To reveal the key mechanism of Citri Reticulatae Pericarpium–Reynoutria japonica Houtt in the treatment of liver cancer and its correlation with lipid metabolism: synergetic effect with network pharmacology, molecular docking and bioinformatics

Abstract

Background

Liver cancer (LC) is a prevalent malignancy characterized by insidious onset, high recurrence rates, and significant mortality. The herbal combination "Citri Reticulatae Pericarpium–Reynoutria japonica" (CR) has traditionally been used for managing LC and related disorders, demonstrating notable therapeutic efficacy. This study aims to identify the molecular targets of CR in LC and assess its pharmacological properties and potential toxicity.

Methods

We used GeneCards, OMIM, and TTD databases to identify targets associated with liver cancer and constructed protein–protein interaction (PPI) and herbal–target–pathway–disease interaction networks. We performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses to explore the molecular mechanisms of CR. We also used differential expression analysis with GEO datasets (GSE76427, GSE87630, and GSE112790) to validate liver cancer-related targets and identify common core targets.

Results

Survival analysis indicated that HSP90AA1 may serve as a potential biomarker for LC. Molecular docking studies showed that rhein, a compound in CR, has a strong binding affinity to the identified target proteins. This research provides robust theoretical support for the clinical application of CR in treating LC, especially in cases related to non-alcoholic fatty liver disease (NAFLD) or Metabolic Lipid Disorders (DLM).

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-02708-8.

Introduction

The liver plays a pivotal role in the human body, serving as the central organ for metabolism, detoxification, and digestion. It regulates biochemical pathways essential for nutrient processing, toxin elimination, and immune responses. Notably, the liver produces bile, which is crucial for the emulsification and absorption of dietary fats in the small intestine. LC is one of the most lethal malignancies of the digestive system, with an increasing global incidence and high mortality rate [1, 2]. Among primary liver cancers, hepatocellular carcinoma (HCC) accounts for more than 80% of cases, particularly in regions such as East Asia and sub-Saharan Africa, where chronic hepatitis B virus and hepatitis C virus infections are highly prevalent [4]. According to the latest global cancer statistics, primary liver cancer ranks sixth in incidence and third in cancer-related mortality, highlighting its significant public health burden [5, 6].

LC is broadly classified into primary and secondary forms. Primary liver cancer originates from hepatic epithelial or mesenchymal tissues, with key risk factors including chronic viral hepatitis, NAFLD, liver cirrhosis, autoimmune disorders, genetic predisposition, obesity, diabetes, smoking, and dietary carcinogens [7, 8]. In contrast, secondary liver cancer results from metastatic spread from malignancies in organs such as the breast, pancreas, colon, lung, and stomach [9, 10].

The rising prevalence of NAFLD, particularly in developed nations, has led to a surge in NAFLD-associated LC, posing a growing challenge to global healthcare systems. Current treatment strategies for LC include surgical resection, radiation therapy, chemotherapy, targeted therapy, immunotherapy, and liver transplantation. However, these approaches are often associated with severe adverse effects, such as immune suppression, tissue damage, and multidrug resistance, which significantly impact patient survival and quality of life [11, 12]. Therefore, there is an urgent need for developing targeted therapies to better manage NAFLD-related LC and improve patient outcomes [13].

Consequently, the development of safer and more effective therapeutic techniques remains a pressing issue in liver cancer research. This pressing issue underscores the urgent need for alternative therapeutic approaches, particularly those derived from medicinal plants with established safety profiles and multi-target effects.

Medicinal plants, known for their safety profile and diverse bioactive components, are increasingly favored for drug discovery. Traditional Chinese Medicine (TCM), characterized by low toxicity, multi-target, and multi-system effects, has shown unique advantages in LC treatment, particularly as adjuvant therapy. Citri Reticulatae Pericarpium (CRP) and Reynoutria japonica Houtt (RJ) are traditional herbal medicines frequently used in combination in clinical practice. Recent studies have suggested that CR exhibits anti-liver cancer effects through various mechanisms, including inhibition of tumor cell proliferation, modulation of oxidative stress, and regulation of apoptosis-related pathways.

Given the increasing global burden of liver cancer and the limitations of conventional therapies, there is an urgent need to investigate the therapeutic potential of CRP and RJ, which have shown promising pharmacological properties in addressing the complex pathophysiology of liver cancer. CRP, obtained from the desiccated rinds of citrus fruits, is rich in flavonoids such as hesperidin, naringin, and nobiletin, which have been identified as key therapeutic agents [14] and has been found to have anticancer, anti-inflammatory, antioxidant, hepatoprotective, and anti-apoptotic properties [15–19].Particularly, the studies have highlighted the potential of these flavonoids in treating liver cancer [21].

Studies have demonstrated that flavonoids in CRP effectively regulate blood glucose, cholesterol, and lipid levels, preventing liver damage [22]. Hesperidin has been shown to induce apoptosis in liver cancer cells by activating the mitochondrial pathway and reducing oxidative stress [23]. Naringin exerts hepatoprotective effects by modulating PI3K/Akt signaling and reducing the production of pro-inflammatory cytokines [24, 25]. Nobiletin, protects against MAFLD and liver cancer. In MAFLD, Nobiletin modulates gut microbiota, enhances bile acid excretion, reduces inflammation, and regulates FXR/SHP and CYP7A1/CYP27A1 pathways to improve cholesterol metabolism. Additionally, CRP flavonoids, including Nobiletin exhibit anti-inflammatory and anti-oxidative effects, contributing to hepatoprotection and cancer prevention. Further studies are needed for clinical applications [26, 27].

RJ is derived from the dried rhizomes and roots of Reynoutria japonica Houtt. It contains active compounds such as polydatin, resveratrol, and rhein [28],which have pharmacological effects including anti-inflammatory, antioxidant, antiviral, anticancer, and hepatoprotective properties [29–33]. For example, polydatin suppresses LC cell proliferation, migration, and invasion by upregulating miR-877-5p expression, reducing the expression of metastasis-associated proteins [35]. Additionally, resveratrol and polydatin protect liver function by improving mitochondrial membrane potential and reducing mitochondrial damage [36]. Rhein inhibits liver cancer by targeting HSP72, HSC70, and GRP78, enhancing artemisinin derivatives’ efficacy and inhibiting tumor growth[37].

In light of the rising incidence of NAFLD-driven HCC and the critical demand for innovative, low-toxicity interventions, exploring the mechanisms of CRP and RJ is both timely and imperative. This study employs network pharmacology and molecular docking to investigate the mechanisms by which CR exerts its therapeutic effects on LC, aiming to identify key active components and their targets. Through a multi-target and multi-pathway approach, we aim to provide new insights into the systemic mechanisms of TCM in managing complex diseases such as LC. The study framework is illustrated in Fig. 1.

Fig. 1.

A detailed workflow of the network pharmacological investigation strategy for CR in the treatment of liver cancer. Five parts include database preparation, network construction, GO and KEGG pathway analysis, molecular docking verification, and validation of core targets in CR

To sum up, CRP has demonstrated promising anticancer potential in liver cancer models, showing inhibitory effects on hepatocellular carcinoma cell proliferation and migration, as well as the ability to enhance chemotherapy sensitivity. However, the precise molecular mechanisms remain unclear, necessitating further investigation using network pharmacology and molecular docking approaches.

