During the past decade, mass spectrometry (MS) and its combination with chromatographic separation techniques have emerged as crucial approaches to describe the chemical profiles of PM, which including high performance liquid chromatography-ion trap-mass spectrometry (HPLC-IT-MS) (Sun et al., 2009a), high performance liquid chromatography-electrospray ionization-mass spectrometry (HPLC-ESI-MS) (Yi et al., 2007; Zhao et al., 2013), ultra performance liquid chromatography-quadrupole-time of flight-mass spectrometry (UPLC-Q-TOF-MS) (Lin et al., 2015c; Wang et al., 2017), and high-performance liquid chromatography/ultra high pressure liquid chromatography-linear ion trap-Orbitrap hybrid mass spectrometry (HPLC/UHPLC-LTQ-Orbitrap-MS) (Xu et al., 2012; Qiu et al., 2013; Wang T. H. et al., 2015), and these results suggest that more than 135 compounds were detected and identified by comparison with the standards or through investigating references. It should be noted that 28 new dianthrone glycosides were characterized for the first time by the means of HPLC-LTQ-Orbitrap-MS (Xu et al., 2012), and their fragmentation behaviors were also proposed. This finding enriched the chemical structure types of PM and offered us more comprehensive information. Actually, several novel dianthrone glycosides from PM were elucidated by the conventional phytochemistry methods 4 years later (Yang J. B. et al., 2016; Yang et al., 2017a,b), which confirmed the existence of dianthrone glycosides in PM. Compared with traditional Q-TOF mass spectrometry, UPLC-LTQ-Orbitrap-MSⁿ provides superior resolution and mass accuracy; meanwhile, with the MSⁿ function, the Orbitrap technology can achieve 10 levels of MS analysis capability (Lin et al., 2015d). In this manner, the phenolic constituents were rapidly screened in the roots of PM, and based on the fragment pattern rules of reference stilbenes and anthraquinone derivatives, 59 constituents were characterized or tentatively identified, of which 22 constituents were the first to be reported in PM and 12 compounds were characterized as potential new compounds (Qiu et al., 2013). Table 3 summarizes the chromatographic approaches for the qualitative analysis of PM.

Stilbenes are one of the major components in PM. To date, more than 20 stilbenes and analogs have been found, of which TSG is the most representative compound. Previous publications have demonstrated that TSG possessed anti-tumor, anti-aging and liver-protective bioactivities (Bounda and Feng, 2015; Lin et al., 2015a; Ling and Xu, 2016), which matched well with the traditional efficacies of PM. Quinones are the other characteristic components in PM, which had anti-microbial, anti-cancer, anti-oxidant and anti-human cytomegalovirus effects (Bounda and Feng, 2015; Lin et al., 2015a; Zhou et al., 2015; Dong et al., 2016; Sun et al., 2016). Due to the long conjugated system existing in basic structures of stilbenes and anthraquinones, their characteristic ultraviolet absorptions are easily screened. Therefore, HPLC in tandem with an ultraviolet detector (UV) or diode array detector (DAD)/photodiode array detector (PDA) are widely applied in the quantitative evaluation of PM (Yi et al., 2007; Han et al., 2009, 2013; Jiao and Zuo, 2009; Yan et al., 2010; Zhao et al., 2013; Liang et al., 2014; Li et al., 2016b) According to the Chinese pharmacopeia, TSG, emodin and physcion were eluted on a C₁₈ column by the means of HPLC. However, due to the lack of standard references of anthraquinone glycosides, the content of combined anthraquinone was determined by an indirect method, an additional acidic hydrolysis step was needed, and then, the resulting aglycones were assessed. In this way, combined anthraquinones in PM were calculated as the total amount of physcion and emodin. In fact, determination of the authentic composition has always played a vital role the in quality control of herbals, and with the increasing availability of anthraquinone glycoside standards, direct quantification of the combined anthraquinones was performed by some researchers. Their contents of five markers, i.e., TSG, EMG, PG, emodin and physcion, were quantitatively evaluated by HPLC with DAD (Yi et al., 2007). The results indicated that TSG and EMG were the predominant compounds in PM, which account for about 2.6~4.2% and 0.2~0.6%, respectively, of the total dry weight, and the contents of the other three constituents were no more than 0.06%. Another HPLC-DAD approach (Han et al., 2013) was also proposed for the simultaneous determination of 8 hydrophilic bioactive compounds of PM including TSG, EMG, gallic acid, catechin, epicatechin, hypaphorine, and proanthocyanidin B1 as well as B2, which results in similar data. On the other hand, when analytes are present in trace amounts or showed poor separation, combined MS and liquid chromatographic techniques are the preferred alternative, which provided higher sensitivity and selectivity. ESI-MS in the negative mode was most commonly used in the quantitative analysis of PM (Liang et al., 2011; Zhu et al., 2012; Lin et al., 2015b; Wang T. H. et al., 2015; Luo et al., 2016a). An HPLC-MS/MS method was developed for the simultaneous determination of 14 compounds including stilbenes, quinones, flavonoids and phenolic acids, which might be the work that quantified most compounds in PM. Apart from approaches utilizing HPLC in tandem with UV, DAD, and MS, capillary gas chromatography coupled with flame ionization and mass spectrometric detection (GC-FID-MS) (Zuo et al., 2008) and micellar electrokinetic chromatography (MECK) (Lao et al., 2013; Luo et al., 2015a,b, 2016b) were also established for the determination of stilbenes and anthraquinones in PM. However, due to the complicated protocols and additional derivatization step, GC and MECK might not be substitutes for HPLC as a routine test method. Table 4 summarizes the chromatographic methods for the quantitative analysis of PM.

