Evaluating the carcinogenic potential and molecular mechanisms of 2,3,7,8-tetrachlorodibenzo-p-dioxin in the human stomach using organoids and bulk sequencing data: a multi-machine learning approach combined with computational simulation

Abstract

Objective

This study investigates the impact of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) on the pathogenesis of gastric cancer and its associated molecular mechanisms, particularly the interaction between TCDD and key targets and pathways.

Methods

We employed various machine learning techniques and online databases to perform differential expression analysis on bulk gastric cancer sequencing data and organoid sequencing data to identify target genes and pathways related to TCDD and gastric cancer. A risk prediction model based on the expression levels of key intersection target genes was constructed. Network toxicology and molecular docking techniques were used to study the binding of TCDD to target proteins.

Results

A total of 24 genes were identified as potential target genes related to TCDD-induced gastric cancer. Machine learning analysis identified 5 core target genes as key intersection target genes of TCDD-induced gastric cancer, with the chemical carcinogenesis-receptor activation pathway, p53 signaling pathway, and IL-17 signaling pathway being the key pathways. Molecular docking revealed specific binding effects and binding sites between TCDD and intersection target proteins.

Conclusion

This study suggests that TCDD may affect the pathogenesis of gastric cancer by targeting specific genes and pathways. Molecular docking simulations indicate that there is a significant binding specificity effect between TCDD and target proteins, which is key to gastric cancer development.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04023-8.

Introduction

Gastric cancer is one of the most common malignant tumors of the digestive system. According to statistical data, the annual incidence rate of gastric cancer worldwide is about 15 per 100,000, particularly prominent in East Asia, posing one of the severe challenges to public health [1, 2]. Its incidence involves many factors, among which the most well-known include Helicobacter pylori infection, dietary habits, and genetic predisposition [3, 4]. Additionally, it is related to socioeconomic factors and the accessibility of medical resources. A comprehensive understanding of its epidemiological characteristics and related risk factors is crucial for the prevention and intervention of gastric cancer occurrence.

In recent years, epidemiological studies on gastric cancer have been increasing, especially in the association between H. pylori infection and gastric cancer. A substantial amount of evidence indicates that infection with specific virulent strains of H. pylori is closely related to the occurrence of gastric cancer [5, 6]. Furthermore, changes in dietary habits, such as the intake of high-salt diets and pickled foods, are also considered significant risk factors for gastric cancer [7]. However, despite numerous studies clarifying the risk factors for gastric cancer, there are many related risk factors that need to be refined and understood to further explore the mechanisms and influencing factors behind them.

In daily life, municipal solid-waste incineration, metal smelting and recycling, and chlorine bleaching of wood pulp all generate TCDD. Environmental TCDD is subsequently ingested by humans mainly through animal-derived foods—meat, fish, and dairy products. In addition, residents near chemical plants, chlorinated chemical production facilities, and hazardous-waste disposal sites may experience occupational or accidental TCDD exposure [8, 9]. TCDD possesses extremely strong chemical and thermal stability. It exists in solid form at room temperature, with a melting point range of 100 ~ 350 °C and a boiling point range of 300 ~ 550 °C. It is widely present in environmental media such as the atmosphere, water bodies, and soil. It has strong resistance to photolysis, chemical decomposition, and biological degradation, making it difficult to naturally break down in the environment, and it can exist for a long time. Due to the polymorphism and complexity of the ecological environment, it can be transmitted and enriched through the food chain, posing a potential threat to ecosystems and human health. For instance, long-term or repeated exposure can lead to dermatitis and chloracne and may also affect the bone marrow, endocrine system, immune system, liver, and nervous system. Moreover, TCDD has carcinogenic tendencies, with studies mentioning its strong carcinogenicity in both humans and animals. Epidemiological tracking survey results show that the incidence rate of cancers such as liver, lymphoid, and hematopoietic systems, as well as the digestive tract, is significantly increased in those exposed, thus it is classified as a Group 1 carcinogen. In addition, it also has immunotoxicity, endocrine toxicity, reproductive toxicity, and developmental toxicity, which can lead to skin diseases and neurological damage.

