Khoa Nguyen Tran
Yeasmin Akter Munni
In-Jun Yang- Department of Physiology, College of Korean Medicine, Dongguk University, Gyeongju, Republic of Korea
Background: The concept of “quality markers” (Q-markers) has emerged as a key solution to address limitations in the evaluation and standardization of traditional herbal medicines. Despite the introduction of various Q-marker identification strategies, methodological inconsistencies and a lack of standardization continue to pose challenges.
Objectives: This review aims to systematically organize and evaluate Q-marker selection strategies published over the past 5 years and propose an optimal approach based on a comparative analysis of their strengths and limitations.
Methods: A comprehensive literature search was performed on the Web of Science and PubMed for studies published between January 2020 and December 2024 using keywords related to Q-marker identification in traditional prescriptions. After removing duplicates and screening for relevance, the eligible studies were systematically reviewed. Key information, including the prescription name, therapeutic targets, methodological steps for Q-marker selection, and the final identified Q-markers, was extracted and organized into summary tables. Based on the analysis, the advantages and limitations of each strategy were evaluated.
Results: The studies were categorized into four representative strategies: [S1] mechanism-driven validation, which relies on network pharmacology and bioassays to align compounds with disease pathways (22 cases, 36.67%); [S2] profile–effect correlation modeling, which uses statistical and machine learning tools to link chemical composition with pharmacodynamic outcomes (24 cases, 40%); [S3] in silico preliminary filtering, which rapidly screens candidate compounds using computational predictions without experimental validation (8 cases, 13.33%); and [S4] multi-criteria decision frameworks, which integrate formulation hierarchy, efficacy, and chemical properties into composite scoring models (6 cases, 10.00%). The average number of Q-markers identified in each strategy was 7.23, 6.61, 8.25, and 7.5, respectively. While each strategy has unique analytical strengths, they often lack consistency and reproducibility when applied in isolation. To overcome this, we recommend a stepwise approach that integrates (1) compound selection based on bioavailability, (2) disease-relevant biomarker selection, (3) correlation modeling, and (4) a multi-criteria scoring framework based on TCM principles. This integrated model accounts for compound bioavailability, specificity, and formulation roles, enabling the identification of functionally relevant Q-markers, including low-abundance constituents.
Conclusion: This review can provide valuable insights to guide future research and development of traditional herbal medicines, particularly in the context of quality control and innovative drug discovery. The proposed framework improves biological relevance and practical applicability and may serve as a scalable model for the quality assessment of multi-component herbal systems and complex pharmacological formulations.
Graphical Abstract.
1 Introduction
The value of traditional herbal prescriptions (THPs) lies in their ability to provide therapeutic benefits through complex natural compounds, many of which have shown potential for integration into modern medical applications. THPs are characterized by inherent complexities, including the composition of multiple compounds and multi-target mechanisms (1). These features pose substantial challenges to achieving consistent quality control and standardization. Furthermore, the composition of THPs can vary significantly due to factors such as soil quality, climate, harvest timing, and postharvest processing techniques (2, 3).
Development strategies for THPs have frequently emphasized high-content bioactive ingredients, primarily due to practical considerations such as ease of analysis, regulatory requirements, established practices, research limitations, and economic factors. This approach often fails to correlate directly with therapeutic efficacy because it overlooks the complex interactions among the various components within a prescription that are essential to its overall effectiveness. To address these limitations, Liu et al. proposed the concept of a quality marker (Q-marker) in 2016 (4). Unlike previous approaches that focused solely on major or bioactive compounds, Q-markers comprehensively consider biological activity, absorption rate, stability, specificity, quantifiability, and compatibility based on the standard principles of traditional medicine theory (5). This innovative concept has opened new horizons in the research and application of traditional herbal medicine, leading to significant progress in quality control, efficacy evaluation, and drug development (6).
Various techniques have been developed to identify Q-markers in THPs. These approaches include computational methods (online databases, predictive tools, and mathematical models) and laboratory testing, such as absorption, stability, and bioactivity (7). The diversity of available techniques has generated multiple strategies for Q-marker determination, each with distinct advantages and limitations. While offering methodological flexibility, this multiplicity of approaches has also introduced significant ambiguities and inconsistencies in the evaluation principles. Currently, the field lacks comprehensive guidelines for selecting the most appropriate and reliable strategies for specific research contexts. Our study systematically reviews and analyzes Q-marker identification approaches published over the past 5 years, thereby suggesting suitable strategies for future studies and contributing to the quality control and standardization of THPs.
2 Materials and methods
This systematic review was conducted according to the updated guidelines of the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) 2020 (8). The Web of Science and PubMed databases were accessed (until January 01, 2025) to search for relevant publications from 2020 to 2024, using the search term: “[(q-marker) OR (quality marker)] AND [(herbal) OR (traditional Chinese medicine) OR (extract) OR (prescription)] NOT (review).”
First, duplicate articles were removed using Microsoft Excel. In the first round of screening, the researchers checked the title and abstract of each article to eliminate records that did not meet the following criteria (1): using traditional prescriptions (2), studying quality markers related to traditional herbal medicines, and (3) specifically providing strategies for selecting quality markers. In the second round, the full text of the remaining articles was evaluated using the following exclusion criteria (1): not specifically mentioning the names of quality markers (2), single-step strategy, and (3) full text not accessible. A single-step strategy indicates studies using only one approach for Q-marker identification, such as only HPLC fingerprinting or only network pharmacology. Studies that combine network pharmacology and molecular docking were classified as “network-pharmacology-only” if they did not include any wet-lab experiments (bioactivity, serum absorption, chromatographic validation). All included studies must employ a multi-step strategy that integrates more than two different analytical procedures, such as computational screening and experimental validation or chromatographic analysis, statistical modeling, and bioactivity testing. Upon confirmation of the final records, the information from each record was extracted into tables, including prescription name, targets, step-by-step strategy for selecting quality markers, and specific final quality markers. All steps in this process were performed by two independent researchers, while a third researcher was consulted in case of disagreement.
3 Results
Using specific search terms, we collected 1,638 records from the Web of Science and 2,234 records from PubMed from January 2020 to December 2024, from which 990 duplicates were eliminated. After screening 2,882 abstracts, we excluded 2,787 abstracts that did not meet our eligibility criteria, including (1) using a prescription (2), studying Q-markers in traditional Chinese medicine (TCM) areas, and (3) specifying a strategy for Q-marker identification. In the full-text evaluation round, we further removed 35 studies belonging to the following categories (1) (1): nonspecific compounds mentioned (2), single-step strategy used, and (3) full text not accessible. Finally, 60 studies were selected for this review (Figure 1).
Figure 1. Flow diagram for the study selection in this systematic review.
Of the 60 selected studies, 19 were published in the first 2 years, followed by 24 studies from 2022 to 2023. The highest number of publications (9) was recorded in 2024 (Figure 2A). As shown in Table 1, these studies, based on their core methodological characteristics, were classified into four strategies: [S1] Mechanism-driven validation strategy (22 cases, accounting for 36.67%); [S2] Profile–Effect correlation strategy (24 cases, 40.00%); [S3] In silico preliminary filtering strategy (8 cases, 13.33%); and [S4] Multi-criteria decision framework strategy (6 cases, 10.00%) (Figure 2B, Table 1). Although some studies involved overlapping features between strategies, the classification was based on the predominant methodology used. Across the four strategies, the average number of Q-markers identified varied. [S1] reported an average of 7.23 Q-markers per study, while [S2] yielded an average of 6.61. [S3] showed the highest average, at 8.25 Q-markers, and [S4] reported an average of 7.50 Q-markers (Figure 2C).
Figure 2. Number of publications regarding (A) Q-marker identification for THPs from 2020 to 2024, together with (B) their classification into different strategies, and (C) the corresponding average Q-marker number as outcomes.
Table 1. Classification criteria for the four research strategies.
3.1 Strategy 1: mechanism-driven validation strategy (22)
A total of 22 studies were classified under this group. Overall, this strategy integrates a wide range of methods to identify Q-markers, including extract composition analysis, metabolomics, network pharmacology, molecular docking, and cell- or animal-based validation experiments. Network pharmacology was employed in all studies, although it was utilized at different stages and for varying purposes depending on the specific research design (Figure 3).
Figure 3. The demonstration of strategy 1: mechanism-driven validation strategy.
3.1.1 Composition analysis
Identifying the compound profiles of herbal extracts was the initial step in any Q-marker determination strategy; however, this step can be performed using different methods. According to the 22 studies, extract composition was obtained through three main approaches (1): direct analysis of the extraction fluid using techniques such as high-performance liquid chromatography (HPLC), ultra-performance liquid chromatography (UPLC), gas chromatography–mass spectrometry (GC–MS), and liquid chromatography–mass spectrometry (LC–MS) (2); indirect identification based on in vivo absorption analysis using biological samples such as urine, plasma, or serum; and (3) referencing compound data from online databases or previously published literature.
HPLC, UPLC, GC–MS, and LC–MS are well-established analytical techniques, each offering distinct advantages in separating, identifying, and characterizing complex chemical mixtures based on their physicochemical properties (10). As shown in Table 1, these techniques are rarely applied in isolation; rather, they are often integrated with advanced detectors or mass spectrometers to enhance sensitivity, resolution, and structural elucidation capabilities. Common combinations include HPLC with diode array detection (HPLC-DAD), ultra-high-performance liquid chromatography (UPLC), or HPLC with quadrupole time-of-flight mass spectrometry (UPLC/HPLC-QToF-MS/MS), and ultra-high performance liquid chromatography coupled with Q-Exactive Orbitrap mass spectrometry (UHPLC-Q-Exactive-Orbitrap-MS). The resulting chromatographic or spectrometric fingerprints served as chemical profiles, reflecting both the qualitative and semi-quantitative compositions of the extract. Each peak in the chromatogram corresponded to a specific compound, enabling comprehensive mapping of the phytochemical profile of the extract.