Materials and methods

Screening of active compounds in CR and associated targets

To identify the active compounds of CRP and RJ, we first retrieved known compounds from the Traditional Chinese Medicine Systems Pharmacology (TCMSP) database [38]. A total of 63 compounds were initially identified from CRP and 62 compounds from RJ. To ensure pharmacokinetic relevance, we applied absorption, distribution, metabolism, and excretion (ADME) filtering criteria, selecting compounds with oral bioavailability (OB) ≥ 30% and drug-likeness (DL) ≥ 0.18. After applying these criteria, 5 compounds from CRP and 10 compounds from RJ were shortlisted. These selected compounds represent 7.93% of the total compounds in CRP and 16.13% of the total compounds in RJ.The potential targets of the filtered compounds were predicted using Swiss TargetPrediction [39] and PharmMapper [40]. The identified targets were then mapped to their official gene symbols using the UniProt database [41] for standardization.

Collection of LC-related targets

To identify potential targets associated with LC, a comprehensive search was conducted using the term "Liver Cancer" across three databases: GeneCards [42], the Therapeutic Target Database (TTD) [43], and the Online Mendelian Inheritance in Man (OMIM) database [44]. Targets with relevance scores exceeding the median value underwent a three-stage filtration process. The union of the target sets from the three databases was compiled, and duplicates were removed to generate a refined list of LC-associated genes.

Collection of NAFLD targets

A systematic search for potential NAFLD-related targets was performed using the keyword "non-alcoholic fatty liver disease" across three databases: GeneCards, TTD, and OMIM. Targets with relevance scores exceeding the median threshold underwent a three-step filtration process. The results from all three databases were merged, and redundant entries were eliminated to obtain a refined set of NAFLD-associated genes.

Collection of DLM targets

To identify potential targets associated with DLM, a comprehensive search was conducted using the term "Metabolic Lipid Disorders" across GeneCards. Targets with relevance scores exceeding the median value underwent a three-stage filtration process.

Identification of common targets between active compounds and LC

To identify shared targets between CR-related and LC-associated genes, a Venn diagram was generated using VENNY 2.1.0 [45], visually illustrating the overlapping gene set.

Construction of PPI network

The intersecting genes of CRP and LC were submitted to the STRING v3.11.1 database [46], with the species filter set to"Homo sapiens"and the interaction confidence threshold established at 0.9 to ensure high-confidence PPIs.The resulting PPI network data were imported into Cytoscape v3.7.1 for visualization and further network analysis. Key targets were identified using the CytoHubba plugin, based on degree centrality (DC). To comprehensively assess the PPI network, three topological parameters—degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC)—were analyzed, with their median values used as threshold criteria. Additionally, the Molecular Complex Detection (MCODE) plugin was employed to identify and evaluate functional biomolecular clusters within the network.

KEGG and GO enrichment analysis

To elucidate the biological functions of CRP, KEGG pathway and GO enrichment analyses were conducted using the DAVID database [47]. Annotations and visual representations of the results were generated using the Bioinformatics platform [48], with cutoff thresholds set at FDR < 0.05 and P < 0.05.Results were visualized using bar charts and bubble plots, facilitating a clearer understanding of the functional roles of the identified overlapping targets.

Network construction

Construction of topological network

To systematically illustrate the relationships between the drug CRP, its active compounds, and their biological targets, a'drug-active compound-target'network was constructed using Cytoscape v3.7.1.This network provides a comprehensive view of how CRP, through its active components, interacts with specific biological targets. A topological analysis of the network was performed to identify key nodes and interactions, thereby enhancing our understanding of the complex relationships involved.

Target-organ network construction

The metabolic processes of CRP in vivo are not fully elucidated, and its therapeutic effects on LC may involve multiple organs and tissues.To investigate the expression patterns of the targets of CRP's active compounds in various organs and tissues, we used the BioGPS database [49] to retrieve mRNA expression levels of these targets [50]. The data were derived from microarray analyses.

Based on the expression data, a'target-organ'network was constructed using Cytoscape v3.7.1, where nodes represent targets and organs, and edges connect targets to the organs in which they are expressed. Node colors and sizes were determined based on DC to highlight important targets and organs in the context of CRP's effects on LC.”

Molecular docking validation and ADMET analysis

The structural files of core ligands in SDF format were obtained from the PubChem database [51]. These files were then converted to PDB format using PyMOL version 2.5.0 [52]. Meanwhile, the crystal structures of the target proteins were acquired from the RCSB Protein Data Bank [53] and processed using PyMOL to isolate the ligands. The small molecule ligands and receptor proteins were loaded into AutoDockTools 1.5.7 [54] for preprocessing, which included dehydration, hydrogen addition, along with charge computation, and then saved in the form of PDBQT files to get ready for molecular docking.

A blind docking approach was employed in this study because it allows for a comprehensive exploration of all possible binding sites on the target protein without prior knowledge of the active binding region. This approach is particularly useful for discovering novel binding sites and evaluating the full potential of ligand-protein interactions. To achieve this, the docking grid box was set to encompass the entire target protein, allowing AutoDock Vina to explore potential binding sites across the entire protein surface. The grid box parameters were defined in Table 1.

Table 1.

Grid box centers and sizes for molecular docking of target genes

Gene name	Grid box center	Grid box size
TP53	x = 122.593, y = 97.379, z = −41.654	x = 24 Å, y = 26 Å, z = 28 Å
ESR1	x = 26.6, y = −12.807, z = 16.742	x = 32 Å, y = 26 Å, z = 34 Å
BCL2	x = −0.141, y = −5.997, z = 5.575	x = 28 Å, y = 22 Å, z = 20 Å
JUN	x = 23.99, y = −20.976, z = −4.06	x = 30 Å, y = 24 Å, z = 26 Å
TNF	x = 16.05, y = 15.494, z = 114.301	x = 30 Å, y = 28 Å, z = 34 Å
HSP90 AA1	x = −43.634, y = 15.277, z = −18.917	x = 18 Å, y = 24 Å, z = 36 Å
AKT1	x = 19.901, y = 18.534, z = 15.857	x = 24 Å, y = 18 Å, z = 26 Å
IL-6	x = 8.185, y = −25.073, z = −0.568	x = 28 Å, y = 26 Å, z = 16 Å

These grid box settings allow for the exploration of all possible binding sites within the protein structure.

Subsequently, molecular docking investigations were carried out by means of AutoDock Vina Version 1.1.2 [55]. The binding free energy results were visualized with the"ggplot2"and"pheatmap"R packages, and the docking interactions were displayed using PyMOL.

To predict drug-likeness properties,machine-learning analyses were carried out using the SwissADME [56] online server [57]. Ten significant compounds were analyzed for ADMET properties [58], and their toxicity profiles were predicted using the free online tool ProToxII [59]. The results were systematically recorded.