Besides the stilbenes and quinones, nucleosides and nucleobases also have been determined by MS detection (Luo et al., 2015c,d,e, 2016c,d; Xu et al., 2015).

According to the Chinese pharmacopeia, PM, Polygonum cuspidatum (PC) and Rheum officinale Baill. (RO) are the most frequently used traditional Chinese medicines in the family polygonaceae, which contribute to a wide range of pharmaceutical properties. Due to the similar types of constituents contained in these similar medicinal plants, several studies focused on the quantification and discrimination have been carried out by means of HPLC in tandem with UV or MS (Avula et al., 2007; Huang et al., 2008; Ma et al., 2012; Li et al., 2014; Feng et al., 2016). In HPLC-MS (Huang et al., 2008; Li et al., 2014; Feng et al., 2016), more than 30 compounds had been identified, which mainly belonged to stilbenes, anthraquinones, phenolic acids, and flavonoids. Among these components, thirteen analyte markers including TSG, EMG, emodin, physcion, aloe-emodin, rhein, chrysophanol, piceid, resveratrol, epicatechin, gallic acid, and sennoside A as well as B were simultaneously determined by an HPLC-variable wavelength detection (VWD) approach (Ma et al., 2012). The vital characteristic components for the quality control of a single herb were found by systematic comparison of the chemical compositions of these three herbs, and the results demonstrated that stilbenes and anthraquinones were the main constituents, while chrysophanol, TSG, and piceid could be used as key markers in the discrimination of RO, PM, and PC, respectively. Besides the routine assays mentioned above, HPLC with fluorescence detection (He et al., 2009) and ¹H-NMR approach (Frederich et al., 2011) were also achieved for the quality assessment of these polygonaceous herbs. In addition, the descriptions of macroscopic and microscopic properties also played significant roles in the authentication of the raw materials and their adulteration (Avula et al., 2007; Liang et al., 2011, 2014). Table 5 summarizes the chromatographic methods for the comparative analysis of PM and its related medicinal plants.

