1 Introduction
Cancer remains a leading cause of death globally, posing a significant threat to human health. Breast cancer has the highest incidence among malignancies in women, with the highest incidence and mortality rates
1–3. Human epidermal growth factor receptor 2 (HER2) positive breast cancer, a distinct subtype, has emerged as a key clinical concern because of its aggressive nature, poor prognosis, and high mortality rate.
4 Trastuzumab, a monoclonal antibody, is designed to bind specifically to the extracellular domain of HER2. This inhibits the formation of receptor homodimers and blocks downstream signaling pathways, ultimately inhibits tumor cell proliferation, making trastuzumab an important therapeutic agent. However, as trastuzumab has become more widely used, its adverse reactions have become more apparent. One such reaction is trastuzumab-induced cardiotoxicity (TIC), which has attracted considerable attention. According to statistics, 0.6% to 30.1% of patients treated with trastuzumab alone experience varying degrees of cardiotoxic reactions, primarily manifested by decreased left ventricular ejection fraction, heart failure, and arrhythmia. In severe cases, treatment must be terminated.
5,6 This threatens trastuzumab's clinical value and poses a significant challenge that requires urgent resolution.
The Qizao Baoxin Formula (QBF) is an empirical prescription developed by a renowned doctor at the Cancer Hospital of the Chinese Academy of Medical Sciences. Derived from the Shengyu Decoction, which was first recorded in
Lan's Secret Collection,
7 QBF consists of seven Chinese medicinal herbs:
Astragali Radix,
Codonopsis Radix,
Cinnamomi Ramulus,
Angelicae Sinensis Radix,
Ziziphi Spinosae Semen,
Salviae Miltiorrhizae Radix and
Chuanxiong Rhizoma. These herbs nourish the blood and tonify the heart
8–14. Previous clinical observations by our research group have shown that this formula has definite preventive and thera-peutic effects on TIC and it may be associated with regulating ferroptosis pathways.
Ferroptosis, a recently discovered form of programmed cell death involving iron-dependent lipid peroxide accumulation, has been identified as a critical mechanism in the pathogenesis of TIC.
15,16 Studies have shown that TRZ enhances iron uptake by increasing the expression of the transferrin receptor in cardiomyocytes, thereby creating conditions that promote ferroptosis.
17 Furthermore, TRZ increases the expression of acyl-CoA synthetase long-chain family member 4, promoting lipid peroxide accumulation.
16,18 It also inhibits the function of solute carrier family 7 member 11, reduces glutathione synthesis and downregulates the expression of glutathione peroxidase 4
18,19. Together, these effects impair antioxidant defence system. However, the active components, targets and related molecular mechanisms by which QBF regulates ferroptosis to alleviate TIC remain unknown. Due to the multi-component and multi-target nature of traditional Chinese medicinal formulas, conventional experimental approaches alone are insufficient to elucidate their holistic mechanisms. Network pharmacology integrated with molecular docking provides an efficient, high-throughput screening method to identify active components and predict their molecular targets.
20,21 This study employed these approaches to investigate how QBF modulates ferroptosis to alleviate TIC, aiming to establish a scientific foundation for the future research on QBF-regulated ferroptosis for treating TIC.
2 Material and methods
2.1 Collection of QBF components and targets
The chemical components of the herbs in the QBF were obtained from the following five Chinese medicinal herb databases: TCMSP 2.0, Herb 2.0, SymMap 2.0, ETCM 2.0, and BATMAN-TCM 2.0. The SMILES notations of the components were retrieved from PubChem. After merging the results from all databases, duplicate components were removed based on identical SMILES strings. Since QBF is administered orally, components must possess favorable pharmacokinetic properties for oral absorption to exert systemic effects. The SwissADME platform was used to screen these components and identify potential bioactive components suitable for oral administration. The specific screening criteria were as follows:
a. Solubility (Log S) >−6
b. Complete compliance with Lipinski's Rule of Five
22c. Gastrointestinal (GI) absorption = “High”
The known targets of the active components were retrieved from the TCMSP 2.0 database. To predict potential targets, the Swis-sTargetPrediction platform was used with the species set to “homo sapiens”. All target identifiers were standardized to official gene symbols using the UniProt database. After merging known and predicted targets, duplicate entries were removed based on gene symbols, and targets without valid human gene annotations were excluded.