The target proteins selected for molecular docking were derived from the PPI network constructed using STRING and analyzed in Cytoscape. Key hub targets were identified based on topological parameters, including DC, BC, and CC. Targets exceeding two times the median DC value were considered pivotal hub proteins. In particular, TP53, AKT1, JUN, IL6, TNF, HSP90 AA1, ESR1, and BCL2 were identified as the most significant nodes in the network and thus selected for docking analysis.

The active compounds used in the docking analysis were selected based on their high relevance in the"drug-active compound-target"network, as well as their documented pharmacological effects against liver cancer. To ensure robust interaction potential, only compounds with high binding affinity (binding free energy ≤ −7 kcal/mol) and favorable pharmacokinetic properties were retained for further analysis.

Acquisition of liver cancer datasets and target validation

Three LC datasets, namely GSE87630, GSE112790, and GSE76427, were chosen from the Gene Expression Omnibus (GEO) database [60].

GSE87630: This dataset includes 64 HCC tumor samples and 30 non-tumor liver samples. The data was generated using expression profiling by array, profiling mRNA expression, DNA copy numbers, and DNA methylation in the same HCC patients. The aim was to identify genes whose transcriptional expression is regulated by genomic and/or epigenetic alterations.

GSE112790: This dataset involves a total of 183 HCC patients who underwent curative hepatic resection between 2006 and 2013 at Tokyo Medical and Dental University Hospital. The analysis utilized expression profiling by array and integrated multiple omics data to characterize the molecular subtypes of HCC. It also provided valuable insights into recurrence and metastasis of HCC.

GSE76427: This dataset contains 115 HCC tumor samples and 52 adjacent non-tumor tissues from patients who underwent liver resection between 2000 and 2013. The data was generated using lllumina HumanHT-12 v4 Expression BeadChip arrays, and linked clinical data were collected for prognostic evaluation. This study aimed to develop a new molecular stratification system for prognosis based on gene expression patterns.

Differential analysis was conducted using the "limma" R package, which includes tools for data normalization and interpretation of gene expression. Bubble plots were generated using the"ggplot2"and"pheatmap"R packages.

Genes that are common to both the CR potential targets and the DEGs identified from the three LC datasets were visualized using a Venn diagram. These intersecting genes were chosen for additional examination.

Survival analysis of LC

Survival analysis was conducted using GEPIA2 [61] based on the common genes identified through differential analysis. The clinical significance of specific genes was assessed according to their expression levels. In order to explore the correlation between gene expression and the overall survival of LC patients, Kaplan-Meier survival curves were plotted. GEPIA2's normalization features facilitated the comparison of relative transcriptional levels of two disparate genes.

Results

Active components of CR and Potential targets in LC

In the research process, a sum of 15 compounds (Table S1) originating from CR were successfully detected by means of the TCMSP, Swiss, and PharmMapper databases. Specifically, the TCMSP database helped to spot 221 potential targets (Table S2) associated with CR compounds, accompanied by their corresponding gene symbols. Subsequently, by integrating disease targets sourced from GeneCards, OMIM, and TTD databases and eliminating duplicate entries, 1,550 targets (Table S3) a related to LC were pinpointed. Through Venn diagram analysis, it was uncovered that there were 123 overlapping targets (Table S4) ashared between the active chemical targets of CR and the LC disease targets (Fig. 2a; Table 2). These intersecting targets are considered promising candidates for CR's anti-LC mechanisms.

Fig. 2.

Screening of CR active compounds and targets in liver cancer. A Venn diagram of 123 common targets between active compound targets of CR and liver cancer disease targets. B Herb–compound–target network

Table 2.

The 123 common targets between active compound targets of CR and liver cancer disease targets

No.	Target	No.	Target	No.	Target	No.	Target	No.	Target	No.	Target	No.	Target
1	AKT1	21	CDK4	41	RXRA	61	ADIPOQ	81	INSR	101	CYP1B1	121	GJA1
2	APOB	22	MYC	42	IL10	62	CYP19A1	82	ESR2	102	HSPB1	122	CAT
3	CASP8	23	MAPK1	43	GSTM1	63	PPARA	83	SPP1	103	CCNA2	123	ODC1
4	CCND1	24	NFE2L2	44	CDK2	64	STAT1	84	E2 F1	104	F3
5	CDKN2A	25	IL6	45	EGF	65	MAPK8	85	RASSF1	105	CCNB1
6	EGFR	26	TNF	46	PIK3CG	66	IGFBP3	86	NOS2	106	PRKCA
7	EPHB2	27	PPARG	47	CDKN1A	67	CCL2	87	HMOX1	107	CREB1
8	ERBB2	28	TGFB1	48	GSTP1	68	PGR	88	MAPK3	108	CTSD
9	ESR1	29	ERBB3	49	ICAM1	69	FASN	89	CASP9	109	NOS3
10	IL1B	30	VEGFA	50	CXCL8	70	CAV1	90	IL4	110	HMGCR
11	IRF1	31	JUN	51	PTGS2	71	ABCC1	91	NFKBIA	111	SOD1
12	MPO	32	BAX	52	CYP1A1	72	PARP1	92	IL1A	112	CD40LG
13	NQO1	33	IGF2	53	IFNG	73	CRP	93	RELA	113	CHUK
14	PTEN	34	MDM2	54	MMP9	74	SERPINE1	94	GSK3B	114	TOP1
15	RB1	35	IL2	55	UGT1A1	75	MMP1	95	MAPK14	115	MMP3
16	TP53	36	HIF1A	56	MMP2	76	TIMP1	96	NR1I2	116	DPP4
17	TOP2A	37	RAF1	57	CASP3	77	BCL2L1	97	MCL1	117	ACACA
18	CHEK2	38	CYP3A4	58	CYP1A2	78	PLAU	98	CHEK1	118	VCAM1
19	MET	39	BCL2	59	BIRC5	79	PCNA	99	SREBF1	119	AHR
20	AR	40	HSP90AA1	60	ABCG2	80	FOS	100	HSPA5	120	BAD

The "Drug-Compound-Target" network was developed utilizing cytoscape, incorporating 2 herbs,15 compounds, and 123 targets (Fig. 2b; Table 2). The edges of the network denote interactions between substances and targets, while the size and shading of nodes reflect the strength of these interactions. Notably, quercetin (MOL000098), luteolin (MOL000006), and naringenin (MOL004328) exhibited the strongest interactions with targets, suggesting that CR's therapeutic mechanisms for LC involve multiple compounds targeting various nodes.

PPI network construction and analysis

To explore the mechanisms underlying CR's treatment of LC, a PPI network based on 123 common targets were developed with the STRING database and visualized in Cytoscape (Fig. 3a).Nodes represent target genes and their respective proteins, while edges denote interactions. Node colors, ranging from light to dark, indicate increasing degree values. DC, CC and BC were employed as primary indicators for network analysis.

Fig. 3.