Molecular docking and network toxicology are essential tools in modern toxicology and drug research. Molecular docking uses computational simulations to predict the binding mode of small molecules with biological macromolecules, providing support for drug design and the study of interactions between biological molecules. Network toxicology constructs biological molecular networks to reveal the multi-target toxicity mechanisms of chemical substances, offering new methods for toxicity prediction and risk assessment. The combination of these two approaches can provide a more comprehensive understanding of the biological effects of chemical substances.

Although studies have mentioned a possible association between TCDD and the occurrence of gastric cancer, there has been no further research to explore this. Based on this, our study uses network toxicology and molecular docking techniques for the first time to explore the molecular mechanisms by which TCDD acts on gastric tissue and causes disease, hoping to provide a reference for the prevention and treatment of gastric cancer in populations exposed to TCDD.

Methods and materials

GEO database data retrieval and download

We retrieved and downloaded gene sequencing expression profiles related to gastric cancer from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), specifically the datasets GSE33335, GSE19826, and GSE63089, along with gastric cancer organoid mRNA-Seq data GSE33335.

Toxicity analysis and target gene acquisition

For the network toxicology analysis, we utilized the online tool ProTox 3.0 for chemical toxicity prediction to analyze the toxicity of TCDD. We used STITCH (http://stitch.embl.de/), SwissTargetPrediction (http://www.swisstargetpredictio-n.ch/), and ChEMBL (https://www.ebi.ac.uk/chembl/) to explore compound targets.

Disease-related gene acquisition

Databases GeneCards (https://www.genecards.org/) and OMIM (https://omim.org/) were used to explore genes related to gastric cancer.

Analysis based on R packages

We performed differential expression analysis using the “limma” package in R and plotted heatmaps and volcano plots. Multiple R packages, including “openxlsx”, “seqinr”, “plyr”, “randomForestSRC”, “glmnet”, “plsRglm”, “gbm”, “caret”, “mboost”, and “BART”, were used to screen diagnostic models using 113 machine learning methods. We used the “e1071/kernlab” package, “randomForest” package, and “glmnet” for feature gene screening with machine learning. The “pROC” package was used to analyze the diagnostic efficacy of the feature genes and plot diagnostic ROC curves. The “rms” and “rmda” packages were utilized to build and validate risk models.

Verification of expression of core intersecting target genes

The Human Protein Atlas (HPA) (https://www.proteinatlas.org/) is a large public database focused on the study of the human proteome. The expression of five core intersecting target genes in gastric cancer was investigated using HPA.

Molecular docking

To confirm the impact of TCDD on hub genes, molecular docking was conducted to explore potential binding interactions. SDF format ligand files and three-dimensional structural models of core target proteins were obtained from PubChem and RCSB Protein Data Bank [10]. CB Dock2 was used for protein-ligand blind docking.

Results

TCDD toxicity and carcinogenicity analysis

Using the online tool ProTox 3.0-Prediction Of Toxicity Of Chemicals (https://tox.charite.de/protox3/), we analyzed the toxicity and mutagenic and carcinogenic potential of TCDD. The results indicated significant mutagenic and carcinogenic effects of TCDD (detailed results are shown in Supplementary Material 1).

Disease-related gene screening

We used GeneCards (Supplementary Material 2) and OMIM (Supplementary Material 3) to explore genes associated with gastric cancer. The union of genes from both databases yielded 1665 genes related to gastric cancer (Fig. 1A). Subsequently, we downloaded three gastric cancer mRNA Seq datasets (GSE33335, GSE19826, and GSE63089) from the GEO database. Batch effects were removed from these datasets, and box plots before and after batch effect removal are shown in Figure S1A and B. To verify the batch effect removal, we performed principal component analysis (PCA) on the included samples, with results shown in Figures S1C and D. Then, differential expression analysis was conducted using the “limma” package (|log(FC)|>0.585), yielding 1268 differentially expressed genes, with a partial heatmap shown in Fig. 1B. Additionally, gastric cancer organoid mRNA-Seq data GSE33335 were retrieved and standardized for differential expression analysis, resulting in 1322 differentially expressed genes, with a partial heatmap shown in Fig. 1C. Furthermore, the mRNA-Seq data included in this analysis were also subjected to WGCNA, with scale independence and mean connectivity being optimal at a soft threshold of 6 (Fig. 1D). At this point, the module heatmap showed that the gray module had the best disease correlation (Fig. 1E, R = 0.19, P = 0.01). The relationship between the gene members of the gray module and the disease is shown in Fig. 1F (relationship = 0.52, 1.5e-06). Subsequently, the intersection of differentially expressed genes and related genes from the above parts was taken, with genes appearing in at least two screening projects considered related to gastric cancer, ultimately yielding 824 genes (Fig. 1G).