2.2 Collection of TIC-related targets
The mRNA expression profile data were obtained from the GEO database under accession number GSE264120. This dataset includes cardiac tissue samples from HER2 positive breast cancer patients treated with trastuzumab, comprising patients with TIC and patients without cardiotoxicity. Transcriptome sequencing was performed using the Illumina NovaSeq 6000 platform (GPL24676). Differential expression analysis was conducted using the RStudio software. Genes with |log2FC|> 1 and P < 0.05 were considered significantly differentially expressed. A volcano plot was also generated using the RStudio software. TIC-related targets were also obtained from GeneCards database and PubMed. In GeneCards database, the keyword “trastuzumab-induced cardiotoxicity” was searched. In PubMed, the same keyword was used. All gene symbols were standardized using UniProt database, and duplicate targets were removed based on gene symbols.
2.3 Acquisition of ferroptosis pathway-related targets
Ferroptosis pathway-related targets were obtained from the FerrDb 2.0 database, which includes genes related to ferroptosis promoters, inhibitors, and markers. Gene symbols were standardized using UniProt database, and duplicate targets were removed using gene symbols.
2.4 Screening of potential targets for QBF for improving TIC via ferroptosis pathways
The QBF targets, TIC-related targets, and ferroptosis pathway-related targets were entered into Venny 2.1.0, and the intersection was taken. The common targets were chosen as potential QBF targets for improving TIC via ferroptosis pathways.
2.5 Construction of the Protein-Protein Interaction (PPI) network
The potential targets identified above were brought into the STRING database with the species set to “homo sapiens” and a confidence score of ≥ 0.4 to construct the PPI network. Cytoscape 3.9.1 was used to make the visualization, and the “CytoNCA” plug-in was used to calculate the degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC) of each node in the network. Nodes with higher centrality values were considered more biologically significant in the network. Targets with DC, BC, and CC values higher than the median were selected as the core targets.
2.6 Enrichment analysis
These potential targets underwent Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses using RStudio software and the “clusterProfiler,” “enrichplot,” and “org.Hs.eg.db” plug-ins, with a screening criterion of Padj < 0.05. The “ggplot2” plug-in was used to make the visualization.
2.7 Construction of the “Chinese Medicine-Component-Target” network
These potential targets were brought into Cytoscape 3.9.1 to construct the “Chinese Medicine-Component-Target” network. The DC, BC, and CC values for topological analysis were calculated using the “CytoNCA” plug-in. As nodes with higher centrality are considered more significant, components with DC, BC, and CC values all exceeding the median were selected as the core components for subsequent analysis.
2.8 Molecular docking
Core targets and core components were selected for molecular docking validation. The protein crystal structures were sourced from the PDB database and prepared in PyMOL software by removing water and ligands, adding hydrogen atoms and optimising the structures using AutoDockTools. The ligand structures were energy-minimised in ChemDraw 3D software. The protein and compound files were converted to PDBQT format, and docking simulations were carried out using AutoDock Vina 1.1.2 software. The results were then visualised in PyMOL software. A binding energy of < −5.0 kcal/mol was used as the threshold to indicate strong binding activity.
3 Results
3.1 Potential targets of QBF for improving TIC via ferroptosis pathways
A total of 1537 chemical components from the seven herbs of QBF were obtained by searching five databases and removing duplicates. This large number reflects the multi-component nature of the QBF formula. After screening with the SwissADME platform, 817 potential active components were identified that met all three criteria: Log S > −6, full compliance with Lipinski's Rule of Five, and GI absorption = “High”. After removing duplicates using gene symbols,1690 potential targets were obtained.