On the basis of TCM theory, most of the herbs need to be processed before their clinical usage, and during this procedure, the appearance characteristics and bioactivities of herbs might be changed. Thus, different herbal medicine forms will be selected precisely according to the diagnostics of patients. PM is one typical medicine that has completely different utilities in the crude and processed forms, and especially, the toxicity-attenuating effect of processing has been verified (Yu et al., 2011; Wu et al., 2012; Lin et al., 2015a,e; Wang J. et al., 2015; Li et al., 2016a; Wang et al., 2016; Cui et al., 2017). These fascinating variations have attracted great attention from scientists, and considerable efforts were performed to explore the processing mechanism of PM. Various approaches were applied to monitor the transformation of principal compounds, and the results suggested that the hydrolysis reaction and Maillard reaction were involved in the steaming process of the root of PM (Liu et al., 2009, 2013; Xu et al., 2011; Yu et al., 2011; Chen et al., 2012; Wu et al., 2012; Yang et al., 2015; Zhai et al., 2016; Sun et al., 2017; Zhao et al., 2017). First, the combined anthraquinones were unequivocally hydrolyzed into free ones. Emodin and physcion were the characteristic components in PMP, and the content of emodin was increased by more than 30% after processing, while the content of EMG decreased. Meanwhile, a similar downswing was found regarding the level of TSG, and approximately 60% of TSG was reduced during PMP preparation (Yu et al., 2011; Chen et al., 2012; Wu et al., 2012; Lin et al., 2015e; Yang et al., 2015). It should be noted that when the steaming time was extended, more than 90% of TSG disappeared (Chen et al., 2012). However, the hydrolysis products of TSG could not be detected by most researchers who focused on the rules of PMP processing, and only one publication indicated that a deglycosylated compound had the [TSG+H-Glu+H₂O]⁺ ion at m/z 245.0, which was observed in a direct ionization mass method (Hu et al., 2012). Actually, exploration the products of TSG during the processing is a research project with important scientific significance, which would play a vital role in improving our understanding on the global processing mechanisms of PM. Apart from stilbenes and quinones, the concentration of gallic acid was significantly increased, which might be associated with the hydrolysis of tannin (Chen et al., 2012; Zhai et al., 2016; Zhao et al., 2017). Secondly, the occurrence of a Maillard reaction would likely be responsible for the changes in the contents of 5-HMF, amino acids, sugars, pH and surface color (Liu et al., 2008, 2009). Among these indicators, polysaccharides were closely correlated to the major biological activities of PMP, and an HPLC-evaporative light scattering detection (ELSD) method was proposed for qualification of the sugars in PMP, with the major constituents assigned as glucose, fructose, and sucrose. The contents of D-fructose and sucrose decreased, while the content of D-glucose increased (Liu et al., 2009). In another report, the content of low molecular weight polysaccharides increased after processing (Qiu et al., 2007).

PMP is one representative processed drug that needs repeated steaming. According to the record of ancient writings (Cui et al., 2017), nine cycles of steaming and solarization were required, but in current practice, this traditional method was commonly replaced with steaming once within a few hours. Moreover, PM and its adulterants stained with black dye to simulate PMP without processing were found in the market. It was noted that the pharmacopeia protocols failed to differentiate PM from PMP mainly due to the poor specificity, as the same targets including TSG, emodin and physcion were determined in both of these two medicines. All of these factors would pose a serious health risk. Therefore, the critical markers should be screened and identified in order to distinguish PM and PMP as well as evaluate the quality of PMP. On this occasion, comparative studies of chemical constituents were carried out based on the platform of HPLC-DAD and mass spectrometry (Liang et al., 2010; Liu et al., 2011). 5-HMF was first proposed as a key ingredient to authenticate PMP and PM on account of it being recognized as a product of the Maillard reaction and being newly formed during processing (Liu et al., 2009). However, some scientists found contrary results that 5-HMF could not be observed in PMP in their HPLC analyses (Wu et al., 2012). Furthermore, 5-HMF was a controversial agent due to its toxicity (Severin et al., 2010; Bauer-Marinovic et al., 2012; Islam et al., 2014), as studies had shown that 5-HMF exhibited cytotoxic, genotoxic and tumoral effects. On the other hand, six ingredients, namely catechin, flavanol gallate dimer, polygoninmitin B, emodin-1-O-glucoside, emodin-8-O-(6′-O-malonyl)-glucoside, and physcion-8-O-(6′-O-malonyl)-glucoside, disappeared or decreased significantly after processing, which were assigned as chemical markers for differentiating PM from PMP (Liang et al., 2010; Liu et al., 2011). Nevertheless, the repeatability and reliability of this strategy needs to be validated in further research, which is mainly due to that the amounts of these markers were relatively minor compared to other components in PM before steaming, and the batches of samples investigated in the studies were limited.

As mentioned above, it was difficult to screen the markers only from the perspective of chemical compounds because variations in the plant origins and processing technologies of PM would result in significant differences of index components. In order to assess the quality of PMP, some new approaches were performed based on chemical profiling combined with a bioactivity assay. Chang et al. (2016) conducted an activity-based integrated UHPLC/Q-TOF-MS-FC method to clarify the effect of the processing time on the lipase inhibitory activity of PMP. Chen et al. (2016) developed an online HPLC-DAD-CL assay based on the three reactive oxygen species to evaluate the quality of PMP. In addition, in some other methods, the chemical compositions were analyzed and toxicity monitoring was established to evaluate the processing technologies and quality of PMP (Pang et al., 2014; Ma Z. J. et al., 2015).

