superior. However, the third-class samples were more discrete, and ve (S3, S13, S16, S17, S18) were closer to the second-class samples. Combining this with CA, we found that the ve samples in class 3 were clustered into a separate category. Further, it could be inferred from the results of the determination of the content of the index components that the main components of the ve samples were consistent with the rst-and second-class samples, but the contents of the index components were extremely low, which could be presumed to be the residual products of Acacia catechu. e ngerprints of the other ten samples in class 3 were di erent from the control samples, and the content of each    component was determined to be low, so it was presumed that the ten samples were counterfeit. us, the third-class samples could be inferred to be substandard samples.

Conclusions
In this study, the established fingerprint and multi-indicator component analysis method were stable and feasible and could be used for the classification and quality evaluation of market Acacia catechu. e results of fingerprint similarity were highly consistent with the results of unsupervised clustering CA and PCA and supervised OPLS-DA, which could systematically classify market Acacia catechu into three classes. e results of the multi-indicator component analysis demonstrated that catechin (A) and the sum of catechin and epicatechin (C) could be used as the preferred indicator parameters for developing quality standards of Acacia catechu grade. All in all, this study provides a new method of Acacia catechu quality evaluation and identification based on fingerprint-chemical pattern recognition, which is an important reference value for the development of Acacia catechu quality standards.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare no conflicts of interest.