Metabolomic analysis, frequently encountered in our dataset, was used for profiling small-molecule metabolites in biological samples from animal subjects following a defined period of drug administration. Metabolites extracted from biosamples such as plasma, serum, urine, bile, and feces were commonly analyzed using advanced techniques like UPLC-MS/MS or UHPLC–MS/MS. The resulting complex spectral data were compared against spectral libraries or databases to identify and quantify the target metabolites. Based on the principle that compounds must be absorbed to exert their action, metabolomics was used not only for profiling drug metabolism but also as a screening tool to eliminate compounds with low bioavailability or negligible systemic exposure. For instance, in study A17, an initial chemical analysis of Guchang Zhixie pills identified 198 compounds. However, following metabolomic profiling, only 17 well-absorbed compounds detected in plasma were selected as candidate Q-markers for further investigation.
The final method for primary component analysis involved gathering data from online databases, such as the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP), the Traditional Chinese Medicine Integrated Database (TCMID), Herb Ingredients’ Targets (HIT), and the Bioinformatics Analysis Tool for Molecular Mechanism of TCM (BATMAN-TCM), which integrate vast amounts of information on herbs and their ingredients. Four of the 22 studies in [S1] utilized this approach to acquire the compound profiles of the herbs. For example, the Lianhua Qingwen capsule (study A06) contained 13 medicinal herbs, and data collected from TCMSP and TDT provided a total of 538 compounds. Similarly, study A02 on Da-Cheng-Qi Decoction (DCQD), which includes four crude herbs, identified 272 potential constituents in three databases: TCMID, TCMSP, and HIT.
3.1.2 Network pharmacology
Network pharmacology was widely used in the reviewed studies to explore interactions between compounds and biological networks. Researchers utilized various online databases (Table 2) to identify compounds, predict their molecular targets, and map associated biological pathways (11). The resulting networks enabled visualization of compound–target–pathway interactions and supported the prioritization of potentially bioactive compounds. In these studies, network pharmacology analyses were typically applied at four hierarchical levels.
Table 2. Strategy 1: mechanism-driven validation strategy.
First, the TCMSP, TCMID, and BATMAN-TCM databases were used to filter compounds based on pharmacokinetic parameters, including molecular weight (MW), drug-likeness (DL), blood–brain barrier (BBB) permeability, and oral bioavailability (OB). For example, TCMSP suggests screening thresholds of 180 ≤ MW ≤ 500 Da, DL ≥ 0.1, BBB > −0.3, and OB ≥ 20%. These parameters are often customized depending on the experimental conditions. In study A01, UHPLC-QTOF-MS/MS analysis identified 148 compounds in the Guan-Xin-Jing extract, of which 46 compounds met the adjusted screening criteria (OB > 30%, DL > 0.18) and were retained for further analysis (Zhang et al., 2020a).
Second, network pharmacology was commonly used to identify overlapping targets between herbal compounds and disease-related genes. For example, in study A07, 62 compounds from Xinkeshu tablets (XKST) were linked to 519 predicted targets using SwissTargetPrediction, PubChem, and STITCH. Disease-associated genes for cardiovascular conditions (n = 275) were retrieved from the OMIM, DrugBank, PharmGKB, and KEGG databases. Intersecting these datasets yielded 62 common targets, and 44 compounds were selected based on this overlap (12). A similar approach was taken in study A02, whereby 82 shared targets between the extract’s compounds and the disease were identified, along with 182 corresponding compounds, followed by chemical analysis (LC/MS, GC/MS, and HPLC/SPD), which confirmed the presence of only 25 compounds in the decoction (13).
At the third level, screening was performed through a topological analysis of the compound–target network to identify key nodes and interactions. This approach evaluated network metrics, such as degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC), to identify influential compounds. Degree centrality refers to the number of direct connections that a node has with other nodes. Nodes with a high degree of centrality are considered hubs because of their numerous direct interactions, indicating their potential importance in the network’s immediate connectivity and influence (14). BC measures the extent to which a node lies on the shortest paths between other nodes, acting as a critical connector or bridge within the network, while CC measures how quickly a node can reach all other nodes. Compounds with high centrality values are more likely to modulate disease-relevant biological pathways (14). In the selected cases, Cytoscape software was used for topological analysis, with DC always considered, while the use of other metrics varied. There was no standardized threshold for centrality values, which were typically determined relative to a specific network structure. For instance, studies A02, A03, A12, and A15 selected compounds that exceeded the median values of DC or BC. In contrast, studies C05 and C01 prioritized compounds with the top six and eight DC scores, respectively (9, 13, 15–18).
The final level involved molecular docking, a computational technique used to predict the binding affinity and interaction sites between compounds and target proteins. This method evaluates how strongly a compound (ligand) can bind to a protein (receptor) and is often based on docking scores or binding energies. While molecular docking can be performed independently, it was frequently used in conjunction with network pharmacology to validate compound–target predictions. In [S1], 6 of 22 studies employed docking analyses, all using candidate compounds and targets identified from prior network pharmacology screening. For example, in study A13, 93 compounds from Simiao Yong’an decoction were docked against five predicted target proteins: IL-17A, C3, C5a, VEGFR2, and STAT3, using docking scores to assess the interaction strength. As a result, glycyrrhizic acid showed a stronger predicted binding to IL-17A (−9.2 kcal/mol) than harpagide (−6.8 kcal/mol), supporting its prioritization as a candidate Q-marker. Additionally, because each compound was docked with five receptors, researchers identified the compounds with the lowest average docking scores, along with reference considerations, to select six compounds as Q-markers (19). Similarly, study A14 compared the docking scores of 12 candidate compounds with those of a reference drug (20). In contrast, the docking outcomes were less discriminative in studies A03 and A15. In A03, all five docked compounds had docking scores ≤ − 6.0 kcal/mol and were retained (15). In A15, although one compound (saikosaponin B2) failed to dock with any of the four targets, it was included based on its experimental activity (20).
3.1.3 In vitro/in vivo studies
In vitro and in vivo experiments play critical roles in Q-marker identification by providing direct evidence of compound activity in laboratory environments, including animal bodies, cells, tissues, or biochemical assays. These experimental validations can be integrated at different stages of a Q-marker discovery strategy, depending on specific research objectives.
In [S1], 6 of 22 studies began with laboratory experiments, which included in vivo metabolomics and Caco-2 cell monolayer transport assays. Such early-stage experiments provided insights into the bioavailability, absorption, and pharmacodynamic relevance of candidate compounds, thus helping to prioritize those that were not only detectable in the extract but also biologically active and therapeutically meaningful. Mid-strategy laboratory experiments were conducted in 12 of the 22 studies. At this stage, experiments were performed to confirm initial predictions, explore mechanisms of action, optimize compound selection, and refine candidates before further analysis. Late-stage in vitro or in vivo studies, conducted in 7 studies, served as final validation steps to assess efficacy, safety, and translational potential.
For example, study A08 identified nine Q-marker candidates from Shuangshen Pingfei using UHPLC-ESI-QTOF-MS/MS and network pharmacology. These compounds were tested in vitro using bleomycin-treated macrophages, where all nine compounds reduced the expression of the inflammatory markers CCL2 and CCR2. Although none were excluded, this step provided confirmatory evidence of their anti-inflammatory potential (21). In study A04, 20 potential compounds were identified from Qingzao Jiufei Decoction via UHPLC-ESI-QTOF-MS and network pharmacology, and in vivo pharmacokinetic experiments revealed that two compounds were undetectable, likely due to poor bioavailability. Therefore, only 18 compounds were retained for further analysis (22). An integrative example is shown in study A02, in which 11 compounds were combined into an active compounds combination (ACC) based on their concentrations in the extract. Both in vitro (STC-1 cells) and in vivo (vincristine-induced paralytic ileus in rats) experiments demonstrated that ACC and the original extract had comparable effects in regulating bowel motility, healing, and reducing intestinal bleeding and weight loss. These findings validated the selected Q-markers and demonstrated that ACC could replicate the therapeutic efficacy of the full herbal extract (13).
3.1.4 Literature review
Six of the 22 studies in [S1] incorporated additional literature-based criteria to further refine the selection of Q-markers. These studies included theories on TCM, toxicity profiles, and pharmacokinetic characteristics. This approach ensures that the proposed Q-markers align not only with experimental data but also with clinical relevance and principles of traditional usage. Most studies referred to the “five principles” of Q-marker determination as a widely accepted framework for final selection. These principles include (1): Traceability: Q-markers should be inherent secondary metabolites present in both raw materials and end products, enabling traceability throughout the quality control process (2); Specificity: Q-markers should be characteristic of the medicinal material, ideally unique to its species or closely related taxa, enabling clear differentiation from other herbal materials (3); Effectiveness: Q-markers must have clearly defined chemical structures and known biological or therapeutic activity (4); Measurability: Q-markers must be detectable and quantifiable through established analytical methods (5); Compatibility: Q-markers should align with TCM theory, prioritizing components in “monarch” herbs and considering those in “minister,” “assistant,” and “guide” herbs (5). However, not all studies applied all five principles uniformly. For example, studies A08, B09, B10, and D02 considered only the “compatibility” principle in the final Q-marker selection (21, 23–25). In study A07 with XKST, after a process of composition analysis, network pharmacology, plasma metabolomics, and validation experiments, nine components (danshensu, salvianolic acid A, salvianolic acid B, salvianolic acid D, ononin, quinic acid, puerarin, daidzein, and biochanin A) were included in the list of candidate Q-markers. Upon applying the “five principles,” salvianolic acid D, ononin, quinic acid, and biochanin A were excluded due to their low concentrations, which rendered them non-quantifiable. Danshensu, salvianolic acid B, and salvianolic acid A belong to the king herb of XKST, while puerarin and daidzein belong to the minister herb of XKST. Therefore, only these compounds were selected as the final Q-markers for XKST (12).
3.2 Strategy 2: profile–effect correlation strategy (24)
Compared to [S1], the studies categorized under [S2] have also implemented fundamental steps, such as component analysis and laboratory experiments. However, network pharmacology was not utilized; instead, chemometric and mathematical analysis methods were emphasized for Q-marker identification. Upon detailed examination, the 24 studies in [S2] can be grouped into sub-strategies (1): the spectrum-bioactivity relationship approach (SBR) (8 cases, 33.4%) (2), the correlation between marker metabolites and serum components (PCMS) model (11 cases, 45.8%) (3), the artificial neural network (ANN) model (3 cases, 12.5%), and (4) others (2 cases, 8.3%) (Figure 4). The SBR group includes studies using chromatographic fingerprint peaks as the X variable and biological activity measurements as the Y variable, followed by statistical correlation analysis. Studies classified into the PCMS group use absorbed serum/urine components as the X variable and disease biomarkers or metabolomic changes as the Y variable, typically involving metabolomic profiling and correlation coefficient calculation. The ANN group includes studies using machine learning architecture with an input layer (chemical/chromatographic data), hidden layer (s), and output layer (biological responses), employing back-propagation or similar training algorithms. Although these sub-strategies differ in strategic detail, they require input from composition analyses and experimental data to evaluate the correlation between the extract components and therapeutic outcomes or metabolomic biomarkers. Based on these correlations, the compounds with the strongest associations were selected as Q-markers.