Currently, the hepatotoxicity of PM has attracted great concern, and a considerable number of experiments related to PM-induced liver injury were carried out, which offered us comprehensive information to understand the mechanisms. In general, the extracts of PMP are considered to be relatively safe, while hepatotoxicity is found in the PM extracts. Several publications have focused on comparatively studying the toxicities of various extraction solvents from PM and PMP, and the results suggested that the ethanol extract could induce hepatic lesions more easily than that of water decocta (Lv et al., 2013, 2015; Lin et al., 2015e). The order of toxicity was described as PM ethanol extract > PM water extract > PMP ethanol extract > PMP water extract, and in another manuscript (Wu et al., 2012), the toxicity order was proposed as follows: PM water extract > PM acetone extract > PMP acetone extract. Although the hepatotoxic chemicals attributing to the hepatic lesions of PM remain in dispute, emodin and its derivates were believed to be the most likely hepatotoxic components (Yu et al., 2011; Ma J. et al., 2015), and they have also gained much attention from researchers. Recent studies indicated that emodin showed severe cytotoxicity against the human liver cell line L-02 in a concentration- and time-dependent manner. Furthermore, a time-dependent intracellular accumulation of emodin was found in cellular toxicokinetic research by using an HPLC-MS method (Li et al., 2012). Subsequently, an experiment emphasizing the multicomponent interactions of PM was conducted by the same research group, and the results suggested that TSG could delay the elimination of emodin, with the mechanism possibly associated with the inhibition of UGT1A8 mRNA expression (Ma et al., 2013). Lv et al. (2015) explored the toxic components of PM based on biospecific hepatocyte extraction, and the results demonstrated that emodin, physcion, EG and PG were proposed as hepatotoxic components. Lin et al. (2015e) established an UPLC-Q-TOF/MS approach coupled with Progenesis QI and Makerlynx XS software to screen the toxic components from extractions of PM, in which the suspected targets were recognized as emodin-O-(malonyl)-hex, emodin-O-glc, emodin, emodin-8-O-glc, emodin-O-(acetyl)-hex, and emodin-O-hex-sulfate. It was noteworthy that some reports speculated that the toxicity of PM might not be correlated with the content of emodin derivatives but depended on the contents of TSG or the relative content of TSG and emodin (Wu et al., 2012; Yang M. et al., 2016), which is mainly because the amount of emodin was relatively small in PM. Actually, the idiosyncratic hepatotoxicity induced by the isomerization of TSG (cis-TSG) in LPS-treated rats was found in the latest publication (Li et al., 2017), which provided us a new perspective on liver injury by PM. In addition, tannin is another major component in PM, which accounts for approximately 15% of the total dry weight. It was hypothesized that tannin is one of the reasons for the induced liver damage mainly because the content of tannin decreased by 9% after processing (Liu et al., 2005), which meanwhile attenuated the toxicity. In a preliminary investigation, the significant changes of liver biochemical indices were observed after the oral administration of tannin extracts of PM (Hu et al., 2010, 2011). However, in current practice, the chemical analysis approaches regarding tannin are still in the beginning stages, and no individual reference is available due to its complicated structure. Furthermore, the influence of some other factors, i.e., specification (Li Y. M. et al., 2016) and geographical areas (Lin et al., 2017), on the hepatocyte toxicity of PM were discussed. Table 6 summarizes the chromatographic methods for the hepatotoxic analysis of PM in vitro and in vivo.

Exploration of the metabolites plays an important role in clarifying the possible mechanism associated with the liver damage resulting from PM. Lin et al. (2015c) established an U-HPLC-Q-TOF/MS method to describe the absorption and metabolism of PM extract in rat plasm after oral administration, and 16 of 23 compounds were indicated as prototype components of PM, while seven compounds were predicted to be metabolites, which included three isomers of stilbene glucoside glucuronidation, two isomers of emodin glucuronidation, torachrysone glucuronidation, and torachrysone. Through an in vitro study, the metabolism of active compounds of PM was investigated in human normal liver cells (L-02) by means of LC-MS (Liu et al., 2015) and HPLC (Lin et al., 2015), and the results suggested that eight phase II metabolites of TSG and emodin were detected. Their chemical structures were elucidated based on their characteristic fragments, including three isomers of the glucuronidation of TSG, three isomers of the glucuronidation of emodin, one sulfation of emodin, and one emodin-cysteine adduct. In addition, the formation of emodin metabolites mediated by cytochrome P450 was investigated (Qin et al., 2016), and three hydroxylation metabolites named 2-hydroxyemodin, 5-hydroxyemodin, and ω-hydroxyemodin as well as three N-acetyl cysteine conjugates and two emodin-derived GSH conjugates were identified. Among these metabolites, emodin-cysteine was suspected to be associated with liver injury due to the formation of an adduct disturbing GSH and fatty acid metabolism in human liver cells. However, further validation should be carried out to confirm this hypothesis. Due to the multiple phenolic hydroxyl groups in TSG and emodin as well as the limitation of the mass spectrum, the combined positions of glucuronide/ sulfatide/cysteine were still uncertain. In addition, the bioactivities and toxicities of these metabolites need to be evaluated.