Fig. 1.

Disease-Related Gene Screening. A The union of gastric cancer-related genes screened from GeneCards and OMIM; B Heatmap of differential expression analysis results for gastric cancer mRNA-seq data; C Heatmap of differential expression analysis results for gastric cancer organoid mRNA-seq data; D Left panel (Scale Independence): Scatter plot showing the Scale-Free Topology Model Fit Index as the Soft Threshold changes. Right panel (Mean Connectivity): Scatter plot showing Mean Connectivity as the soft threshold changes; E Heatmap of the relationship between modules and traits; F Relationship between Module Membership and Gene Significance within the gray module

Screening of 2,3,7,8-tetrachlorodibenzo-p-dioxin related target genes

Subsequently, online tools STITCH (for detailed results, see Supplementary Material 4), SwissTargetPrediction (for detailed results, see Supplementary Material 5), and ChEMBL (for detailed results, see Supplementary Material 6) were used to explore TCDD-related targets. The union of targets obtained from the three databases resulted in a total of 320 TCDD-related targets (Fig. 2A). The intersection of these 320 TCDD-related targets with the 1665 gastric cancer-related genes yielded 24 cross-target genes (Fig. 2B), and a network diagram of these 24 cross-target genes was constructed, as shown in Fig. 2C.

Fig. 2.

TCDD Target Gene Screening. A A Venn diagram showing the union of target genes obtained from STITCH, SwissTargetPrediction, and ChEMBL; B A Venn diagram showing the intersection of disease-related genes with TCDD target genes; C A network diagram of the 24 intersecting target genes; D GO functional analysis of the 24 intersecting target genes; E KEGG pathway analysis of the 24 intersecting target genes

GO/KEGG enrichment analysis

Enrichment analysis was performed on the 24 core intersecting target genes for GO functions and KEGG pathways. The GO enrichment analysis results indicated that at the BP level, they are mainly involved in processes such as response to xenobiotic stimulus, positive regulation of cell cycle process, negative regulation of cell cycle process, regulation of mitotic cell cycle phase transition, and positive regulation of cell cycle; at the CC level, they are mainly involved in chromosomal region, melanosome, pigment granule, and ficolin-1-rich granule lumen; and in terms of MF, they are mainly involved in ubiquitin-like protein ligase binding, heat shock protein binding, histone deacetylase binding, ubiquitin protein ligase binding, and Hsp90 protein binding (Fig. 2D). The KEGG analysis results also indicated that these intersecting target genes are mainly enriched in multiple signaling pathways such as Lipid and atherosclerosis, Viral carcinogenesis, Chemical carcinogenesis - receptor activation, Drug metabolism - cytochrome P450, p53 signaling pathway, Metabolism of xenobiotics by cytochrome P450, and IL-17 signaling pathway (Fig. 2E).

Construction of diagnostic model based on 113 machine learning methods

Subsequently, a diagnostic model was constructed based on a combination of 113 machine learning methods, using the union of the three datasets as the training group and each of the three datasets as the validation set. The results are shown in Fig. 3A, where the “glmBoost + GBM” machine learning combination showed the best diagnostic performance. The diagnostic model’s results for the training and validation sets were visualized using ROC curves, as shown in Fig. 3B-E. To verify the diagnostic efficacy of the model, a confusion matrix was used for validation, which also showed high diagnostic performance (Fig. 3F-I). The “glmBoost + GBM” model was ultimately chosen as the best development plan, and 12 key intersecting target genes were identified: CTSB, WDHD1, ATAD2, MAOA, HIF1A, CA9, MMP14, ALDH3A1, CYP2B6, AHR, MMP3, and MMP1. The expression of these 12 key intersecting target genes in gastric cancer is shown in Fig. 3J. Figure 3K shows the diagnostic ROC curves for these 12 genes, from which it can be seen that all 12 genes have good diagnostic efficacy for gastric cancer (ROC AUC > 0.75) and are potential diagnostic targets.

Fig. 3.