Figure 4. The demonstration of strategy 2: profile–effect correlation strategy.
3.2.1 Spectrum–bioactivity relationship strategy
This strategy integrates chromatographic fingerprinting with bioactivity evaluation to correlate specific chemical components of an herbal extract with its therapeutic effects. As shown in Table 3, all eight studies began with analytical techniques such as HPLC, UPLC, UHPLC, and two-dimensional liquid chromatography (2DLC) coupled with LTQ Orbitrap Mass Spectrometry (LTQ-Orbitrap-MS) to provide the chemical profile of the extracts, serving as the first input for SBR analysis. Subsequently, in vitro and in vivo assays were conducted to evaluate the biological activities of the extracts, including anti-inflammatory, antioxidant, and antimicrobial effects. The raw data from these experiments served as the second input for the SBR analysis. Statistical methods, such as Pearson correlation analysis or Grey correlation analysis (GCA), were then applied to identify the chromatographic peaks significantly associated with the observed bioactivities. Peaks with a high correlation coefficient (r) were considered potential Q-markers. Additionally, Partial Least Squares (PLS) models and their variants, including partial least squares regression (PLSR) or orthogonal partial least squares discriminant analysis (OPLS-DA), were sometimes used (5/8 cases) as supportive methods. These multivariate statistical techniques model the relationships between sets of variables. PLS constructs latent variables that maximize the covariance between predictor variables (X) and response variables (Y), enabling regression in PLSR and classification in OPLS-DA (26). Compounds with variable importance in projection (VIP) scores > 1 and t-tests indicating significant intergroup changes (p < 0.05) were identified as influential in the observed effects and, therefore, potential Q-markers. Finally, validation experiments such as in vitro/in vivo tests, quantitative analysis, or molecular docking were conducted (5/8 cases) to further confirm the Q-marker list.
Table 3. Strategy 2: profile–effect correlation strategy.
For instance, in study B06, HPLC was used to generate chromatographic fingerprint profiles from 12 batches of Kai-Xin-San (KXS) extracts, identifying 25 common peaks with good consistency. These peaks were used as the first input variables for SBR. The antioxidant activity of each batch was evaluated using the DPPH free radical scavenging assay, and the resulting data served as the second input. Pearson’s correlation coefficients were calculated to determine the linear relationships between the peaks in the HPLC chromatograms and the antioxidant activities of the different batches. Thirteen peaks were positively correlated with DPPH free radical scavenging capacity (r > 0.7). In addition, OPLS-DA was applied using the same variables and revealed another 13 peaks with VIP values > 1. Finally, seven common peaks from both analyses were selected as Q-markers for KXS (27).
3.2.2 Plotting of correlation between marker metabolites and serum components model
The PCMS model is a correlation-based analytical approach that evaluates the relationship between components absorbed into the bloodstream following administration and disease-related biomarkers. All 11 studies in this sub-strategy began with urine or serum metabolomic analyses to obtain profiles of the bioavailable herbal components and lists of potential biomarkers relevant to the disease under investigation. Here, principal component analysis (PCA) and the OPLS-DA mathematical model were used as pattern recognition tools. PCA reduces high-dimensional data into a smaller number of principal components, while OPLS-DA maximizes the separation between experimental groups by filtering out variability unrelated to the phenotype. This approach helps identify which absorbed components and biomarkers contribute most strongly to group discrimination (28). Subsequently, Pearson correlation analysis was used to calculate the correlation coefficient (r) between the serum components and marker metabolites, with higher correlation coefficients indicating stronger relationships (29). From this, lists of high-potential Q-markers and related biomarkers can be determined. For the validation step, 6/11 studies relied on literature reviews and TCM principles such as compatibility, specificity, and traceability to obtain the final Q-marker list, while 4/11 studies performed in vitro tests such as quantitative analysis, real-time qPCR, cell viability, and western blot.
In study B13, after establishing the abnormal uterine bleeding (AUB) model and treating with Baoyin Jian (BYJ), urine and serum samples were collected from control, model, and BYJ-treated rats. Urine metabolomic data were first analyzed using PCA, which revealed clear clustering within groups and notable separation between the control and model groups, indicating substantial metabolic alterations in the disease state. OPLS-DA was subsequently applied, and 32 ions with VIP scores > 1 and p < 0.05 (t-test) were identified as potential disease biomarkers. These metabolites served as the first variables in the PCMS model. In parallel, 59 serum components from BYJ-treated rats were characterized using the UPLC-G2-Si/MSE system combined with UNIFI software, representing the second variable. Pearson’s correlation analysis was used to calculate the correlation coefficients between the 59 serum compounds and the 32 biomarkers. Seven of the 59 compounds with _r_ > 0.7 and at least five biomarkers were ultimately identified as Q-markers of BYJ (30). In study B19, the Q-markers of Qifu decoction (QFD) for heart failure treatment were identified using a similar approach, with minor modifications in the biomarker selection strategy. A metabolomic analysis identified 21 serum biomarkers associated with heart failure, and only 14 that were significantly reversed by QFD treatment were selected as the initial input for PCMS. Simultaneously, 24 QFD-derived constituents were detected in the serum of the treated rats using UHPLC-Q-TOF-MS combined with shared and unique structures (SUS) plot analysis, which served as the second input. From Pearson correlation analysis, 11 compounds exhibiting strong correlations (|r| ≥ 0.6) were identified as potential bioactive components. Subsequent in vitro assays for cardioprotective activity confirmed that neoline and calycosin were potential Q-markers of QFD.
3.2.3 Back-propagation artificial neural network model
ANNs are computational models inspired by the architecture of the human brain, which predict outcomes (outputs) based on input variables (predictors) using various training algorithms. In the reviewed studies, the back-propagation (BP) algorithm was the most commonly employed method for training ANN models (31). All cases within this sub-strategy specifically used the back-propagation artificial neural network (BP-ANN) approach.
A BP-ANN comprises an input layer, one or more hidden layers, and an output layer. Each layer consists of a set of interconnected nodes (neurons) linked by weights. The input layer receives raw data, which are then passed through the hidden layers. In these layers, each node calculates the weighted sum of its inputs and applies an activation function to introduce nonlinearity. The resulting information is then transmitted to the output layer. If the predicted output deviates from the actual result, the model enters the back-propagation phase, where the connection weights are iteratively adjusted to minimize the error between the predicted and observed outcomes. This training process continues until the network reaches an acceptable level of accuracy (32, 33). From the selected studies, ANN models required inputs of characteristic peak area data from herbal extracts (independent variables) and the corresponding bioactivity measurements (dependent variables). These data were used to compute specific indices, such as the degree of influence (ID) or mean impact values (MIVs), which contribute to the evaluation of the bioactive components. Finally, quantitative analyses such as UPLC-MS/MS or HPLC-UV were performed (in 2/3 cases) on the suggested compounds for reconfirmation.
For instance, in study B20, 18 chromatographic peaks from the herbal pair Chuanxiong Rhizoma and Cyperi Rhizoma extract (CRCR) were identified using HPLC-ESI-Q-TOF-MS/MS. Subsequently, the pharmacological effects of CRCR were evaluated in a migraine rat model by measuring six biomarkers: 5-hydroxytryptamine, calcitonin gene-related peptide, β-endorphin, vasoactive intestinal peptide, and nitric oxide synthase. A BP-ANN model was constructed with an input layer of 18 neurons (corresponding to the 18 characteristic peaks), a hidden layer of eight neurons, and an output layer comprising six neurons (representing the six pharmacological indicators). A sensitivity analysis was applied to compute the ID value, where variables with ID > 5% were considered to have significant effects on the pharmacological outcomes. Based on these criteria, 18 compounds were identified as candidate Q-markers. Additionally, a PLSR model was employed to further explore the correlation between chromatographic peaks and pharmacodynamic markers. This analysis identified 12 peaks with positive contributions across all six biomarkers. Finally, the eight compounds commonly identified by both the ANN and PLSR models were designated as Q-markers for CRCR (34). In study B22, a similar strategy was used to investigate the chemical profile and bioactivity of Jinqi Jiangtang. Instead of the PLSR model, a ReliefF-based feature selection algorithm was used to pre-select 18 representative compounds with potential antidiabetic activity. These compounds were then used as input variables in the ANN model, which consisted of 18 neurons in the input layer (peak intensities), 7 neurons in the hidden layer, and 3 neurons in the output layer (representing bioactivities). The MIV analysis revealed that 10 of the 18 preselected compounds with summed MIV > 0 were identified as potential Q-markers for Jinqi Jiangtang (35).
3.3 Strategy 3: in silico preliminary filtering strategy (8)
As shown in Table 4, the studies classified under [S3] adopted relatively simplified strategies, typically involving only two to three methodological steps. These approaches primarily relied on the in vitro determination of compound profiles, online databases, and computational tools for Q-marker identification without incorporating any validation experiments using cell lines or animal models. Analytical techniques, such as HPLC-MS or UPLC-MS, were commonly used to identify the chemical constituents (8 cases), which were subsequently analyzed through network pharmacology (7 cases), molecular docking (2 cases), or chemometric analyses, such as PCA and OPLS-DA (3 cases).
Table 4. Strategy 3: in silico preliminary filtering strategy.