In last few years, studies used metabolomics methods integrated with pattern recognition to investigate the potential hepatotoxicity of PM have been constantly reported, which provided preliminary information on the mechanisms of liver injury induced by PM (Dong et al., 2015; Zhang et al., 2015; Li et al., 2016; Zhang C. E. et al., 2016; Ma et al., 2017; Xia et al., 2017). It should be noted that in these experiments, the rats were usually administrated a high dose of PM orally for more than 28 days. In a targeted metabolomics study, the perturbation of nine bile acids (BAs) associated with PM-induced liver injury were evaluated. The glycodeoxycholic acid (GDCA) in bile and hyodeoxycholic acid (HDCA) in serum significantly decreased and were assigned as potential biomarkers for PM-induced liver injury in rats (Dong et al., 2015). In untargeted metabolomics research, 16 possible endogenous metabolites in serum along with 10 metabolites in liver tissue samples were identified by means of GC-MS, and these markers were involved in amino acid, fatty acid, and energy metabolism pathways (Zhang et al., 2015; Xia et al., 2017). Additionally, 16 significantly disturbed biomarkers in urine samples were authenticated by the LC-MS assay, and the pathway analysis showed that vitamin B6 metabolism, tryptophan metabolism and the citrate cycle might be the most important pathways involved in the PM-induced hepatotoxicity (Zhang C. E. et al., 2016). Furthermore, 21 potential metabolomics biomarkers related to the idiosyncratic hepatotoxicity of PM were detected by an UHPLC-MS approach, which is mainly associated with the tricarboxylic acid cycle and sphingolipid metabolism pathways (Li et al., 2016). Table 7 summarizes the chromatographic methods for the metabolomics analysis of PM.

Exogenous contaminants (e.g., mycotoxins, heavy metals, and pesticides) are also considered to be the main reasons for the cause of herbal (Drug-Induced Liver Injury) DILI, especially, the process of steaming would make aflatoxins easily appear if the solarization of PMP was not in time, which usually is known as its acute hepatotoxicity and carcinogenic feature. Several publications focused on the determination of these mycotoxins by means of UHPLC-MS, including aflatoxins B1, B2, G1, G2, M1, and M2 (Han et al., 2010a), ochratoxins A and B (Han et al., 2010b), fumonisins B1, B2, and B3 (Han et al., 2010c), five type B trichothecenes, which contained deoxynivalenol (DON), 3-acetyldeoxynivalenol (3-ADON), 15-ADON, nivalenol and fusarenon X (Han et al., 2010d), zearalenone (ZEN) and its derivatives (Han et al., 2011). It should be noted that the isotope dilution method was employed in all of the experiments mentioned above, which is attributed to the advantage that the isotopic IS had similar behavior to the target during the sample pretreatment and ionization process. Thus, in this way the matrix effects were minimized, and the recoveries were calibrated. The result revealed positive findings of ZEN (1.1 μg/kg) as well as fumonisins B1 (1.25 μg/kg) and B2 (0.82 μg/kg) in randomly selected PM samples, respectively. On the other hand, the heavy metals and inorganic elements were evaluated by using inductively coupled plasma mass spectrometry (ICP-MS) (Luo et al., 2014, 2015f,g) and atomic absorption spectrometry (AAS) (Shi et al., 2011), and the results showed that parts of samples with detected Hg, As and Pb exceeded the safety limits specified by the Green Trade Standards of Importing and Exporting Medicinal Plants and Preparations of China (Pb ≦ 5 μg/g, Cd ≦ 0.3 μg/g, Hg ≦ 0.2 μg/g, Cu ≦ 20 μg/g, and As ≦ 2 μg/g), which should raise significant concerns regarding this issue. In another translational medicine study, five batches of identified PM or PMP were collected from patients with suspected PM DILI, and hazardous materials of these samples, which are comprised of heavy metals, mycotoxins, and pesticides, were determined according to the Chinese pharmacopeia or European Union standards. The laboratory reports demonstrated that there were no targets exceeding the safety limits (Wang J. et al., 2015).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.