Differential expression analysis was performed on the GSE264120 dataset. Based on the screening criteria of |log2FC| > 1 and P < 0.05, a total of 620 differentially expressed genes were identified, of which 304 were up-regulated and 316 were down-regulated. The volcano plot of the differential expression results is shown in Fig. 1. These differentially expressed genes were merged with TIC-related targets retrieved from GeneCards database and PubMed. After removing duplicates using gene symbols, 856 targets remained.
A sum of 969 targets related to ferroptosis pathways were obtained by searching the FerrDb database.
An intersection analysis was performed on the three sets of targets. Fig. 2 shows that a total of 52 common targets were identified. These targets were considered potential QBF targets for improving TIC and are associated with the regulation of ferroptosis pathways.
3.2 PPI network
The PPI network of these 52 targets was obtained via the STRING platform. As shown in Fig. 3, the network contains 52 nodes and 677 edges. To identify the most biologically significant nodes, DC, BC, and CC values were calculated. Targets with values higher than the median for all three metrics were selected as core targets. Using these stricter criteria, 25 core targets were identified, including TP53, BCL2, STAT3, HIF1A, AKT1, EGFR, TNF, MYC, IL6, ESR1, PTEN, CTNNB1, MAPK3, MTOR, SRC, MAPK8, PTGS2, PARP1, ATM, NFE2L2, CD44, IFNG, MAPK1, CAV1, and HSPA5.
3.3 Enrichment analysis
GO enrichment analysis was performed to evaluate the functions of the 52 potential targets. A total of 1634 significantly enriched terms were identified, including 1443 biological processes (BPs), 59 cellular components (CCs), and 132 molecular functions (MFs). The top 10 terms of each category are shown in Fig. 4. BP terms include the cellular response to chemical stress, response to oxidative stress, intrinsic apoptotic signaling pathway, and regulation of cell proliferation, and more. CC terms are primarily related to the membrane raft, membrane microdomain, focal adhesion, and cell-substrate junction. MF terms are primarily associated with the DNA-binding transcription factor binding, protein serine/threonine kinase activity, and ubiquitin-protein ligase binding. KEGG enrichment analysis was performed to evaluate the signaling pathways involved in the 52 potential targets. A total of 157 signaling pathways were identified. The top 20 enriched signaling pathways are shown in Fig. 5 and include proteoglycans in cancer, endocrine resistance, EGFR tyrosine kinase inhibitor resistance, and the FoxO signaling pathway, and more.
3.4 “Chinese Medicine-Component-Target” network
The “Chinese Medicine-Component-Target” network was constructed for these 52 potential targets using Cytoscape 3.9.1, and the results are shown in Fig. 6. Nodes with higher centrality values are considered more biologically significant within the network. Core components were chosen based on DC, BC, and CC values all being higher than the median. A sum of 268 core components were found, including main chemical components such as quercetin, kaempferol, luteolin, calycosin, formononetin, and ferulic acid.
3.5 Molecular docking
Due to the large number of core targets and components identified, we focused on the potential key mechanisms and selected representative candidates for molecular docking. Based on the enrichment analysis that highlighted oxidative stress as a key regulator of ferroptosis, BCL2, HIF1A, and NFE2L2 three targets were selected
23–26. Six components of herbs of QBF, quercetin, kaempferol, luteolin, calycosin, formononetin, and ferulic acid, were selected based on their abundance in QBF
27–30. All six components exhibited strong binding activity with these three targets, with binding energies all < −5.0 kcal/mol. Figs. 7–9 show the specific binding energy and two-dimensional interactions.
4 Discussion
Cancer remains a major global health concern. Breast cancer, the leading cause of cancer-related death among women, has garnered significant attention, especially the HER2 positive subtype. Trastuzumab is the important therapeutic drug for HER2 positive breast cancer, which can significantly improve patients' survival periods and is widely used in clinical practice. However, this drug often causes TIC after clinical application, which affects patient safety. Clinical studies have shown that the QBF can improve TIC, and its mechanism may be associated with the ferroptosis pathway. However, the main active components and molecular mechanisms by which the QBF regulates these pathways are not fully understood.