3.4 Strategy 4: multi-criteria decision framework strategy (6)
This strategy follows the concept of the “five principles” of TCM (Traceability, Compatibility, Efficacy, Specificity, and Measurability) to determine Q-markers (5). This model implemented appropriate methodologies to generate datasets representing the essential properties of a Q-marker. In this model, each data set corresponded to one “variable” (dimension). After gathering the data for each dimension, the values were normalized to a “0–1 scale” or “1–4 scale” to mitigate the analytical variations caused by the disparity in the criteria of different dimensions. These normalized values were then integrated into a “multi-dimensional network,” in which the points corresponding to the different variables were interconnected. A regression analysis was then applied to compute a characteristic index for the network, such as the regression area (RA) or shaded area (QMI). Compounds with higher index values were considered to contribute more significantly to the therapeutic effects of the extract and were selected as Q-markers (Figure 5).
Figure 5. The demonstration of strategy 4: multi-criteria decision framework strategy.
As shown in Table 5, the “content dimension” can be interpreted as representing both the measurability and traceability principles in Q-marker selection. For instance, in study D01, the content dimension was established directly through the quantitative analysis of compounds across multiple batches using UPLC-PDA (36). In contrast, study D03 integrated both metabolic and chemical profiling, selecting 19 compounds shown to enter the systemic circulation after administration, with their concentrations quantified in the prescription using UHPLC-Q/TOF-MS/MS analysis (37). Notably, although the “stability dimension” is not formally included in the “five principles” of TCM Q-marker selection, it was utilized in 4 out of 6 reviewed studies to construct multi-dimensional networks. This was based on the rationale that a stable Q-marker would strengthen the application of other principles and support the efficacy of the entire extract. In these studies, stability was assessed by measuring changes in the relative content of compounds under stress conditions, such as strong light, high temperature, and high humidity. However, inconsistencies have been observed in the interpretation of this attribute. While studies D01 and D04 prioritized highly stable compounds, D02 selected compounds exhibiting larger stability fluctuations, arguing that such instability may better reflect the intrinsic quality of a prescription (25, 36, 38). The “compatibility dimension” categorizes each compound according to its role in the TCM prescription hierarchy, namely, King, Minister, Assistant, or Guide herbs. King herbs target the primary disease and provide the main therapeutic action; minister herbs assist in enhancing or broadening this effect; assistant herbs help balance the formula or reduce side effects; and guide herbs direct the formula’s action to specific organs or channels, collectively contributing to a harmonized and effective formulation (39). The “efficacy dimension” evaluates the biological activity of candidate compounds, such as enzyme inhibition or molecular target binding. As shown in Table 5, five of six studies employed in vitro assays, while only study D05 utilized an in vivo model to assess the analgesic effects of the candidate compounds in mice (40). Furthermore, studies D02, D03, and D05 incorporated topological node degree values and molecular docking-derived binding affinities to enhance efficacy assessments (25, 37). Lastly, the “specificity dimension” was addressed only in study D05. To obtain such data, the biosynthetic pathways of the Q-marker candidates were analyzed, and information on the plant families to which each candidate compound belonged was collected. Compounds occurring in fewer plant families were considered more specific and were assigned correspondingly higher converted scores, in alignment with the principle that chemical specificity enhances the authenticity and traceability of herbal sources (40).
Table 5. Strategy 4: multi-criteria decision framework strategy.
For instance, study D03 aimed to identify Q-markers in Qiliqiangxin capsules (QLQX) for the treatment of chronic heart failure. Based on a previous in vivo metabolomics investigation, 19 compounds were selected for their presence in the systemic circulation following administration. The “content dimension” was established by quantifying the concentrations of these 19 compounds in QLQX using UHPLC-Q/TOF-MS/MS, followed by normalization to a 4-point scale using a mathematical formula that accounted for the maximum and minimum values in the dataset. Sinapine thiocyanate, which had the highest average content (2363.13 μg/g), was assigned a score of 4, while ginsenoside Rb1 (286.75 μg/g) was normalized to 1.36. Subsequently, the “compatibility dimension” was quantified by assigning values according to TCM prescription hierarchy: compounds derived from king herbs were scored as 4, while those from minister, assistant, and guide herbs received scores of 3, 2, and 1, respectively. For example, formononetin from the king herb Astragali radix was assigned a value of 4, whereas alisol A from the assistant herb Alismatis rhizoma was assigned a value of 2. The “efficacy dimension” was evaluated via three factors. The anti-heart failure effects of the compounds were assessed using an isoprenaline-induced myocardial injury model in rat H9c2 cardiomyocytes, and their relative activity was normalized against the effective dose of QLQX. A topological analysis was also performed to calculate the DC of each compound in the compound–target interaction network of QLQX, along with their in vivo-predicted bioavailability retrieved through name-based queries in the TCMSP database. All dimensional values were normalized to a scale of 1–4, and the regression area (RA) and coefficient of variation (CV) were calculated for each compound. As a result, 7 compounds with RA ≥ 0.8× and CV ≤ 1.2× were selected as Q-markers of QLQX (37).
4 Discussion
Over the past 5 years, research on Q-markers in THPs has increased rapidly; however, methodological coherence remains lacking. Despite the growing number of studies, the field remains fragmented, with diverse strategies developed in isolation and lacking a unified conceptual framework to guide comparative evaluation (41). In comparison with previously published review articles on Q-marker identification, the present study offers several notable advances in scope and methodological rigor. First, whereas earlier works mainly adopt a narrative review format, our study represents the first systematic review of Q-marker research conducted according to PRISMA 2020 guidelines, allowing structured extraction, comparison, and evaluation of methodological details across studies. Second, by analyzing 60 original articles published between January 2020 and December 2024, this review provides an updated synthesis that captures new concepts and advanced analytical approaches that were not comprehensively discussed in previous reviews, including modern computational and statistical models such as similarity analysis (SA, HCA, PCA), correlation analysis (BCA, CCA, GRA, Pearson correlation), and advanced computational models like PLSR, OPLS-DA, and BP-ANN. Third, unlike prior reviews that focus narrowly on a single analytical domain (e.g., chromatographic analysis or network pharmacology), our work integrates all major analytical approaches currently used in Q-marker discovery and organizes them into a four-strategy classification system that reflects actual research practice. Finally, by critically examining the strengths and limitations of each strategy, this review proposes a refined and more operational framework for future Q-marker identification, offering clearer methodological guidance than has been available to date.
In this review, we classified 60 studies into four representative strategic models: [S1] mechanism-driven validation, [S2] profile–effect correlation modeling, [S3] in silico preliminary filtering, and [S4] multi-criteria decision frameworks. Strategies [S1], [S2], and [S3] typically follow a sequential logic, beginning with compound profiling, followed by computational or statistical filtering to narrow down candidate compounds. In contrast, [S4] adopts a concurrent evaluation model that integrates multiple criteria, such as bioavailability, efficacy, and prescription hierarchy, from the outset. Although common techniques such as network pharmacology, metabolomics, and experimental validation are used across all strategies, their specific roles and analytical weights differ substantially. More critically, no study has resolved the conflicting Q-marker results that emerge when different methods are applied. The absence of methodological interoperability is a central limitation of the current literature. To move forward, it is essential to reassess the internal logic of each strategy as well as its structural compatibility for future integrative models.
The mechanism-driven validation strategy ([S1]) exemplified the recurring structural inconsistencies observed in Q-marker research. This approach typically uses network pharmacology as an initial screening tool, followed by experimental assays to validate predicted compound–target interactions. Although several studies have reported biologically meaningful outcomes using this framework (42, 43), its overall methodological structure lacks consistency. Screening parameters, such as network centrality scores, oral bioavailability thresholds, and compound–target overlap ratios, are applied inconsistently across studies and are rarely grounded in statistically defined or empirically validated benchmarks (11, 44). Furthermore, most network pharmacology analyses rely on static network architectures derived from public databases, which cannot capture dynamic biological changes, such as time-dependent gene expression, post-translational modifications, or disease progression (45). Importantly, many databases are updated slowly and may not incorporate newly characterized compounds or targets, reducing their relevance for contemporary analyses. Moreover, newer databases or updated versions often build directly upon older resources, inheriting outdated or redundant information (46, 47). Because literature-derived entries frequently contain false positives or false negatives, researchers must critically evaluate data provenance and reliability when interpreting network pharmacology outputs. A further methodological concern, related to molecular docking, is the absence of a universal docking score cutoff to define “strong,” “medium,” or “weak” binding (15, 19). Studies apply widely varying thresholds, often based on convenience or precedent rather than validated correlations with experimental affinity. This inconsistency, combined with known variability between docking tools, limits the interpretability of docking results and highlights the need for experimental validation. The positioning of experimental validation varies; some studies incorporate it as an early filter for refining candidates, whereas others treat it as a downstream confirmation step detached from the screening logic. This asymmetry between prediction and validation undermines the internal coherence of [S1] and limits its reproducibility. To enhance the reliability of this strategy, future studies must establish clearer functional workflows and adopt standardized thresholds based on biological and pharmacological relevance.
The profile–effect correlation strategy ([S2]) offers a distinct contrast to the mechanism-focused strategy [S1] by emphasizing data-driven inference. It applies statistical and machine learning techniques to identify quantitative associations between chemical composition and pharmacological outcomes. Within this model, sub-strategies such as SBR, BP-ANN, and PCMS differ in their analytical structures. SBR typically uses PCA and PLS-DA to capture linear trends, whereas BP-ANN models nonlinear patterns through iterative training on labeled datasets (33, 48). Although a BP-ANN may yield a larger number of candidate markers, this does not guarantee higher accuracy, and the risks of overfitting and interpretability limitations remain significant (49, 50). Moreover, reliance on peak intensity data may bias the results toward high-abundance compounds, neglecting low-concentration constituents with pharmacological relevance (25). The PCMS addresses these issues by focusing on bioavailable serum compounds and correlating them with disease-related metabolic biomarkers to improve mechanistic plausibility. However, it may overlook active constituents below detection thresholds or those unrelated to the selected biomarkers. It is also worth noting the inconsistency in how different studies define correlation strength. Some studies categorized 0.6 < |r| < 0.7 as “high correlation” and |r| ≥ 0.7 as “extremely high correlation”; in other cases, 0.7 ≤ |r| < 0.8 indicated “high correlation” and 0.8 ≤ |r| ≤ 1 indicated “extremely high correlation” (51, 52). Thus, the predictive capacity of [S2] depends on model architecture as well as experimental design elements such as compound exposure, concentration bias, and biomarker specificity. To increase its reliability, [S2] may benefit from integrating multi-criteria frameworks, such as [S4], which account for both chemical relevance and biological context, thereby enabling a more balanced evaluation of Q-marker.