Identification of candidate targets through protein–protein interaction (PPI) analysis. A PPI analysis. B PPI network based on cluster analysis using the MCODE plugin. C Identification of top 28 core targets based on DC ≥ 2 times the median DC. D Bar chart of the top 28 core targets

Using topology-based screening criteria, 28 fundamental targets were discovered with a direct current threshold ≥2 times the median DC(Fig. 3b). The 28 primary targets are presented in a horizontal bar chart(Fig.3c; Table 3),with TP53 (96), AKT1 (64), JUN (56), IL6 (50), TNF (48), HSP90AA1 (46), ESR1 (46), and BCL2 (42) exhibiting the highest degrees and thus identified as pivotal hub targets. The MCODE plugin in Cytoscape conducted clustering analysis, identifying seven strongly interconnected sub-networks (Fig. 3d).

Table 3.

Top 28 targets, ranked by degree in the compound–target–pathway network of liver cancer

Gene name	Degree	Gene name	Degree
TP53	96	RELA	32
AKT1	64	CASP3	32
JUN	56	MDM2	32
IL6	50	FOS	28
TNF	48	MAPK8	28
HSP90AA1	46	CDKN1A	28
ESR1	46	CXCL8	26
BCL2	42	IFNG	26
MAPK3	40	CCL2	26
CCND1	40	HIF1A	26
MAPK1	40	CDKN2A	24
MYC	38	CDK4	24
EGFR	34	BCL2L1	24
IL1B	34	MAPK14	24

KEGG and GO enrichment analysis

KEGG pathway enrichment analysis discovered 178 critical LC-related signaling pathways. Among the top 30 enriched pathways, those most relevant to LC include hepatitis B, hepatitis C, HCC, and the PI3K/Akt signaling pathway (Fig. 4a). Sankey diagrams illustrated the relationships between target genes and the top 30 enriched pathways (Fig. 4b),highlighting the multifunctionality of CR's therapeutic targets in modulating complex biological processes.

Fig. 4.

Results of GO and KEGG enrichment analysis. A Barplot and bubble chart of the top 30 pathways based on KEGG enrichment analysis. B Barplot and bubble chart of GO functional enrichment analysis. C GO enrichment analysis. In the first ten enrichment analysis, green, orange purple represent the enrichment analysis of the biological process, cellular component and molecule function respectively. D Sankey diagram of the KEGG pathway to analyse the targets of CR in liver cancer treatment. The left rectangle, right rectangle and line represent the targets, the KEGG pathway and the pathway acted by the target, respectively

Functional annotation and enrichment analyses identified 732 Biological Processes (BP), 72 Cellular Components (CC) and 142 Molecular Functions (MF) (Fig. 4c).Bubble plots were used to visualize the top 10 items in each category, ranked by p-values and counts. BP analysis indicated significant enrichment in gene expression regulation, cellular responses to external stimuli, apoptosis regulation, and cell proliferation.

Analysis of compound–target–pathway interactions and core target identification in liver cancer

We selected the top 30 KEGG pathways, along with their corresponding targets and compounds, to develop a compound-target-pathway network using cytoscape (Fig. 5A). Table 3 lists the 16 highest-ranked targets (Table 4) based on their degree within the network. Five overlapping core targets were identified by intersecting the top 28 targets from the PPI network with the top 16 targets from the compound–target–pathway network, illustrated in the Venn diagram (Fig. 5B). Additionally, the top 10 compounds sorted by network degree were extracted for detailed analysis (Table 5).

Fig. 5.

Network pharmacology analyse results. A compound–target–pathway network. B Venn diagram of common core targets from liver cancer–PPI and drug–target–path analysis. C Target–path network

Table 4.

Top 16 targets, ranked by degree in the compound–target–pathway network of liver cancer

Gene name	Degree	Gene name	Degree
RB1	18	AR	5
PTGS2	11	GSTP1	4
HSP90AA1	11	IL10	4
TOP2A	9	DPP4	4
GSK3B	8	ESR1	4
JUN	7	BCL2	4
PIK3CG	7	CASP9	4
PPARG	6	CASP3	4

Table 5.

Details of the top 10 compounds in the compound–target–pathway network of liver cancer

Rank	Component ID	Component name	Structure	CAS	Degree	OB(%)	DL
1	MOL000098	Quercetin		117-39-5	97	46.43	0.28
2	MOL000006	Luteolin		491-70-3	45	36.16	0.25
3	MOL004328	Naringenin		67604-48-2	25	59.29	0.21
4	MOL005828	Nobiletin		478-01-3	23	61.67	0.52
5	MOL000358	Beta-sitosterol		83-46-5	13	36.91	0.75
6	MOL013287	Physovenine		6091-05-0	12	106.21	0.19
7	MOL013281	6,8-Dihydroxy-7-methoxyxanthone		87339-74-0	9	35.83	0.21
8	MOL000492	(+)-catechin		154-23-4	6	54.83	0.24
9	MOL005100	5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)chroman-4-one		520-33-2	5	47.74	0.27
10	MOL002268	Rhein		478-43-3	5	47.07	0.28

Target-organ analysis

We extensively assessed the mRNA expression levels of all 123 PPI targets utilising data from the BioGPS Database, with the objective of analysing the metabolism of CR in vivo during its LC administration. A Target-Organ Network comprising 1633 nodes and 178 edges was generated using Cytoscape. The organs related to CR-LC targets included Smooth Muscle (37 targets), Bronchial Epithelial Cells (36 targets), Liver (24 targets), and others (Fig. 5C).

This finding suggests that CR may act on multiple organs and cell types. However, how these organs contribute to its anti-LC effects requires further investigation. Apart from the digestive system, it also involves the activation of immune responses across diverse cell types, suggesting a broader spectrum of biological mechanisms in LC treatment.

Analysis of shared pathways and core targets in NAFLD and DLM-related LC

Recent clinical statistics indicates that NAFLD and DLM play critical functions in the advancement and evolution of LC. To explore these shared mechanisms, 99 common targets (Table S5) related to NAFLD and DLM Selected proteins were utilized to develop a PPI Network (Fig. 6A). Taking advantage of a cutoff value of at least twice the median DC, the five primary targets were picked (Fig. 6B).

Fig. 6.

Network pharmacology analyse results. A PPI analysis. B Bar chart of the top 5 core targets. C Compound–target–pathway network. D Venn diagram of common core targets from NAFLD+DLM PPI and drug–target–path analysis

To further elucidate the relationship between the Chinese herbal compound CR and LC associated with NAFLD and DLM, a compound–target–pathway network was established incorporating the active compounds and their corresponding pathways (Fig. 6C).The top 21 key targets were ranked by degree (Table 6). Subsequently, a Venn diagram intersecting the top five PPI targets with the top 21 network targets revealed two overlapping core targets (Fig. 6D).

Table 6.

Top 21 targets in the compound–target–pathway network of NAFLD-LMDs, ranked by degree

Gene name	Degree	Gene name	Degree
RB1	15	BCL2	4
PTGS2	11	CDKN1A	3
HSP90AA1	11	NOS3	3
JUN	9	RXRA	3
GSK3B	7	BAX	3
IL10	7	AKT1	3
PIK3CG	7	MMP9	3
PPARG	6	MAPK1	3
ADRB2	4	TP53	3
DPP4	4	GSTP1	3
ESR1	4

Validation via molecular docking and ADMET analysis

Molecular docking

The docking targets were selected based on PPI network analysis, where key hub genes were identified through topological screening (DC ≥ 2 × median DC). The final docking analysis focused on TP53, AKT1, JUN, IL6, TNF, HSP90AA1, ESR1, and BCL2, as they demonstrated the highest connectivity and functional significance in liver cancer pathogenesis.