Diagnostic Model Based on 113 Machine Learning Selections. A Heatmap of various diagnostic models; B–E ROC curves showing the predictive performance of each diagnostic model on the training and validation cohorts; F–I Confusion matrices for the diagnostic models on the training and validation cohorts; J Box plot of differential expression analysis of genes in the diagnostic model in gastric cancer; K Diagnostic ROC curves of the genes in the diagnostic model for gastric cancer

Core intersection target gene screening

Furthermore, this analysis conducted a more in-depth examination of the 12 key intersection target genes. Lasso regression analysis (Fig. 4A), random forest algorithm (RF, Fig. 4B and C), and support vector machine algorithm (Fig. 4D) were used to screen these 12 key intersection targets. The RF was set with a seed of 123 and ntree = 500, achieving the minimum error at ntree = 29, selecting the top five genes with the highest importance scores; Lasso regression identified 11 genes when the cross-validation error was minimized; the SVM algorithm also identified 11 genes at the point of minimum cross-validation error. By taking the intersection of the three machine learning methods, five core intersection target genes were ultimately determined: CA9, WDHD1, ATAD2, MAOA, and MMP14, as shown in Fig. 4E.

Fig. 4.

Core Intersection Target Gene Screening. A Screening of core target genes based on Lasso regression analysis; B, C Screening of core target genes based on the random forest algorithm; D Screening of core target genes based on the support vector machine algorithm; E Intersection of screening results from the three machine learning methods

Expression distribution of core genes

Subsequently, the expression of five core intersecting target genes in gastric cancer was investigated using the Human Protein Atlas (HPA) (https://www.proteinatlas.org/). Immunohistochemical results indicated that WDHD1, MMP14, and ATAD2 showed an upregulation trend in expression in gastric cancer compared to normal gastric tissue, while CA9 and MAOA exhibited a downregulation trend in expression in gastric cancer compared to normal gastric tissue. Specific details are shown in Fig. 5. This analysis also explored the intracellular expression distribution of five core intersecting target genes based on the HPA database. The results showed that ATAD2 and WDHD1 are expressed in both the cytoplasm and the nucleus, while CA9, MMP14, and MAOA are primarily expressed in the cytoplasm, as shown in Fig. 6.

Fig. 5.

Immunohistochemical results of five core intersecting target genes in gastric cancer

Fig. 6.

Immunofluorescence results of five core intersecting target genes in gastric cancer

Risk model construction

Subsequently, based on the five core intersection targets, we constructed a risk prediction model, as shown in Fig. 7A. The model indicates that the upregulation of WDHD1, MMP14, and ATAD2, and the downregulation of CA9 and MAOA are associated with the risk of gastric cancer occurrence. To verify the predictive performance of this risk prediction model, we plotted a calibration curve, and the results showed a good fit between the model’s predictive curve and the standard curve, suggesting that the model has good efficacy for predicting gastric cancer caused by TCDD (Fig. 7B). Decision Curve Analysis (DCA) was used to assess the net benefit of different predictive models at various threshold probabilities. This analysis also employed DCA to analyze the net benefit of the predictive model, and the results indicated that the model can bring a higher net benefit, especially at lower threshold probabilities (Fig. 7C). Additionally, we plotted clinical decision curves to evaluate the potential impact of the predictive model in clinical practice, and the results suggested a good degree of conformity between the predicted outcomes and actual events as the threshold changes, indicating the model’s potential for clinical application (Fig. 7D).

Fig. 7.

Risk model construction and validation. A A risk prediction model nomogram constructed based on five core intersection target genes; B Calibration curve; C Decision curve; D Clinical impact curve