The in silico preliminary filtering strategy ([S3]) relied primarily on compound screening based on database mining, drug-likeness scoring, and molecular docking simulations. Although this model allows for the rapid triage of large chemical libraries with minimal experimental burden, its methodological simplicity often limits its strategic value. In contrast to [S1], which incorporates biological validation, and [S2], which relies on statistical correlation with experimental data, [S3] typically concludes at the computational prediction stage. Without empirical feedback, the functional relevance of predicted Q-marker candidates remains speculative. Furthermore, filtering thresholds such as docking scores or ADME criteria are often applied inconsistently across studies and are rarely justified by standardized benchmarks. Most critically, [S3] fails to address key pharmacological dimensions, such as systemic bioavailability, metabolic transformation, or multi-compound interactions inherent to traditional herbal formulations. These limitations hinder its use as an independent identification model. Instead, [S3] may be best positioned as an auxiliary pre-screening step within a broader analytical pipeline. Its operational speed can support early-stage filtering, but robust Q-marker evaluation ultimately requires integration with experimental validation, multi-criteria frameworks, and concepts more fully developed in strategy [S4].
The multi-criteria decision framework strategy ([S4]) adopts a structured scoring system to evaluate candidate Q-markers across multiple dimensions, including biological efficacy, content measurability, formulation role, chemical stability, and compound specificity. In contrast to single-criterion approaches, [S4] quantitatively integrates these variables, enabling a composite evaluation that reflects the overall functional relevance of each compound. Notably, the “efficacy” dimension incorporates experimental data from in vitro and in vivo studies, allowing for the inclusion of low-abundance but pharmacologically potent constituents (25, 53). In addition, [S4] is uniquely capable of capturing not only the direct effects of individual compounds but also their synergistic or supportive functions within the broader prescription matrix (54). For example, in the evaluation of Shenqi Jiangtang granules, astragaloside IV was selected as the final Q-marker because of its high efficacy score (28.31) despite a low content score (1.8), whereas ginsenosides Rh1 and Rb3, although abundant, were excluded due to insufficient bioactivity (25). Despite its strengths, [S4] faces several implementation challenges. The scoring criteria for each dimension are not standardized and often rely on literature-based or database-derived proxies instead of direct experimental validation. This leads to variability in scoring interpretations and weightings across studies (25, 36). The “compatibility” dimension, for instance, is frequently defined subjectively, while the “stability” dimension has been interpreted inconsistently; some studies prioritize stable compounds, whereas others emphasize instability as a proxy for formulation sensitivity. In fact, “stability” is a crucial dimension in Q-marker selection. According to the WHO standards, quality markers should remain chemically stable under defined storage conditions to ensure they function as reliable indicators (55). Unstable markers undermine batch-to-batch consistency, a core requirement for pharmaceutical quality control, and complicate manufacturing, storage, and shelf-life determination. Moreover, from a safety standpoint, chemically unstable constituents may degrade into reactive or even toxic metabolites under stress conditions. Additionally, [S4] demonstrates substantial value when used in combination with predictive models. In a study on Qiliqiangxin capsules, candidate compounds were initially screened using pharmacological criteria from [S2] and subsequently re-evaluated using [S4] scoring metrics, such as content, compatibility, and efficacy (37). This sequential integration allows researchers to rediscover low-abundance but functionally important compounds that would have been excluded by content-based screening alone.
Analysis of the average number of Q-markers identified across strategies revealed notable methodological differences. In silico preliminary filtering ([S3]) yielded the highest average number of Q-markers (8.25), likely reflecting its broad, hypothesis-generating nature and reliance on large-scale computational screening rather than empirical validation. In contrast, the profile–effect correlation strategy ([S2]) produced the lowest average number (6.61), consistent with its emphasis on statistically robust correlations that may filter out weaker or less consistent candidates. The mechanism-driven validation strategy [(S1)] showed a moderately high average (7.23), balancing biological plausibility with experimental confirmation. Similarly, the multi-criteria decision framework [(S4)] identified a comparable number of Q-markers (7.50) despite its more complex scoring and weighting system. Collectively, these patterns indicate that computationally oriented approaches tend to generate a broader set of putative markers, while experimentally grounded or statistically stringent methods yield more selective but potentially more reliable Q-marker sets.
Based on the strengths and limitations identified across the current strategies (Table 6), we suggest an integrative strategy that future studies can benefit from through Q-marker identification to ensure both reliability and practicality (Figure 6). This strategy should begin with (1) the selection of candidate compounds detected in the serum or urine following in vivo administration to ensure bioavailability, followed by (2) the identification of disease-relevant biomarkers (3). PCMS modeling can then be used to evaluate compound–biomarker correlations, screening out important compounds and biomarkers (4). Finally, a multi-dimensional scoring system incorporating key TCM principles can be implemented: measurability (via analytical methods such as HPLC, UPLC-MS/MS), efficacy (via biological activity in in vitro/in vivo tests and molecular docking scores with candidate biomarkers), compatibility (via ranking of the original herb as King, Minister, Assistant, or Guide herb), and stability (via checking under strong light, high temperature, and high humidity). Study B10 can serve as a close example for the proposed strategy, which identified Q-markers of the Wenxin formula (WXF) for the treatment of cardiac disease (24). First, the authors focused on bioavailable components by identifying 37 WXF-derived compounds detected in serum after administration. Next, using UPLC-HDMS, 25 metabolites were identified as biomarkers for myocardial ischamia (MI). They then performed PCMS modeling with Pearson correlation analysis between serum constituents and MI-associated biomarkers, thereby identifying 8 active constituents highly related to the therapeutic effect. Finally, although not expressed as a formal scoring system, B10 implicitly evaluated these candidates according to some key TCM principles; for example, berberrubine-o-glucuronide was excluded because it is a compound generated by internal metabolism and is not an inherent component of the crude drugs. Ginsenoside Rb1, cinnamic acid, paeoniflorin, and berberine were selected as the Q-markers based on their traceability back to the crude drugs and their reflection of the compatibility principle of WXF’s formulation (representing the sovereign, minister, and assistant drugs). In fact, our proposed procedure is quite similar to that of the studies classified into the PCMS group of [S2] in this review, but in selecting the final Q-marker list based on TCM principles, we emphasize that the experimental methods should receive greater attention and be weighted equally using specific quantitative measures.
Table 6. Summary of the characteristics, advantages, and disadvantages of the four strategies.
Figure 6. The demonstration of the proposed integrative strategy for Q-marker identification.
This proposed strategy would address the major limitations of existing models by minimizing bias toward high-abundance compounds, ensuring biological relevance, and integrating data-driven predictions with experimental verification. As highlighted in the 2020 edition of the Chinese Pharmacopoeia and WHO standards, quality markers should reflect both chemical consistency and biological efficacy. Accordingly, the proposed framework aligns closely with international regulatory guidelines for herbal medicine quality control (55, 56). In particular, the bioavailability criterion ensures that selected Q-markers are not only present in raw materials but are also systemically absorbed, addressing the emphasis on transitivity and traceability throughout the production chain, from raw materials to finished products. Moreover, the multidimensional scoring system corresponds directly with the “five principles” framework advocated by Chinese regulatory authorities. From an industrial perspective, this framework offers systematic marker selection with regulatory-compliant documentation, provides robust evidence of biological activity, and facilitates traceability across all stages of production.
Notably, all 60 studies included in this review were conducted by research groups based in China, indicating that English-language databases effectively capture mainstream TCM research. However, as our search strategy was restricted to English-language articles, the potential consequences of excluding Chinese literature may include omitting specific methodological details available only in local databases. Therefore, future reviews should expand the scope to regional databases and additional languages to ensure a more comprehensive evaluation.
5 Conclusion
In conclusion, the 60 eligible studies published between 2020 and 2024 demonstrated increasing research activity, with the highest output in 2024. Four predominant methodological strategies were identified, with profile–effect correlation (S2) and mechanism-driven validation (S1) representing the majority of cases, followed by in silico preliminary filtering (S3) and multi-criteria decision frameworks (S4). The number of Q-markers identified varied across these strategies, ranging from an average of 6.61 Q-markers in S2 to 8.25 in S3, indicating clear quantitative differences in Q-marker outputs among the methodological approaches.
Although each strategy offers specific strengths, reliance on single, isolated methods often leads to inconsistent selection criteria and limited reproducibility. To advance Q-marker research, future studies should adopt more coherent, stepwise frameworks that integrate prediction, validation, and standardization. We therefore propose an integrated strategy combining (1) bioavailability-based compound selection (2), disease-relevant biomarker identification (3), correlation modeling, and (4) a multi-dimensional scoring system grounded in key TCM principles. This approach would provide a practical and reproducible foundation for improving Q-marker discovery and supporting quality control in traditional medicine.
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Author contributions
KT: Conceptualization, Methodology, Investigation, Data curation, Visualization, Writing – original draft, Writing – review & editing. GM: Data curation, Investigation, Visualization, Writing – review & editing. YM: Writing – review & editing. I-JY: Conceptualization, Supervision, Writing – review & editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This research was funded by the Dongguk University Research Fund (2023), National Research Foundation (NRF-2021R1I1B0748979), and Korean Health Technology R&D Project of the Korean Ministry of Health and Welfare (RS-2022-KH127455 and RS-2024-00443739).