Molecular Docking studies were conducted to validate the anticipated networks and investigate the interaction mechanisms of CR in the prevention and treatment of LC. The software AutoDock Vina (https://vina.scripps.edu/) was applied to evaluate the binding interactions that occur between specific active compounds and key target proteins retrieved from the RCSB Database.

Eight from protein–protein interaction networks (AKT1, JUN, TP53, TNF, HSP90AA1, BCL2, IL6, and ESR1) and their corresponding active compounds (e.g., Quercetin, Luteolin, Naringenin, Rhein) were chosen for docking studies. Binding free energy and the quantity of hydrogen bonds served as key metrics for docking evaluation. Compounds with binding free energy below −7 kcal/mol were deemed to possess substantial binding potential.

The heatmap results (Fig. 7; Table7) showed 38 ligand–receptor pairs with binding energies below −7 kcal/mol. Among these, TP53, TNF, HSP90AA1, and ESR1 had the most negative binding energies, suggesting that they might be crucial targets for CR-mediated LC treatment. PyMOL visualizations highlighted interactions such as hydrogen bonds, hydrophobic forces, and π–π stacking, supporting the stability of these complexes (Fig. 8).

Fig. 7.

Validation and screening of molecular docking. Heatmap of liver cancer molecular docking scores

Table 7.

Molecular docking binding affinities (kcal/mol) between active compounds and target proteins

ID	Compounds	AKT1 (kcal/mol)	JUN (kcal/mol)	TP53 (kcal/mol)	TNF (kcal/mol)	HSP90AA1 (kcal/mol)	ESR1 (kcal/mol)	BCL2 (kcal/mol)	IL6 (kcal/mol)
MOL000358	Beta-sitosterol	−7.0	−5.9	−6.1	−7.6	−7.9	−6.7	−7.7	−6.3
MOL013281	6,8-Dihydroxy-7-methoxyxanthone	−6.0	−6.1	−7.2	−7.1	−8.9	−7.4	−6.4	−6.0
MOL000492	(+)-catechin	−6.4	−5.9	−7.8	−8.9	−8.7	−7.3	−6.7	−6.4
MOL005100	5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)chroman-4-one	−6.1	−6.4	−8.2	−8.8	−9.0	−7.5	−7.1	−6.8
MOL002268	rhein	−6.7	−6.7	−8.3	−8.9	−9.6	−8.6	−7.5	−6.9
MOL000006	luteolin	−6.1	−6.5	−8.0	−8.8	−9.3	−8.0	−6.9	−7.2
MOL004328	naringenin	−6.2	−6.2	−7.8	−8.4	−9.1	−8.6	−6.7	−6.6
MOL000098	quercetin	−6.0	−6.4	−7.9	−9.0	−9.1	−7.3	−6.9	−7.1
MOL005828	nobiletin	−5.7	−5.8	−6.9	−7.3	−8.3	−6.9	−6.2	−6.0
MOL013287	Physovenine	−6.0	−6.4	−6.8	−6.2	−7.5	−7.5	−6.8	−5.5

Fig. 8.

a Docking pattern of BCL2, a core liver cancer target, and active compounds. b Docking pattern of ESR1, a core liver cancer target, and active compounds. c Docking pattern of HSP90AA1, a core liver cancer target, and active compounds. d Docking pattern of IL-6, a core liver cancer target, and active compounds. e Docking pattern of TNF, a core liver cancer target, and active compounds. f Docking pattern of TP53, a core liver cancer target, and active compounds

ADMET analysis

ADMET analysis is essential in drug discovery, assessing the pharmacokinetic and toxicological characteristics of prospective medicinal molecules. Using the SwissADME Database, we assessed the pharmacokinetic profiles of 10 key active compounds in CR. Table8 summarizes the results, indicating that all compounds exhibited favorable pharmacokinetic characteristics, encompassing gastrointestinal uptake, penetrability of the blood-brain barrier, the substrate activity of P-glycoprotein, as well as oral bioavailability. Notably, all compounds demonstrated excellent biocompatibility across various toxicity prediction models and were predicted to be non-hepatotoxic. However, Quercetin, Luteolin, and Rhein showed potential mutagenicity in specific predictive models. Additionally, Quercetin and Luteolin displayed carcinogenic potential in their predictions. Despite these findings, it is essential to interpret the results within the context of in vitro and in vivo validations, as computational models may overestimate potential risks without experimental evidence.

Table 8.

ADMET profiling of compounds

Compounds	Quercetin	Luteolin	Naringenin	Nobiletin	Physovenine	6,8-Dihydroxy-7-methoxyxanthone	(+)-catechin	5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)chroman-4-one	Rhein
GIabsorption	High	High	High	High	High	High	High	High	High
BBBpermeant	No	No	No	No	Yes	No	No	No	No
Pgpsubstrate	No	No	Yes	No	Yes	No	Yes	Yes	No
CYP1A2 inhibitor	Yes	Yes	Yes	No	No	Yes	No	Yes	No
CYP2C19 inhibitor	No	No	No	No	No	No	No	No	No
CYP2C9 inhibitor	No	No	No	Yes	No	No	No	No	No
CYP2D6 inhibitor	Yes	Yes	No	No	Yes	Yes	No	No	No
CYP3A4 inhibitor	Yes	Yes	Yes	Yes	No	Yes	No	Yes	No
PredictedLD50	159 mg/kg	3919 mg/kg	2000 mg/kg	5000 mg/kg	2 mg/kg	3800 mg/kg	10,000 mg/kg	2000 mg/kg	5000 mg/kg
ToxicityReversemutationAmestest	Non-Toxic	Non-Toxic	Non-Toxic	Non-Toxic	Non-Toxic	Non-Toxic	Non-Toxic	Non-Toxic	Non-Toxic
Hepatotoxicity	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive
Carcinogenicity	Active (0.68)	Active (0.68)	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive
Mutagenicity	Active (0.51)	Active (0.51)	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive	Active (0.76)
Cytotoxicity	Inactive	Inactive	Active (0.59)	Inactive	Inactive	Inactive	Inactive	Inactive	Inactive

These insights underline the importance of further experimental evaluation to comprehensively assess the safety and efficacy profiles of CR’s active compounds. The findings reinforce the therapeutic promise of CR while highlighting specific areas requiring additional scrutiny, such as mutagenicity and carcinogenicity predictions for certain compounds.