Owing to the compositional complexity and target diversity of Chinese medicinal formulas, this study employed network pharmacology for high-throughput screening. Through intersection analysis, 52 potential targets were identified. Further narrowing these down using PPI network analysis, we identified 25 core targets. GO analysis revealed significant enrichment in processes such as the response to oxidative stress and the intrinsic apoptotic signaling pathway. KEGG pathway analysis revealed pathways such as endocrine resistance, EGFR tyrosine kinase inhibitor resistance, and the FoxO signaling pathway. These findings suggest that QBF has diverse functions. Oxidative stress is a key regulator of ferroptosis, promoting heme oxygenase-1 (HO-1) expression, driving iron release via heme degradation, facilitating iron accumulation, and triggering subsequent lipid peroxidation.
31 Reactive oxygen species generated by oxidative stress also serves as a key trigger in lipid peroxidation.
32 Thus, the study focused on oxidative stress. Three targets related to oxidative stress, BCL2, HIF1A, and NFE2L2, were selected from the core targets for further analysis. BCL2 maintains normal mitochondrial homeostasis, including mitochondrial membrane permeability, which inhibits hydrogen peroxide leakage from mitochondria and protects cells against oxidative stress-induced damage.
23,24 HIF1A upregulates HO-1 expression and participates in the regulation of iron metabolism under oxidative stress conditions.
25 NFE2L2 modulates the expression of HO-1, FTH1, and AKR1B1, acting as a key regulator of inflammatory and oxidative stress signals.
26This study screened out a total of 268 core components. Due to the large number of core components, six main components of herbs of QBF, including quercetin, kaempferol, luteolin, calycosin, formononetin, and ferulic acid, were selected for further exploration of the mechanism
27–30. Molecular docking results showed that all six components strongly bound to BCL2, HIF1A, and NFE2L2. These computational predictions suggest that these components could modulate the activity of these key targets. Therefore, QBF may regulate ferroptosis pathways and improve TIC. Quercetin, a natural flavonoid found in
Astragali Radix and
Salviae Miltiorrhizae Radix in QBF, exerts antioxidant, anti-inflammatory, and lipid-regulating effects that confer cardioprotection.
33 Previous studies have shown that quercetin is an active component that improves cardiotoxicity induced by doxorubicin, another common therapeutic drug for HER2 breast cancer.
33 This study showed that quercetin may improve TIC by regulating oxidative stress and ferroptosis pathways. This provides a research basis for using quercetin to improve cardiotoxicity induced by the combined use of the two drugs. Calycosin, another main component of
Astragali Radix and the monarch drug in QBF, has extensive antioxidant and anti-inflammatory effects.
34 Previous studies have shown that calycosin can improve doxorubicin-induced cardiotoxicity.
34 This study showed that calycosin may improve TIC by regulating oxidative stress and ferroptosis pathways. This suggests that calycosin may be a Chinese medicinal component for improving cardiotoxicity induced by the combined use of the two drugs. This study provides a basis for this idea. Formononetin, another main component of
Astragali Radix, has extensive antioxidant, anti-inflammatory, and other effects.
35 Currently, research on its ability to improve cardiotoxicity induced by anti-HER2 positive breast cancer therapeutic drugs is limited. This study showed that formononetin may improve TIC by regulating oxidative stress and ferroptosis pathways. This provides a reference for researching its ability to improve anti-tumor drug-induced cardiotoxicity.
This study, while providing a systems-level perspective, has limitations. These findings, being computational predictions, warrant experimental validation. Future in vitro and in vivo experiments can verify the interaction between the core components and targets through which QBF regulates ferroptosis pathways and improves TIC.
5 Conclusions
Through quercetin, kaempferol, luteolin, calycosin, formononetin, and ferulic acid, QBF may target BCL2, HIF1A, and NFE2L2, thereby regulating oxidative stress, modulating ferroptosis, and alleviating TIC. This study provides a theoretical basis for future experimental research on QBF-regulated ferroptosis for treating TIC.
2026 The Authors. Publishing services by Elsevier B.V. on behalf of Higher Education Press and KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).