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed.2025.1709969/full#supplementary-material
References
1. Wang, H, Chen, Y, Wang, L, Liu, Q, Yang, S, and Wang, C. Advancing herbal medicine: enhancing product quality and safety through robust quality control practices. Front Pharmacol. (2023) 14:1265178. doi: 10.3389/fphar.2023.1265178,
2. Hubert, C, Tsiaparas, S, Kahlert, L, Luhmer, K, Moll, MD, Passon, M, et al. Effect of different postharvest methods on essential oil content and composition of three Mentha genotypes. Horticulturae. (2023) 9. doi: 10.3390/horticulturae9090960
3. Barbouchi, M, Benzidia, B, and Choukrad Mb,. Chemical variability in essential oils isolated from roots, stems, leaves and flowers of three Ruta species growing in Morocco. J King Saud Univ Sci. (2021) 33:101634. doi: 10.1016/j.jksus.2021.101634
4. Liu, C-x, Cheng, Y-y, Guo, D-a, Zhang, T-j, Li, Y-z, Hou, W-b, et al. A new concept on quality marker for quality assessment and process control of Chinese medicines. Chin Herb Med. (2017) 9:3–13. doi: 10.1016/S1674-6384(17)60070-4
5. Wang, Z, Chang, H, Zhao, Q, Gou, W, Li, Y, Tu, Z, et al. Mass spectrometry imaging for unearthing and validating quality markers in traditional Chinese medicines. Chin Herb Med. (2024) 17:31–40. doi: 10.1016/j.chmed.2024.04.005,
6. Dai, X-m, Cui, D-n, Wang, J, Zhang, W, Zhang, Z-j, and Xu, F-g. Systems pharmacology based strategy for Q-markers discovery of Huangqin decoction to attenuate intestinal damage. Front Pharmacol. (2018) 9. doi: 10.3389/fphar.2018.00236,
7. Ren, J-l, Zhang, A-H, Kong, L, Han, Y, Yan, G-L, Sun, H, et al. Analytical strategies for the discovery and validation of quality-markers of traditional Chinese medicine. Phytomedicine. (2020) 67:153165. doi: 10.1016/j.phymed.2019.153165,
8. Page, MJ, McKenzie, JE, Bossuyt, PM, Boutron, I, Hoffmann, TC, Mulrow, CD, et al. The Prisma 2020 statement: an updated guideline for reporting systematic reviews. BMJ. (2021) 372:n71. doi: 10.1136/bmj.n71,
9. Wang, Q, Chen, G, Chen, X, Liu, Y, Qin, Z, Lin, P, et al. Development of a three-step-based novel strategy integrating Dmpk with network pharmacology and bioactivity evaluation for the discovery of Q-markers of traditional Chinese medicine prescriptions: Danlou tablet as an example. Phytomedicine. (2023) 108:154511. doi: 10.1016/j.phymed.2022.154511,
10. Parmar, I, Rathod, H, and Shaik, S. A review: recent trends in analytical techniques for characterization and structure elucidation of impurities in the drug substances. Indian J Pharm Sci. (2021) 83:789. doi: 10.36468/pharmaceutical-sciences.789
11. Li, L, Yang, L, Yang, L, He, C, He, Y, Chen, L, et al. Network pharmacology: a bright guiding light on the way to explore the personalized precise medication of traditional Chinese medicine. Chin Med. (2023) 18:146. doi: 10.1186/s13020-023-00853-2,
12. Wei, Y, Nie, L, Gao, L, Zhong, L, Sun, Z, Yang, X, et al. An integrated strategy to identify and quantify the quality markers of Xinkeshu tablets based on spectrum-effect relationship, network pharmacology, plasma pharmacochemistry, and pharmacodynamics of zebrafish. Front Pharmacol. (2022) 13. doi: 10.3389/fphar.2022.899038,
13. Li, D, Lv, B, Wang, D, Xu, D, Qin, S, Zhang, Y, et al. Network pharmacology and bioactive equivalence assessment integrated strategy driven Q-markers discovery for Da-Cheng-Qi decoction to attenuate intestinal obstruction. Phytomedicine. (2020) 72:153236. doi: 10.1016/j.phymed.2020.153236,
14. Zhang, L, Shi, X, Huang, Z, Mao, J, Mei, W, Ding, L, et al. Network pharmacology approach to uncover the mechanism governing the effect of Radix Achyranthis Bidentatae on osteoarthritis. BMC Complement Med Ther. (2020) 20:121. doi: 10.1186/s12906-020-02909-4,
15. Duan, S, Niu, L, Yin, T, Li, L, Gao, S, Yuan, D, et al. A novel strategy for screening bioavailable quality markers of traditional Chinese medicine by integrating intestinal absorption and network pharmacology: application to Wu Ji Bai Feng pill. Phytomedicine. (2020) 76:153226. doi: 10.1016/j.phymed.2020.153226,
16. Chen, Q, Zhao, Y, Li, M, Zheng, P, Zhang, S, Li, H, et al. Hplc-Ms and network pharmacology analysis to reveal quality markers of Huo-Xue-Jiang-Tang Yin, a Chinese herbal medicine for type 2 diabetes mellitus. Evid Based Complement Alternat Med. (2021) 2021:1072975. doi: 10.1155/2021/1072975,
17. Wang, X, Zhou, W, Wang, Q, Zhang, Y, Ling, Y, Zhao, T, et al. A novel and comprehensive strategy for quality control in complex Chinese medicine formula using Uhplc-Q-Orbitrap Hrms and Uhplc-Ms/Ms combined with network pharmacology analysis: take Tangshen formula as an example. J Chromatogr B Anal Technol Biomed Life Sci. (2021) 1183:122889. doi: 10.1016/j.jchromb.2021.122889,
18. Li, Z, Xiong, H, Li, N, Zhao, L, Liu, Z, Yu, Y, et al. Integrated Uplc-Q-Tof-Ms and network pharmacology approach-driven quality marker discovery of Danggui Shaoyao san for primary dysmenorrhea. Biomed Chromatogr. (2023) 37:e5608. doi: 10.1002/bmc.5608,
19. Cai, Z, He, J, Jiang, J, Zhao, Z, and Shu, Y. Systematic investigation of the material basis, multiple mechanisms and quality control of Simiao yong'an decoction combined with antibiotic in the treatment of Sepsis. Phytomedicine. (2023) 116:154910. doi: 10.1016/j.phymed.2023.154910,
20. Bi, S, Liu, Y, Lv, T, Ren, Y, Liu, K, Liu, C, et al. Preliminary exploration of method for screening efficacy markers compatibility in Tcm prescriptions based on Q-markers: anti-inflammatory activity of Dachaihu decoction as an example. J Ethnopharmacol. (2023) 312:116539. doi: 10.1016/j.jep.2023.116539,
21. Yeqing, C, Xinsheng, F, Liping, Z, Fangyuan, H, and Pengli, W. Screening and evaluation of quality markers from Shuangshen Pingfei formula for idiopathic pulmonary fibrosis using network pharmacology and Pharmacodynamic, phytochemical, and pharmacokinetic analyses. Phytomedicine. (2022) 100:154040. doi: 10.1016/j.phymed.2022.154040,
22. Wang, T, Lin, S, Li, H, Liu, R, Liu, Z, Xu, H, et al. A stepwise integrated multi-system to screen quality markers of Chinese classic prescription Qingzao Jiufei decoction on the treatment of acute lung injury by combining 'network pharmacology-metabolomics-Pk/Pd Modeling'. Phytomedicine. (2020) 78:153313. doi: 10.1016/j.phymed.2020.153313,
23. Zhao, Q, Gao, X, Yan, G, Zhang, A, Sun, H, Han, Y, et al. Chinmedomics facilitated quality-marker discovery of Sijunzi decoction to treat spleen Qi deficiency syndrome. Front Med. (2020) 14:335–56. doi: 10.1007/s11684-019-0705-9,
24. Wang, Z-w, Liu, C, Zhang, A-h, Yan, G-l, Sun, H, Han, Y, et al. Discovery of Q-markers of Wenxin formula based on a Chinmedomics strategy. J Ethnopharmacol. (2022) 298:115576. doi: 10.1016/j.jep.2022.115576,
25. Zhang, H, Chen, R, Xu, C, Zhang, G, Guan, Y, Feng, Q, et al. An integrated approach to discriminate the quality markers of traditional Chinese medicine preparation based on multi-dimensional characteristic network: Shenqi Jiangtang granule as a case. J Ethnopharmacol. (2021) 278:114277. doi: 10.1016/j.jep.2021.114277,
26. Lin, W-S, Yang, C-M, and Kuo, B-J. Classifying cultivars of rice (Oryza sativa L.) based on corrected canopy reflectance spectra data using the orthogonal projections to latent structures (O-Pls) method. Chemom Intell Lab Syst. (2012) 115:25–36. doi: 10.1016/j.chemolab.2012.04.005
27. Shan, X, Yang, X, Li, D, Zhou, L, Qin, S, Li, J, et al. Research on the quality markers of antioxidant activity of Kai-Xin-san based on the Spectrum-effect relationship. Front Pharmacol. (2023) 14:1270836. doi: 10.3389/fphar.2023.1270836,
28. Hu, C, and Xu, G. Metabolomics and traditional Chinese medicine. TrAC Trends Anal Chem. (2014) 61:207–14. doi: 10.1016/j.trac.2014.06.007
29. Tang, H, Guo, KX, Huang, KE, Li, YF, Chen, W, Wei, HY, et al. An assessment of the antihyperlipidemic ingredients of Qi Ge decoction based on metabolomics combined with serum pharmacochemistry. Biomed Chromatogr. (2024) 38:e5922. doi: 10.1002/bmc.5922,
30. Li, Q, Ren, J, Yang, L, Sun, H, Zhang, X, Yan, G, et al. Parsing the Q-markers of Baoyin Jian to treat abnormal uterine bleeding by high-throughput chinmedomics strategy. Pharmaceuticals (Basel). (2023) 16. doi: 10.3390/ph16050719,
31. Deng, Y, Zhou, X, Shen, J, Xiao, G, Hong, H, Lin, H, et al. New methods based on Back propagation (Bp) and radial basis function (Rbf) artificial neural networks (Anns) for predicting the occurrence of Haloketones in tap water. Sci Total Environ. (2021) 772:145534. doi: 10.1016/j.scitotenv.2021.145534,
32. Hou, L, Zhang, Q, and Du, Y. Width estimation of hidden cracks in tunnel lining based on time-frequency analysis of Gpr data and back propagation neural network optimized by genetic algorithm. Autom Constr. (2024) 162:105394. doi: 10.1016/j.autcon.2024.105394