Molecular docking

Subsequently, to confirm the impact of TCDD on hub genes, potential binding interactions were explored using molecular docking. SDF format ligand files and three-dimensional structural models of core target proteins were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/) and RCSB Protein Data Bank (https://www.rcsb.org/). Online tool CB Dock2 (https://cadd.labsh-are.cn/cb_dock2/index.php) was utilized for blind protein-ligand docking, and the results showed the most tightly bound binding sites (lowest scores, Score < −7). The results indicated the TCDD-ATAD2 docking sites as: I1074, N1064, Y1063, V1008, F1009, K1011, D1030, M1029, P1028, V1013, P1012, E1017, V1018, and Y1021, etc. (Fig. 8A); TCDD-CA9 docking sites as: P54, L53, Q52, G136, G206, Q205, G85, E195, A127, T125, S124, Y88, E87, G139, and R860, etc. (Fig. 8B); TCDD-MMP14 docking sites as: P191, Q390, H395, K388, G387, S400, F401, A311, D193, F198, G197, C399, and V306, etc. (Fig. 8C); These sites revealed the TCDD-WDHD1 molecular docking sites: Y626, D642, S643, G645, A628, W629, V677, W676, R696, Y675, H674, S437, T438, P439, and F697, etc., which may bind to specific amino acid residues of the protein (Fig. 8D). This is crucial for understanding how TCDD affects the function of these proteins and their potential biological effects.

Fig. 8.

Presentation of Molecular Docking Results. A TCDD-ATAD2 docking sites; B TCDD-CA9 docking sites; C TCDD-MMP14 docking sites; D TCDD-WDHD1 docking sites

Discussion

Although numerous studies have clarified the associated risk factors for gastric cancer, such as Helicobacter pylori infection and intake of nitrites [11–13], there are differences in research findings across different regions and populations [14, 15]. This is related to the functional characteristics of the stomach itself, different food and cooking habits faced by different regional populations, and exposure to various environmental risk factors, which contribute to the high incidence of gastric cancer and poor prognosis [16, 17]. Therefore, it is essential to further explore the mechanisms and influencing factors behind this.

As one of the high-risk carcinogens, TCDD has been associated with an increased incidence of cancer in some occupational exposure studies [18]. For instance, in an industrial accident in Italy, workers exposed to TCDD had a significantly higher incidence of gastric cancer over the subsequent 15 years than expected [19]. Additionally, a study of Korean and Vietnamese veterans found that those exposed to tactical herbicides had a significantly increased risk of gastric cancer and liver cancer [20]. TCDD binds to the aryl hydrocarbon receptor (AhR), activating a series of gene transcription and cellular signaling pathways, thereby affecting cell growth, differentiation, and metabolism [21]. These mechanisms play a key role in the occurrence and development of various cancers. Studying the relationship between TCDD and gastric cancer helps to understand its carcinogenic mechanisms and provides a theoretical basis for cancer prevention and treatment. TCDD exposure may also promote cancer occurrence by inducing inflammatory responses. For example, TCDD activates AhR, leading to the high expression of inflammatory factors such as IL-6, which play an important role in the occurrence and progression of gastric cancer [22, 23]. Understanding the carcinogenic mechanisms and exposure pathways of TCDD can help develop more effective cancer prevention strategies. For instance, reducing environmental emissions of TCDD, promoting low-toxicity alternatives, and strengthening occupational exposure protection may all potentially reduce the incidence of gastric cancer.

This study focuses on the potential impact of 2,3,7,8-tetrachlorodibenzodioxin on gastric cancer. Through multidimensional bioinformatics analysis and experimental validation, it systematically explores the toxic mechanisms of TCDD and its role in the occurrence of gastric cancer. By comprehensively analyzing TCDD’s toxicity, target gene screening, pathway analysis, and molecular docking, this study provides a new perspective on understanding TCDD’s carcinogenic mechanisms.

Functional enrichment analysis shows significant enrichment of TCDD-related target genes in the cell cycle regulation process. Specifically, these genes are involved in the positive and negative regulation of the cell cycle, affecting cell proliferation and apoptosis [24]. This finding reveals the potential mechanism by which TCDD promotes the occurrence and development of gastric cancer by affecting cell cycle regulation.