Validation of CR-related targets using GEO liver cancer datasets

We performed differential analysis of CR-related targets exploiting three LC datasets from the GEO database: GSE112790, GSE87630, and GSE76427. The"LIMMA"R package was employed to identify potential targets, retaining only those genes that met the adjusted criteria for DEG with adj. P < 0.05 and |log2(fold change)| ≥ 1. Volcano plots were made use of visually represent the distribution of differentially expressed genes (DEGs). In these plots, the upregulated genes were indicated by red dots, while the green dots represented the downregulated genes. Furthermore, the DEG distribution was depicted in a heatmap, with the top 50 DEGs across several groups (Fig. 9).

Fig. 9.

Volcano and heatmap plot of A GSE76427, B GSE87630, C GSE112790 datasets. The red color represents genes that are upregulated, the blue color represents genes that are downregulated

Among the four hub genes identified with strong binding affinity in molecular docking analysis, two genes (HSP90AA1 and ESR1) were found in these datasets. ESR1 was present and up-regulated across all three datasets, while HSP90AA1 was down-regulated in GSE76427 dataset. To further understand the clinical significance of these findings, we conducted survival analysis for these three genes. The findings demonstrated substantial correlations between these hub genes’ expression levels and the prognoses of patients, bolstering the likelihood that they could serve as viable therapeutic targets for LC.

Survival analysis

Survival analysis was performed using the Kaplan-Meier estimator to evaluate the survival function. The Kaplan-Meier curve, a widely used graphical representation of this function, depicts the probability of an event occurring at a specific time point. After generating the Kaplan-Meier survival curves, log-rank tests were conducted to compare survival rates between the high-expression and low-expression groups (Fig. 10A, B).

Fig. 10.

Using Pearson’s coefficient analysis at p < 0.05, the relationship between the three genes A HSP90AA1, B ESR1 were examined. Genes with no correlation are indicated in blue, genes with negative correlation are shown in red, and genes with positive correlation are shown in green. Core target’s expression in LIHC C HSP90AA1, D ESR1. E Survival map of hazardous ratio

Among the two analyzed genes, HSP90AA1 and ESR1, a p-value of 0.05 was observed, suggesting a potential correlation between their expression levels and LC survival outcomes (Fig. 10C, D). Further survival heatmap analysis revealed that HSP90AA1 functions as a favorable prognostic gene for LC (Fig. 10E), highlighting its potential role in predicting improved survival outcomes in LC patients.

Discussion

TCM and network pharmacology in LC

In recent years, natural herbal medicines have gained significant attention due to their multi-target and multi-pathway therapeutic effects in oncology. Network pharmacology has emerged as a powerful approach to elucidate the intricate interactions between bioactive compounds and their molecular targets [62, 63]. The clinical management of LC remains challenging due to drug resistance, severe toxic side effects, and limited therapeutic options. TCM has demonstrated substantial potential in enhancing treatment efficacy, mitigating toxicity, improving patients'quality of life, and prolonging survival. As a result, TCM has been increasingly recognized in the field of oncology [64].

The clinical management of LC remains challenging due to drug resistance, severe toxic side effects, and limited therapeutic options. TCM has shown promise in enhancing treatment efficacy, mitigating toxicity, and improving patient outcomes. Among TCM formulations, the CR herbal pair has demonstrated broad pharmacological activities, including anti-inflammatory, antioxidant, and tumor-suppressing effects. However, the specific bioactive compounds and molecular mechanisms underlying its anti-LC effects remain unclear.

This study employed network pharmacology and molecular docking to systematically investigate the active compounds, core targets, and key pathways of CR in LC therapy. The findings provide new insights into the molecular basis of CR’s anti-cancer effects, reinforcing its potential as a multi-target therapeutic strategy [65, 66].

Pathogenesis of LC and the therapeutic potential of CR

LC development is driven by genetic mutations and dysregulated signaling pathways, making targeted therapy a promising approach [67]. Major risk factors for LC include hepatitis virus infection, liver cirrhosis, chronic alcohol consumption, metabolic syndrome, diabetes, obesity-related NAFLD, and exposure to aflatoxin B1 [68].

The CR herbal pair, a commonly used prescription formulated by Professor Wang Lingtai for liver cancer treatment, has demonstrated significant therapeutic potential in both traditional medicine and modern biomedical research. Recent studies have shown that CR exerts a wide range of pharmacological effects, including alleviating hepatic inflammatory injury [69], reducing cholesterol gallstone formation [70], modulating tumor-associated inflammatory responses, and inhibiting cholesterol metabolism to prevent hepatic tumor metastasis[71]. Additionally, it has been found beneficial in improving cardiovascular health [72, 73], alleviating oxidative stress in hepatocytes [74, 75], regulating cholesterol biosynthesis[76], and treating chronic biliary tract infections [77, 78].

Accumulating evidence further underscores the therapeutic efficacy of CR's bioactive components in managing various digestive system disorders and cancers, such as gastrointestinal ulcers [80, 81], LC [82, 83], gastric cancer [85], and colorectal cancer [86–89]. Pharmacological studies suggest that CR inhibits cancer cell proliferation, induces tumor cell apoptosis, and suppresses migration and invasion, highlighting its potential as an effective anti-LC therapy [91–93].

Identification of key targets

To elucidate the molecular mechanisms of CR, we identified 15 bioactive compounds and 221 potential targets using TCMSP, Swiss, and PharmMapper. Further integration of GeneCards, OMIM, and TTD databases revealed 1550 LC-related targets, of which 123 overlapped with CR-associated targets.

PPI network analysis identified 28 core targets, including TP53, AKT1, JUN, IL6, TNF, HSP90AA1, ESR1, and BCL2. These targets regulate tumor progression, cell cycle control, inflammatory responses, and apoptosis, suggesting that CR may exert its therapeutic effects through these targets.

Functional enrichment and key signaling pathways

GO enrichment analysis revealed that CR primarily regulates gene expression, cell proliferation, apoptosis, and cellular responses to external stimuli. KEGG analysis identified nine major cancer-associated pathways, including Hepatitis B, Hepatitis C, the Cancer Signaling Pathway, the Hepatocellular Carcinoma Pathway, the PI3 PI3K/Akt Signaling Pathway, the AGE-RAGE Signaling Pathway, the IL-17 Signaling Pathway, the TNF Signaling Pathway, and the p53 Signaling Pathway (Fig. 4A). Among these, the PI3K/Akt signaling pathway is particularly significant, as it regulates cell survival, invasion, migration, and apoptosis resistance, and plays a key role in lipid metabolism. These findings suggest that CR may exert its anti-LC effects through modulation of these pathways. Additionally, genes such as HSP90AA1, RB1, and PTGS2 were significantly enriched (Fig. 4D). These genes not only play roles in the top 30 pathways, but are also involved in apoptosis [95], NAFLD [96], Th17 cell differentiation [97, 98], C-type lectin receptor signaling pathway [99], the HIF Signaling Pathway [100], and various cancer-associated pathways.

Compound–target–pathway analysis

To further explore the molecular mechanisms of CR, we constructed a compound-target-pathway network, identifying ten key bioactive compounds with potential anti-LC effects. These include quercetin, luteolin, naringenin, nobiletin, β-sitosterol, physovenine, 6,8-dihydroxy-7-methoxyxanthone, (+)-catechin, 5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)chroman-4-one, and rhein. Notably, these compounds have been shown to suppress tumor growth and metastasis by modulating inflammation-associated pathways, particularly the PI3K/Akt and JAK2/STAT signaling cascades.