33. Lin, S, Zhou, Y, Hu, J, Sun, Z, Zhang, T, and Wang, M. Exploration for a Bp-Ann model for gas identification and concentration measurement with an ultrasonically radiated catalytic combustion gas sensor. Sensors Actuators B Chem. (2022) 362:131733. doi: 10.1016/j.snb.2022.131733
34. Guo, L, Gong, M, Wu, S, Qiu, F, and Ma, L. Identification and quantification of the quality markers and anti-migraine active components in chuanxiong Rhizoma and Cyperi Rhizoma herbal pair based on Chemometric analysis between chemical constituents and pharmacological effects. J Ethnopharmacol. (2020) 246:112228. doi: 10.1016/j.jep.2019.112228,
35. Yang, L, Xue, Y, Wei, J, Dai, Q, and Li, P. Integrating metabolomic data with machine learning approach for discovery of Q-markers from Jinqi Jiangtang preparation against type 2 diabetes. Chin Med. (2021) 16:30. doi: 10.1186/s13020-021-00438-x,
36. Zhang, J, Wang, D, Zhang, X, Yang, J, Chai, X, and Wang, Y. Application of “spider-web” mode in discovery and identification of Q-markers from Xuefu Zhuyu capsule. Phytomedicine. (2020) 77:153273. doi: 10.1016/j.phymed.2020.153273,
37. He, L, Liu, Y, Yang, K, Zou, Z, Fan, C, Yao, Z, et al. The discovery of Q-markers of Qiliqiangxin capsule, a traditional Chinese medicine prescription in the treatment of chronic heart failure, based on a novel strategy of multi-dimensional "radar chart" mode evaluation. Phytomedicine. (2021) 82:153443. doi: 10.1016/j.phymed.2020.153443,
38. Wang, D, Ding, J, Feng, X, Chai, X, Yang, J, Liu, C, et al. Identification of Q-markers from Hedan tablet by employing “spider-web” mode and taking compounds’ hepatotoxicity into account. Chin Herb Med. (2022) 14:612–21. doi: 10.1016/j.chmed.2021.08.007,
39. Liu, Y, Li, X, Chen, C, Ding, N, Ma, S, and Yang, M. Exploration of compatibility rules and discovery of active ingredients in Tcm formulas by network pharmacology. Chin Herb Med. (2024) 16:572–88. doi: 10.1016/j.chmed.2023.09.008,
40. Gu, X, Li, Z, Feng, Y, Hu, Y, Xie, K, Shang, J, et al. A mathematical multiple-dimensional strategy for Q-markers identification based on “five principles”: Tianshu capsule for migraine treatment as an example. J Ethnopharmacol. (2025) 337:118848. doi: 10.1016/j.jep.2024.118848,
41. Noor, F, Tahir ul Qamar, M, Ashfaq, UA, Albutti, A, Alwashmi, ASS, and Aljasir, MA. Network pharmacology approach for medicinal plants: review and assessment. Pharmaceuticals. (2022) 15. doi: 10.3390/ph15050572,
42. Zhang, DY, Peng, RQ, Wang, X, Zuo, HL, Lyu, LY, Yang, FQ, et al. A network pharmacology-based study on the quality control markers of antithrombotic herbs: using Salvia Miltiorrhiza - Ligusticum chuanxiong as an example. J Ethnopharmacol. (2022) 292:115197. doi: 10.1016/j.jep.2022.115197,
43. Zhou, Y, Niu, M, Zhang, D, Liu, Z, Wu, Q, Chen, J, et al. Screening for anti-inflammation quality markers of Lianhua Qingwen capsule based on network pharmacology, Uplc, and biological activity. Front Pharmacol. (2021) 12:648439. doi: 10.3389/fphar.2021.648439,
44. Li, X, Ren, J, Zhang, W, Zhang, Z, Yu, J, Wu, J, et al. Ltm-Tcm: a comprehensive database for the linking of traditional Chinese medicine with modern medicine at molecular and phenotypic levels. Pharmacol Res. (2022) 178:106185. doi: 10.1016/j.phrs.2022.106185,
45. Yue, R, and Dutta, A. Computational systems biology in disease Modeling and control, review and perspectives. NPJ Syst Biol Appl. (2022) 8:37. doi: 10.1038/s41540-022-00247-4,
46. Fan, M, Jin, C, Li, D, Deng, Y, Yao, L, Chen, Y, et al. Multi-level advances in databases related to systems pharmacology in traditional Chinese medicine: a 60-year review. Front Pharmacol. (2023) 14:1289901. doi: 10.3389/fphar.2023.1289901,
47. Xu, H, Zhang, Y, Wang, P, Zhang, J, Chen, H, Zhang, L, et al. A comprehensive review of integrative pharmacology-based investigation: a paradigm shift in traditional Chinese medicine. Acta Pharm Sin B. (2021) 11:1379–99. doi: 10.1016/j.apsb.2021.03.024,
48. Zafeiris, D, Rutella, S, and Ball, GR. An artificial neural network integrated pipeline for biomarker discovery using Alzheimer's disease as a case study. Comput Struct Biotechnol J. (2018) 16:77–87. doi: 10.1016/j.csbj.2018.02.001,
49. Liu, M, Qiu, Z, Zhao, X, Sun, L, Wang, L, Ren, X, et al. Assessing the quality of calyx of Physalis alkekengi L. var. franchetii based on quantitative analysis of Q-marker combined with chemometrics and machine learning algorithms. J Chem. (2021) 2021:8502929. doi: 10.1155/2021/8502929
50. Chu, J, Liu, X, Zhang, Z, Zhang, Y, and He, M. A novel method overcomeing overfitting of artificial neural network for accurate prediction: application on thermophysical property of natural gas. Case Stud Therm Eng. (2021) 28:101406. doi: 10.1016/j.csite.2021.101406
51. Wang, J, Wei, F, Wang, Y, Liu, Q, He, R, Huang, Y, et al. Exploring the quality markers and mechanism of Bushen Huoxue prescription in prevention and treatment of diabetic retinopathy based on Chinmedomics strategy. J Ethnopharmacol. (2023) 306:116131. doi: 10.1016/j.jep.2022.116131,
52. Li, T, Wu, F, Zhang, A, Dong, H, Ullah, I, Lin, H, et al. High-throughput Chinmedomics strategy discovers the quality markers and mechanisms of Wutou decoction therapeutic for rheumatoid arthritis. Front Pharmacol. (2022) 13:13. doi: 10.3389/fphar.2022.854087,
53. Perricone, M, Arace, E, Corbo, MR, Sinigaglia, M, and Bevilacqua, A. Bioactivity of essential oils: a review on their interaction with food components. Front Microbiol. (2015) 6:76. doi: 10.3389/fmicb.2015.00076,
54. Vaou, N, Stavropoulou, E, Voidarou, CC, Tsakris, Z, Rozos, G, Tsigalou, C, et al. Interactions between medical plant-derived bioactive compounds: focus on antimicrobial combination effects. Antibiotics (Basel). (2022) 11:1014. doi: 10.3390/antibiotics11081014,
55. Organization WH. Who guidelines for assessing quality of herbal medicines with reference to contaminants and residues. Geneva: World Health Organization (2007).
56. China PotPsRo, editor. Pharmacopoeia Commission of the Ministry of Health of the People’s Republic of China. 2020 edition ed: Beijing: China Medical Science Press (2020).
57. Zhang, G, Wang, Q, Liu, X, Wang, X, Xie, W, Liu, K, et al. An integrated approach to uncover quality markers of traditional Chinese medicine underlying chemical profiling, network target selection and metabolomics approach: Guan-Xin-Jing capsule as a model. J Pharm Biomed Anal. (2020) 190:113413. doi: 10.1016/j.jpba.2020.113413,
58. Liang, G, Yang, J, Liu, T, Wang, S, Wen, Y, Han, C, et al. A multi-strategy platform for quality control and Q-markers screen of Chaiqin Chengqi decoction. Phytomedicine. (2021) 85:153525. doi: 10.1016/j.phymed.2021.153525,
59. Cheng, Y, Xiao, M, Chen, J, Wang, D, Hu, Y, Zhang, C, et al. Quality assessment and Q-markers discovery of Tongsaimai tablet by integrating serum Pharmacochemistry and network pharmacology for anti-atherosclerosis benefit. Chin Med. (2022) 17:103. doi: 10.1186/s13020-022-00658-9,
60. Chen, J, Liu, Y, Gai, X, Ye, Q, Zhou, S, Tian, C, et al. Studies on quality markers of Kaihoujian spray for anti-inflammation based on Gray correlation analysis strategy. Evid Based Complement Alternat Med. (2022) 2022:1–26. doi: 10.1155/2022/1273066,
61. Liu, S, Ding, P, Wu, M, Zhu, Z, Tao, J, Wang, J, et al. Screening quality markers (Q-markers) of Xiaoer chaige tuire oral liquid by in vitro sequential metabolism and in vivo biopharmaceutical analysis. Phytomedicine. (2023) 116:154844. doi: 10.1016/j.phymed.2023.154844,
62. Yan, Y, Li, J, Zhang, Y, Wang, H, Qin, X, Zhai, K, et al. Screening the effective components of Suanzaoren decoction on the treatment of chronic restraint stress induced anxiety-like mice by integrated Chinmedomics and network pharmacology. Phytomedicine. (2023) 115:154853. doi: 10.1016/j.phymed.2023.154853,
63. Zhang, Y, Qu, Q, Lei, X, Zhao, X, Zhang, X, Wei, X, et al. Quality markers of Guchang Zhixie pills based on multicomponent qualitative and quantitative analysis combined with network pharmacology and Chemometric analysis. J Pharm Biomed Anal. (2024) 240:115934. doi: 10.1016/j.jpba.2023.115934,
64. Liu, L-W, Tang, M, Zhang, Z-B, Zhou, P-P, Xue, L-P, Jia, Q-Q, et al. A stepwise integrated strategy to explore quality markers of Qishen Yiqi dripping pills against myocardial ischemia. Phytomedicine. (2024) 135:156182. doi: 10.1016/j.phymed.2024.156182,
65. Chen, S-H, Wu, H-S, Jiang, X-F, Zhou, C, Bian, X-R, He, X, et al. Bioinformatics and Lc-Qtof-Ms based discovery of Pharmacodynamic and Q-markers of Pitongshu against functional dyspepsia. J Ethnopharmacol. (2024) 329:118096. doi: 10.1016/j.jep.2024.118096,