TCDD-related target genes are significantly enriched in lipid metabolism and atherosclerosis pathways [25]. Lipid metabolism is an essential biochemical process within cells, involving the synthesis, breakdown, and transport of fatty acids. TCDD-induced lipid metabolism disorders may lead to the accumulation of lipids within cells, affecting membrane fluidity and signal transduction, thereby promoting malignant transformation; reactive oxygen species (ROS) produced during lipid metabolism may further induce DNA damage and gene mutations, promoting cancer occurrence; moreover, lipid metabolism disorders may also trigger chronic inflammatory responses, providing a favorable microenvironment for cancer occurrence. This analysis also found significant enrichment of TCDD-related target genes in the chemical carcinogenesis-receptor activation pathway. As a typical chemical carcinogen, TCDD’s carcinogenic mechanism is mainly achieved by activating the aryl hydrocarbon receptor (AhR). KEGG analysis also showed significant enrichment of TCDD-related target genes in drug metabolism pathways [25]. Cytochrome P450 (CYP450) is an important group of drug-metabolizing enzymes involved in the metabolism and detoxification of various exogenous substances. TCDD, as an exogenous carcinogen, also relies on the CYP450 enzyme system for its metabolic process. In this study, we found that several genes related to drug metabolism (such as CYP2B6, ALDH3A1, etc.) showed abnormal expression under TCDD exposure. The abnormal expression of these genes may promote the occurrence of gastric cancer through the following mechanisms; these TCDD-related intersection target genes are also enriched in the IL-17 signaling pathway. IL-17 is an important inflammatory factor involved in chronic inflammatory responses and immune regulation [26]. The abnormal activation of the IL-17 signaling pathway may lead to chronic inflammatory responses, providing a favorable environment for cancer occurrence, and may also lead to immune regulation imbalance, weakening the host’s immune surveillance against cancer cells, while also promoting cell proliferation [27, 28]. In addition, these TCDD-related intersection target genes are also enriched in the classic cancer-related p53 signaling pathway. The abnormal activation of the p53 signaling pathway may lead to uncontrolled cell cycles [29], promoting cell proliferation and survival; its abnormal inhibition may lead to a decrease in the cell’s ability to repair DNA damage, further promoting gene mutations and cancer occurrence [30, 31].

This study also constructed a diagnostic model for gastric cancer based on a combination of 113 machine learning methods. Through analysis of the training and validation sets, we found that the “glmBoost + GBM” combination showed the best diagnostic performance [32]. Machine-learning algorithms can efficiently identify a small set of genes with genuine diagnostic or prognostic value, thereby avoiding the false positives often produced by traditional single-factor analyses. When applied to model construction, machine-learning-based models significantly outperform conventional approaches in performance metrics, enhancing the reliability of analytical results. The results of ROC curves and confusion matrices further confirmed the diagnostic efficacy of this model. Ultimately, we identified 12 key intersection target genes, which showed significant expression differences in gastric cancer and good diagnostic efficacy. This result indicates that these genes can serve as potential diagnostic markers for gastric cancer, providing new targets for early diagnosis and intervention.

Based on five core intersection target genes, we constructed a risk prediction model for gastric cancer. The results of calibration curves and decision curve analysis indicate that this model has good predictive performance and clinical application potential [33]. This finding suggests that by analyzing the expression levels of these core target genes, the risk of gastric cancer in populations exposed to TCDD can be effectively predicted. This discovery provides important theoretical support for health monitoring and early intervention in populations exposed to TCDD.

WDHD1 (WD repeat and HMG-box DNA-binding protein 1) safeguards genomic stability; its over-expression accelerates G1/S transition and is associated with poor prognosis across multiple cancers [34, 35]. MMP14 (matrix metallopeptidase 14) degrades extracellular matrix components, thereby promoting tumor invasion and angiogenesis [36]. ATAD2 (ATPase family AAA domain-containing 2) epigenetically reprograms chromatin to augment the activities of transcription factors such as c-Myc and AR, driving cell-cycle progression and stemness. In gastric cancer, ATAD2 is highly expressed and strongly correlates with the diffuse Lauren subtype, elevated Ki-67 index, and lymph-node metastasis; its silencing suppresses GC cell invasion [37]. CA9 (carbonic anhydrase 9) modulates the pH of the hypoxic tumor microenvironment, helping tumor cells adapt to low oxygen and evade immune destruction. In gastric cancer, CA9 expression is up-regulated by HIF-1α and positively associated with tumor hypoxia scores, VEGF expression, and adverse prognosis, making it a potential dual anti-angiogenic/immunotherapeutic target [38]. MAOA (monoamine oxidase A) generates ROS via catecholamine degradation, activates NF-κB signaling and the epithelial–mesenchymal transition (EMT) program, and promotes tumor invasion. Elevated MAOA expression in gastric tumors correlates with larger tumor size, distant metastasis, and shortened patient survival; MAOA inhibition reverses EMT and increases chemotherapeutic sensitivity [39]. Collectively, these five genes not only mediate escape from TCDD-induced oncogenic signaling but also play pivotal roles in gastric-cancer initiation, progression, and prognosis, offering a novel combinatorial target set for risk prediction and precision therapy of TCDD-related gastric cancer.