Target-organ analysis

BioGPS database analysis suggested that the CR herbal pair may exert multi-target effects in LC treatment, possibly modulating immune system responses and potentially being involved in systemic immune mechanisms. This suggests a potential broader therapeutic mechanism that could extend beyond localized liver-specific actions, though further experimental validation is required to confirm these findings.

NAFLD, lipid metabolism, and CR's potential role

NAFLD is a well-established risk factor for LC, as lipid metabolism dysregulation promotes chronic inflammation, oxidative stress, and apoptosis imbalance, all contributing to LC progression [101–103].

In this study, we found that CR effectively modulates lipid metabolism disorders, alleviates oxidative stress, and suppresses inflammatory responses, potentially mitigating NAFLD-induced LC progression. Given that the PI3K/Akt pathway plays a central role in both lipid metabolism and LC progression, we propose that CR’s anti-cancer effects may be partially mediated through this pathway.

Molecular docking and pharmacokinetic properties

Molecular docking analysis revealed strong binding affinities between TP53, TNF, HSP90AA1, ESR1, and their respective bioactive compounds, supporting their potential therapeutic effects.

Furthermore, ADMET analysis indicated that all ten compounds exhibited no predicted hepatotoxicity, while nine (excluding physovenine) demonstrated high LD50 values, suggesting low acute toxicity and favorable pharmacokinetics.

Rhein showed the strongest binding affinity. We propose that Rhein’s anti-cancer effects may be mediated through the PI3K/Akt pathway, influencing lipid metabolism and apoptosis in LC cells. Experimental validation is necessary to confirm this hypothesis.

Emerging evidence highlights the PI3K/Akt signaling pathway as a critical regulator of LC progression, influencing cell proliferation, invasion, migration, and apoptosis inhibition [105, 106]. Additionally, this pathway plays a pivotal role in lipid metabolism, regulating lipid synthesis, storage, and oxidation, which in turn contributes to lipid accumulation and metabolic imbalances in hepatocytes. These dysregulated mechanisms have been implicated in the progression of NAFLD and lipid metabolism disorders, thereby influencing LC development and progression [107, 108].

In combination with the above research, four key targets are discussed.HSP90AA1 demonstrated a high docking score in molecular docking analyses. As a heat shock protein, HSP90AA1 is essential for reversing or preventing protein misfolding under stress conditions, thereby contributing to cellular processes such as growth, apoptosis, cancer progression, metastasis, and survival [109, 110]. Dong et al. [112] reported that HSP90AA1 exacerbates oxidative stress in hepatocytes, impairs mitochondrial respiratory chain function, induces inflammation, and increases LC risk. Furthermore, Wei et al. [113].demonstrated that HSP90AA1 activates the VEGFR2-PI3K/Akt pathway, thereby accelerating cell proliferation and inhibiting apoptosis. Additionally, numerous studies have reported significant overexpression of HSP90AA1 in LC patients.

ESR1, a transcription factor involved in DNA binding, hormonal regulation, and transcriptional activation, has been shown to inhibit tumor growth and delay inflammatory progression [114–116]. Its regulatory functions in LC operate through multiple pathways, including the IL-6 signaling pathway, lipid metabolism-related pathways, and non-coding RNA mechanisms [118, 119].

Similarly, TNF and ESR1 play critical roles in inflammatory and regulatory pathways. Tumor necrosis factor-alpha (TNF-α), a key pro-inflammatory cytokine, is integral to liver inflammation, cell proliferation, apoptosis, and necrosis. By activating the NF-κB signaling pathway, TNF-α modulates gene expression and induces apoptotic processes [121].

TP53, a well-established tumor suppressor gene, is an independent risk factor for poor LC prognosis [122]. It suppresses malignancy by regulating DNA damage repair, cell cycle inhibition, senescence, and apoptosis [123, 124].

These findings highlight the clinical significance of these targets and suggest that CR may exert therapeutic effects by modulating these key molecules, offering potential biomarkers for prognosis prediction and novel therapeutic targets (Tables 9 and 10).

Table 9.

Overview of data sets/tools/figures/tables

Label	Name of data file/data set	File types (file extension)	Data repository and identifier (DOI or accession number)
Data set 1	TCMSP database	Database	https://old.tcmsp-e.com/tcmsp.php
Data set 2	SwissTargetPrediction database	Database	http://www.swisstargetprediction.ch/
Data set 3	PharmMapper database	Database	https://www.lilab-ecust.cn/pharmmapper/index.html
Data set 4	The UniProt database	Database	https://www.uniprot.org/
Data set 5	GeneCards database	Database	http://www.genecards.org/
Data set 6	Therapeutic Target Database(TTD)	Database	http://db.idrblab.net/ttd
Data set 7	the Online Mendelian Inheritance in Man (OMIM) database	Database	https://omim.org/
Data set 8	The STRING v3.11.1 database	Database	https://string-db.org
Data set 9	The Gene Expression Omnibus (GEO) database	Database	https://www.ncbi.nlm.nih.gov/geo
Data set 10	DAVID database	Database	https://david.ncifcrf.gov/
Data set 11	BioGPS database	Database	https://biogps.org
Data set 12	The PubChem database	Database	https://pubchem.ncbi.nlm.nih.gov
Tool 1	VENNY 2.1.0	Web-based tool	https://bioinfogp.cnb.csic.es/tools/venny/index.html
Tool 2	The Bioinformatics platform	Web-based tool	https://bioinformatics.com.cn
Tool 3	PyMOL version 2.5.0	Web-based tool	https://pymol.org/2/
Tool 4	The RCSB Protein Data Bank	Web-based tool	https://www.pdb.org/
Tool 5	The AutoDock Tools 1.5.7	Web-based tool	https://ccsb.scripps.edu/mgltools/downloads/
Tool 6	AutoDock Vina Version 1.1.2	Web-based tool	https://vina.scripps.edu
Tool 7	SwissADME online server	Web-based tool	http://www.swissadme.ch/
Tool 8	ProToxII	Web-based tool	https://tox.charite.de/protox3/
Tool 9	GEPIA2	Web-based tool	http://gepia2.cancer-pku.cn/
Figure 1	A detailed workflow of the network pharmacological investigation strategy for CR in the treatment of LC	Image (.png)	Uploaded via journal submission system
Figure 2	A LC and CR target Venn diagram. B Herb–compound–target network	Image (.png)	Uploaded via journal submission system
Figure 3	A PPI analysis. B top 28 core targets. C Bar chart. D PPI network based on cluster analysis	Image (.png)	Uploaded via journal submission system
Figure 4	A KEGG enrichment analysis. B Sankey diagram of the KEGG pathway. C GO functional enrichment analysis. D GO enrichment analysis, exhibit BP, CC and MF respectively	Image (.png)	Uploaded via journal submission system
Figure5	A Compound-target-pathway network. B LC-PPI and Drug–Target–Path analysis Venn diagram. C Target-organ network	Image (.png)	Uploaded via journal submission system
Figure6	A PPI analysis. B Bar chart of the top 5 core targets. C Compound–target–pathway network. D NAFLD+DLM PPI and Drug-Target-Path analysis Venn diagram	Image (.png)	Uploaded via journal submission system
Figure 7	Heatmap of LC molecular docking scores	Image (.png)	Uploaded via journal submission system
Figure 8	Docking patterns of core targets and active compounds of Liver Cancer.	Image (.png)	Uploaded via journal submission system
Figure 9	Volcano and Heatmap plot. A GSE76427, B GSE87630, C GSE112790 datasets	Image (.png)	Uploaded via journal submission system
Figure 10	Pearson’s coefficient analysis A HSP90 AA1, B ESR1. Core targets’s Expression in LIHC. C HSP90 AA1, D ESR1 E Survival map of Hazardous Ratio	Image (.png)	Uploaded via journal submission system
Table 1	Grid Box Centers and Sizes for Molecular Docking of Target Genes	Table (.docx)	Uploaded via journal submission system
Table 2	The 123 common targets between active compound targets of CR and Liver Cancer disease targets	Table (.docx)	Uploaded via journal submission system
Table 3	Top 28 targets, ranked by degree in the compound–target–pathway network of liver cancer	Table (.docx)	Uploaded via journal submission system
Table 4	Top 16 targets, ranked by degree in the compound–target–pathway network of Liver Cancer	Table (.docx)	Uploaded via journal submission system
Table 5	Details of the top 10 compounds in the compound–target–pathway network of Liver Cancer.	Table (.docx)	Uploaded via journal submission system
Table 6	Top 21 targets in the compound-target-pathway network of NAFLD-LMDs, ranked by degree	Table (.docx)	Uploaded via journal submission system
Table 7	ADMET profiling of compounds	Table (.docx)	Uploaded via journal submission system
Table 8	Molecular docking binding affinities (kcal/mol) between active compounds and target proteins.	Table (.docx)	Uploaded via journal submission system
Table 9	Overview of data sets/Tools/Figures/Tables	Table (.docx)	Uploaded via journal submission system
Table 10	Abbreviations	Table (.docx)	Uploaded via journal submission system