66. Chu, Y, Zhang, X, Zuo, L, Wang, X, Shi, Y, Liu, L, et al. Establishment of a multi-strategy platform for quality control and quality markers screen of Mailuoshutong pill. J Pharm Biomed Anal. (2024) 243:116070. doi: 10.1016/j.jpba.2024.116070,
67. Wang, Q, Wang, Q, Huang, Q, Zhang, X, Qin, Z, Yu, Y, et al. Five-layer-funnel filtering mode discovers effective components of Chinese medicine formulas: Zhishi-Xiebai-Guizhi decoction as a case study. Phytomedicine. (2024) 129:155678. doi: 10.1016/j.phymed.2024.155678,
68. Li, Q, Pei, J, Wang, L, Li, J, Qin, D, Mo, H, et al. Study on pharmacodynamic substances and quality control of Jiuwei Jiangtang oral liquid based on network pharmacology, fingerprint and in vitro experiment. Nat Prod Commun. (2024) 19:1934578X241280871. doi: 10.1177/1934578X241280871
69. Wu, X, Liu, Q, Chen, D, Qin, W, Lu, B, Bi, Q, et al. Identification of quality control markers in Suhuang antitussive capsule based on Hplc-Pda fingerprint and anti-inflammatory screening. J Pharm Biomed Anal. (2020) 180:113053. doi: 10.1016/j.jpba.2019.113053,
70. He, F, Xiao, CX, Wang, CJ, Liang, J, Cheng, QQ, Zhang, L, et al. Discovery of quality markers in Hugan Qingzhi formula by integrating a lipid-lowering bioassay with Uhplc-Qqq-Ms/Ms. Evid Based Complement Alternat Med. (2020) 2020:1594350. doi: 10.1155/2020/1594350,
71. Chen, J, Gai, X, Xu, X, Liu, Y, Ren, T, Liu, S, et al. Research on quality markers of Guizhi Fuling prescription for endometriosis treatment based on gray correlation analysis strategy. Front Pharmacol. (2020) 11:588549. doi: 10.3389/fphar.2020.588549,
72. Du, K-z, Cui, Y, Chen, S, Yang, R, Shang, Y, Wang, C, et al. An integration strategy combined progressive multivariate statistics with anticoagulant activity evaluation for screening anticoagulant quality markers in Chinese patent medicine. J Ethnopharmacol. (2022) 287:114964. doi: 10.1016/j.jep.2021.114964,
73. Yu, Y, Zhu, Z, Xie, M, Deng, L, Xie, X, and Zhang, M. Investigation on the Q-markers of Bushen Huoxue prescriptions for Dr treatment based on Chemometric methods and Spectrum-effect relationship. J Ethnopharmacol. (2022) 285:114800. doi: 10.1016/j.jep.2021.114800,
74. Wu, S, Huang, Q, Sheng, F, Zhang, L, Zou, L, Yang, L, et al. Identification of potential quality markers of Zishen Yutai pill based on Spectrum-effect relationship analysis. Front Pharmacol. (2023) 14:1211304. doi: 10.3389/fphar.2023.1211304,
75. Song, C, Zhang, S, Yun, P, Dong, X, Yang, M, Sun, G, et al. A novel evaluation mode of Nao an capsules using multi-wavelength fusion fingerprint combined with quantitative analysis of multi-components by a single marker method and antioxidant activity. Microchem J. (2024) 206:111409. doi: 10.1016/j.microc.2024.111409
76. Chu, Y-j, Wang, M-l, Wang, X-b, Zhang, X-y, Liu, L-w, Shi, Y-y, et al. Identifying quality markers of Mailuoshutong pill against thromboangiitis obliterans based Chinmedomics strategy. Phytomedicine. (2022) 104:154313. doi: 10.1016/j.phymed.2022.154313,
77. Wang, Y, Yang, L, Zhang, X, Sun, Y, Sun, H, Yan, G, et al. Quality marker discovery of Danggui Jianzhong decoction for treating primary dysmenorrhoea based on Chinmedomics strategy. Phytomedicine. (2023) 115:154724. doi: 10.1016/j.phymed.2023.154724,
78. Yu, J-J, Wang, Y, Yan, X, Yan, X, Zhang, Y, Wei, J-J, et al. An integrated strategy to discover the quality control markers of herbal formulae of Danning tablet with anti-cholestasis applications. Arab J Chem. (2024) 17:105640. doi: 10.1016/j.arabjc.2024.105640
79. Zhang, Y, Yang, Y, Ren, J, Yan, G, Yang, L, Wu, X, et al. Chinmedomics strategy for elucidating the effects and effective constituents of Danggui Buxue decoction in treating blood deficiency syndrome. Front Mol Biosci. (2024) 11:1376345. doi: 10.3389/fmolb.2024.1376345,
80. Wan, S, Xie, X, Yang, G, and Feng, F. Discovery of the toxicity-related quality markers and mechanisms of Zhi-Zi-Hou-Po decoction based on Chinmedomics combined with differentially absorbed components and network pharmacology. J Ethnopharmacol. (2024) 320:117408. doi: 10.1016/j.jep.2023.117408,
81. Zhang, X, Liao, W, Ding, X, Zhang, Y, Long, C, Zhou, Q, et al. Multiple-matrices metabolomics combined with serum pharmacochemistry for discovering the potential targets and active constituents of Qifu decoction against heart failure. J Pharm Biomed Anal. (2024) 244:116114. doi: 10.1016/j.jpba.2024.116114,
82. Zheng, Y, Fan, C, Liu, M, Chen, Y, Lu, Z, Xu, N, et al. Overall quality control of the chemical and bioactive consistency of Shengmai formula. J Pharm Biomed Anal. (2020) 189:113411. doi: 10.1016/j.jpba.2020.113411,
83. Li, M, Tang, D, Xu, R, Zhang, S, Chen, Y, and Peng, W. Uncovering quality markers of Yiqi-Tongluo capsule against myocardial ischemia and optimization of its extraction process. J Chromatogr B. (2023) 1230:123926. doi: 10.1016/j.jchromb.2023.123926,
84. Guan, H, Wang, Q, Mei, Y, Ran, J, Zeng, F, Cai, H, et al. A multistep approach for exploring quality markers of Shengjiang Xiexin decoction by integrating plasma pharmacochemistry-pharmacokinetics-pharmacology. J Pharm Biomed Anal. (2024) 241:115999. doi: 10.1016/j.jpba.2024.115999,
85. Fan, Z, Han, X, Xia, L, Xu, X, Xie, J, Zhang, Q, et al. Screening of quality markers of Jie-Geng decoction based on integration of multiple methods. Trop J Pharm Res. (2022) 20. doi: 10.4314/tjpr.v20i4.24
86. Yu, G, Zhang, T, Xu, H, Bi, Y, Feng, X, Wang, J, et al. Investigating the active substance and mechanism of san-Jiu-Wei-tai granules via Uplc-Qe-Orbitrap-Ms and network pharmacology. Evid Based Complement Alternat Med. (2022) 2022:1487903. doi: 10.1155/2022/1487903,
87. Xu, Y, Liu, RR, Yu, XJ, Liu, XN, Zhang, X, Jiang, ZH, et al. Quality markers of Dajianzhong decoction based on multicomponent qualitative and quantitative analysis combined with network pharmacology and chemometric analysis. Phytochem Anal. (2024) 35:146–62. doi: 10.1002/pca.3281,
88. Wang, L, Wu, W, Li, G, Chen, H, Fan, Y, Chen, W, et al. An integrated approach to identify the Q-markers of Banxia-Houpo decoction based on nontargeted multicomponent profiling, network pharmacology, and chemometrics. Phytochem Anal. (2024) 36:832–45. doi: 10.1002/pca.3471,
89. Zhu, X, Lin, M, Chi, Y, Li, X, Jiang, Z, Jian, X, et al. Qualitative and quantitative analysis of chemical components in Qianggan capsule by Uhplc-Q-Tof-Ms/Ms and Lc-Smrm. J Chromatogr A. (2024) 1728:465020. doi: 10.1016/j.chroma.2024.465020,
90. Hu, L, Chen, G, Chen, J, Zou, Z, Qiu, Y, Du, J, et al. Quantitative ternary network-oriented discovery of Q-markers from traditional Chinese medicine prescriptions: Bu-Zhong-Yi-Qi-Tang as a case study. Phytomedicine. (2024) 133:155918. doi: 10.1016/j.phymed.2024.155918,
Glossary
BP-ANN - Back-propagation artificial neural network
CV - Coefficient of variation
GC/MS - Gas chromatography–mass spectrometry
HCA - Hierarchical cluster analysis
HPLC - High-performance liquid chromatography
LC/MS - Liquid chromatography–mass spectrometry
LTQOrbitrap-MS - Linear trap quadrupole Orbitrap mass spectrometry
MIV - Mean impact value
MS/MS - Tandem mass spectrometry
OPLS-DA - Orthogonal partial least squares-discriminant analysis
PCA - Principal component analysis
PCMS - Plotting of correlation between marker metabolites and serum components
PLSR - partial least squares regression
Q/TOF - Quadrupole time-of-flight
Q-marker - Quality marker
QMI - Quality marker index
RA - Regression area
[S1]–[S4] - Strategy 1 to strategy 4
SBR - Spectrum–bioactivity relationship
TCM - Traditional Chinese medicine
TCMSP - Traditional Chinese Medicine Systems Pharmacology Database
THP - Traditional herbal prescription
UHPLC - Ultra-high-performance liquid chromatography
UHPLC-Q-Exactive-Orbitrap-MS - Ultra-high-performance liquid chromatography coupled with Q-Exactive Orbitrap mass spectrometry
UPLC - Ultra-performance liquid chromatography
VIP - Variable importance in projection
Keywords: BP-ANN, metabolomics, multi-dimensional network, network pharmacology, Q-marker, strategy
Citation: Tran KN, Mac GL, Munni YA and Yang I-J (2026) Q-marker identification strategies in traditional Chinese medicines: a systematic review of research from 2020 to 2024. Front. Med. 12:1709969. doi: 10.3389/fmed.2025.1709969
Edited by:
Pengcheng Shi, Southern Medical University, ChinaReviewed by:
Chenghao Fei, Nanjing Agricultural University, ChinaSushma Jaiswal, Guru Ghasidas Vishwavidyalaya, India
Copyright © 2026 Tran, Mac, Munni and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: In-Jun Yang, aW5qdW55QGRvbmdndWsuYWMua3I=
†ORCID: Gia Linh Mac, orcid.org/0009-0004-2147-9740