Through molecular docking analysis, we further confirmed the binding sites of TCDD with core target genes. The discovery of these binding sites reveals the potential impact of TCDD on the function of these genes, providing an important molecular basis for understanding TCDD’s carcinogenic mechanisms. This result is consistent with existing research, where TCDD binds to specific target genes, affecting their function and promoting the occurrence and development of cancer [40].

This study comprehensively reveals the potential impact of TCDD on gastric cancer and its carcinogenic mechanisms through multi-database integration, systematic analysis, machine learning model construction, and molecular docking validation. Although this study has made significant findings, there are still some limitations that need to be further improved in future research: (1) The study is mainly based on bioinformatics analysis and in vitro experiments, lacking in vivo experimental validation in animal models and clinical samples, making it difficult to fully confirm TCDD’s carcinogenic mechanisms [41]; (2) The sample size is small, especially in the high TCDD exposure group, which may affect the statistical power and reliability of the results [42]; (3) The study did not fully consider the interactive effects of TCDD with other environmental factors and genetic factors; future research should conduct gene-environment interaction analysis to identify high-risk populations [43]; (4) The data mainly come from public databases and may have batch effects and sample biases, affecting the accuracy and reliability of the results. Insufficient functional validation of target genes: Although the binding sites were confirmed through molecular docking analysis, there is a lack of in-depth validation of the functions of the target genes. Future research should further verify their functions and mechanisms of action through in vitro and in vivo experiments [44].

Conclusion

This study, through systematic analysis and integration of multiple databases, has revealed the potential impact of TCDD on gastric cancer and its carcinogenic mechanisms, providing new targets(WDHD1, MMP14, ATAD2, and CA9) and tools for early diagnosis and intervention. However, the limitations of the study suggest that future research will need to further verify the relationship between TCDD and gastric cancer, as well as its underlying mechanisms, through in vivo experiments, large-scale cohort studies, and multifactorial interaction analysis.

Abbreviations

TCDD

2,3,7,8-Tetrachlorodibenzo-p-dioxin

IL-17

Interleukin-17

H. pylori

Helicobacter pylori

GEO

Gene expression omnibus

TCGA

The Cancer Genome Atlas

HPA

Human protein atlas

PCA

Principal component analysis

WGCNA

Weighted gene co-expression network analysis

STITCH

Search tool for interacting chemicals

GO

Gene ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

BP

Biological process

CC

Cellular component

MF

Molecular function

Hsp90

Heat shock protein 90

CTSB

Cathepsin B

WDHD1

WD repeat and HMG-box DNA binding protein 1

ATAD2

ATPase family AAA domain-containing protein 2

MAOA

Monoamine oxidase A

HIF1A

Hypoxia-inducible factor-1α

CA9

Carbonic anhydrase 9

MMP14

Matrix metalloproteinase 14

ALDH3A1

Aldehyde dehydrogenase family 3 member A1

CYP2B6

Cytochrome P450 2B6

AHR

Aryl hydrocarbon receptor

MMP3

Matrix metalloproteinase 3

MMP1

Matrix metalloproteinase 1

ROC

Receiver operating characteristic

AUC

Area under the concentration-time curve

AhR

Aryl hydrocarbon receptor

ROS

Reactive oxygen species

Author contributions

Youfu Tian: Investigation, Data curation, Visualization, Formal analysis, Investigation, Methodology, Writing – original draft. Dede Ma: Conceptualization, Funding acquisition, Investigation, Validation. Ke Yan: Data curation, Visualization, Validation. Bing Xiao: Formal analysis, Investigation, Methodology. Jie Liu: Data curation, Validation. Miao Tan: Conceptualization, Supervision, Writing – review & editing.

Funding

This study was supported by the Natural Science Foundation of Shaanxi Province, 2024JC-YBQN-0988.

Data availability

The data used in this study were obtained from the GEO (https://www.ncbi.nlm.nih.gov/geo/) database and Human Protein Atlas (HPA) (https://www.proteinatlas.org/), both of which are available in publicly available databases. This study complies with its data use and publication rules.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.