Table10.

Abbreviations

Full name	Abbreviations
Absorption, distribution, metabolism, and excretion	ADME
betweenness centrality	BC
Citri Reticulatae Pericarpium	CRP
Citri Reticulatae Pericarpium-Reynoutria japonica	CR
Closeness centrality	CC
Degree centrality	DC
Drug likeness	DL
Gene Ontology	GO
Hepatocellular carcinoma	HCC
Kyoto Encyclopedia of Genes and Genomes	KEGG
Liver cancer	LC
Metabolic Lipid Disorders	DLM
Non-alcoholic fatty liver disease	NAFLD
Oral bioavailability	OB
Protein–protein interaction	PPI
Reynoutria japonica Houtt	RJ
The differentially expressed genes	DEGs
The Gene Expression Omnibus	GEO
The Molecular Complex Detection	MCODE
The Online Mendelian Inheritance in Man	OMIM
Therapeutic Target Database	TTD
Traditional Chinese Medicine	TCM
Traditional Chinese Medicine Systems Pharmacology	TCMSP
Tumour necrosis factor-alpha	TNF-α

GEO dataset validation and survival analysis

Expression analysis using GEO datasets confirmed that TP53, HSP90AA1, TNF, and ESR1 were significantly differentially expressed in LC tissues, further reinforcing their biological relevance in LC pathogenesis.

The survival analysis results further emphasize the clinical significance of CR-associated targets. Notably, HSP90AA1 and ESR1 overexpression correlated with poorer LC prognosis, suggesting their role in tumor progression and resistance to apoptosis and their potential as prognostic biomarkers. HSP90AA1 overexpression exacerbating oxidative stress, mitochondrial dysfunction, and inflammation. ESR1, known for its hormonal and transcriptional regulation functions, has been shown to suppress tumor growth and inflammatory progression.

Given that CR’s bioactive compounds interact with these targets, our findings indicate that CR may influence patient survival by modulating key oncogenic pathways. These results provide a strong rationale for further preclinical and clinical investigations to validate the therapeutic relevance of CR in LC treatment.

Clinical implications and future perspectives

These findings highlight the potential clinical significance of these targets, suggesting that CR may exert therapeutic effects by modulating key signaling molecules, offering novel biomarkers for prognosis prediction and treatment strategies.

Overall, this study provides compelling evidence supporting CR's multi-target effects in LC, reinforcing its potential as a complementary treatment strategy. Future research should focus on: Experimental validation of CR’s molecular mechanisms; Clinical trials assessing its efficacy and safety; Exploring its role in immune modulation and systemic therapeutic effects.

Conclusion

This study employed bioinformatics-based approaches, integrating network pharmacology and molecular docking techniques, to comprehensively analyze the pharmacological and molecular mechanisms of the CR herbal pair in LC treatment.

The findings provide strong evidence that bioactive compounds, including Quercetin, Luteolin, Naringenin, Nobiletin, Beta-Sitosterol, Physovenine, 6,8-Dihydroxy-7-methoxyxanthone, (+)-Catechin, 5,7-Dihydroxy-2-(3-hydroxy-4-methoxyphenyl)chroman-4-one, and Rhein, play a critical role in CR’s anti-LC effects.

The therapeutic mechanisms of CR involve multiple signaling pathways, including the PI3K/Akt signaling pathway, AGE-RAGE pathway, TNF pathway, IL-17 pathway, p53 signaling pathway, and pathways related to lipid metabolism and atherosclerosis. Among these, HSP90AA1 emerged as a promising therapeutic target, showing significant potential in regulating tumor proliferation in LC, with Rhein demonstrating the strongest binding affinityto this target. Furthermore, HSP90AA1’s involvement was closely linked to pathways associated with NAFLD and lipid metabolism disorders, further emphasizing the multi-component, multi-target nature of CR’s mechanism of action.

These research findings provide valuable insights into the therapeutic potential of CR in LC treatment, particularly in the context of NAFLD and DLM. Moreover, they lay a solid foundation for further in vivo and clinical studies, offering scientific guidance for the future development and application of CR as a therapeutic agent for LC.

Author contributions

This research was a joint endeavor among the authors. Each made specific and essential contributions. In writing, C.L. was responsible for the original draft preparation and told about the research ideas, methods, and initial outcomes. In visualization, C.L. made different figures and graphical things. Also, she did validation work. Y.J.W. and J.Y. W. were important in writing. They did reviewing and editing. They also got the funding. X. L. and C.F.R. helped in writing by doing supervision. Y.F.W. and Y.Z.M. were in writing through project administration. J.H.Z. and C.Y.G. were in visualization. All authors declare that they have no financial or personal interests that could have influenced the work reported in this manuscript. This study was funded by the Shanghai Municipal Health Commission and funded by J.Y.W.

Funding

Shanghai Municipal Health Commission, ZY (2018-2020)-RCPY-1011.

Data availability 

Data is provided within the manuscript or supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.