Revealing the Impact of Mono(2‐ethylhexyl) Phthalate (MEHP) on Prostate Cancer Based on Network Toxicology and Molecular Docking Approaches

ABSTRACT

Mono(2‐ethylhexyl) phthalate (MEHP) is a ubiquitous environmental contaminant and endocrine‐disrupting chemical (EDC), identified as a potential carcinogen. Emerging studies have begun to elucidate the impact of MEHP on prostate cancer (PCa), yet its pathogenic effects and the underlying molecular mechanisms remain unclear. This study seeks to explore the molecular basis through which MEHP affects the onset and progression of PCa. Using network toxicology and bioinformatics, we identified MEHP‐related pathogenic genes in PCa. An innovative predictive model was developed by employing multiple machine learning ensemble algorithms, and its performance was validated using the area under the receiver operating characteristic (ROC) curve. Furthermore, at the single‐cell resolution, the role of key MEHP‐associated molecules, including several critical genes, in the oncogenic progression of PCa was identified. Through the construction of an environmental pollutant–key gene–PCa network, we investigated the interactions between environmental pollutants and the key genes VGF, ASPN, FOXS1, APLN, and AMH. Molecular docking studies demonstrated that the APLN, FOXS1, and ASPN genes exhibited favorable binding energies and high affinities for MEHP. The findings of this study provide a theoretical foundation for understanding the pathogenic role of MEHP in PCa and its potential molecular mechanisms. They also promote the application of network toxicology, molecular docking, machine learning, and single‐cell analysis in the study of environmental pollutants.

Short abstract

Mono(2‐ethylhexyl) phthalate (MEHP), an environmental endocrine‐disrupting chemical (EDC), exhibits potential carcinogenicity. This study employed network toxicology and bioinformatics to identify MEHP‐associated pathogenic genes in prostate cancer (PCa). Single‐cell analysis elucidated the role of MEHP‐related key molecules in PCa progression. An environmental pollutant–key gene–PCa interaction network was constructed to analyze associations between MEHP and critical genes (VGF, ASPN, FOXS1, APLN, AMH). Molecular docking confirmed high binding affinity between MEHP and APLN/FOXS1/ASPN proteins.

1. Introduction

Phthalates (PAEs) are commonly used as plasticizers in industrial processes and are regarded as endocrine disruptors due to their hormonal disruption. Among these, di(2‐ethylhexyl) phthalate (DEHP) and its primary toxic metabolite mono(2‐ethylhexyl) phthalate (MEHP) are the most prevalent examples (Y. Zhang, Guo, et al. ²⁰²¹). While the environmental presence and toxicity of PAEs have been well‐documented, there is limited comprehensive summary of the current understanding regarding their metabolites.

MEHP is the primary metabolite of DEHP and has attracted considerable attention because of its widespread environmental presence and potential risks to human health. Research has demonstrated that MEHP can be detected in various environmental matrices, such as serum in medical PVC bags (Wowkonowicz 2023), plastic containers, plasma products, and water in PVC tubing (Sanjuan et al. 2023). MEHP concentrations detected in urban freshwater and seawater range from 0.01 to 50 mg/L (Park et al. 2020). In certain contaminated soils, MEHP concentrations can reach levels as high as 13.0–166.7 ng/g (A. Wang et al. ²⁰¹²). Undoubtedly, MEHP is pervasive in daily life, present in everything from personal care products and plastic toys to medical devices and takeaway packaging. Therefore, MEHP persists in the environment over extended periods, presenting potential risks to human health.

Prostate cancer (PCa) is one of the most prevalent malignancies in men, with the highest incidence among male cancers and the second‐highest mortality rate (Siegel et al. 2025). Recent studies have increasingly highlighted that the onset and progression of PCa are influenced by a range of environmental pollutants, including heavy metals (cadmium, arsenic), polycyclic aromatic hydrocarbons (PAHs), and endocrine‐disrupting chemicals (EDCs) (Ikediobi et al. 2004; Kizu et al. ²⁰⁰³; Prajapati et al. ²⁰¹⁴; Rybicki et al. ²⁰⁰⁶; Treas et al. ²⁰²²). Among various environmental pollutants, EDCs have garnered significant attention due to their widespread exposure in the environment and their disruption of hormonal signaling pathways. Bisphenol A (BPA) is one of the most extensively studied EDCs, as it mimics estrogen and binds to estrogen receptors, thereby influencing the development and proliferation of prostate tissue (Pellerin et al. 2020). However, the development and progression of PCa are more reliant on androgen signaling pathways, meaning that focusing solely on BPA may not fully elucidate the role of EDCs in PCa (Dai et al. 2017). Recently, research has increasingly focused on another widely prevalent class of EDCs—PAEs. PAEs primarily affect the androgen receptor (AR) function by disrupting the androgen signaling pathway, thereby increasing the risk of PCa development (M. Zhu, Huang, et al. ²⁰¹⁸). Among various PAEs, DEHP is the most commonly used, and its metabolite MEHP demonstrates greater bioactivity and is widely distributed in the environment (Kratochvil et al. 2019). Research indicates that MEHP induces oxidative stress, leading to cytotoxicity, DNA damage, and an imbalance in the antioxidant system in LNCaP cells, thereby facilitating carcinogenesis (Erkekoğlu et al. 2010). At the epigenetic level, MEHP decreases 5‐mC methylation levels, triggering global DNA methylation abnormalities that may result in the dysregulation of key gene expression, thus accelerating PCa progression (Wu et al. 2017). Additionally, MEHP and its metabolites can accelerate cancer cell proliferation, regulate miRNA biosynthesis, and enhance migration and invasion capabilities, potentially promoting distant metastasis of PCa (Cavalca et al. 2022).

Current studies have provided preliminary evidence that MEHP may mediate the progression of PCa through multiple pathways, with a primary focus on oxidative stress and genetic damage (Cavalca et al. 2022; Erkekoğlu et al. ²⁰¹⁰; Wu et al. ²⁰¹⁷). However, the precise mechanisms through which MEHP exerts its effects remain largely unclear, including the modulation of AR signaling, interactions with other hormonal systems, the impact of different exposure windows on biological effects, epigenetic regulatory mechanisms, and potential synergistic or cumulative effects with other environmental pollutants. To date, no research has comprehensively analyzed how MEHP facilitates the progression of PCa through multiple levels and pathways in a systematic manner. Hence, a deeper investigation into the molecular mechanisms linking MEHP to PCa holds considerable scientific significance and clinical value.

Network toxicology encompasses the interdisciplinary integration of diverse scientific fields, including bioinformatics, big data analysis, genomics, and related technologies, offering a comprehensive framework to understand how chemicals disrupt biological molecular networks and impair cellular functions, potentially contributing to disease development (Huang 2023). Conversely, molecular docking simulates the complex binding patterns of plasticizers with protein targets at the atomic level, elucidating potential mechanistic pathways through which these chemicals could contribute to cancer pathogenesis (Sukumaran et al. 2024). Therefore, the integration of network toxicology and molecular docking techniques presents a highly promising analytical strategy. At the same time, machine learning techniques have proven effective in predicting and identifying unknown data through model training, optimization, and evaluation processes. Studies have demonstrated that machine learning algorithms can accurately predict disease outcomes by analyzing relevant genetic information (Verga et al. 2022). Similarly, single‐cell analysis characterizes cell states and activities by integrating multiple single‐modal omics approaches, including transcriptomics, genomics, epigenomics, epitranscriptomics, proteomics, metabolomics, and other omics (Vandereyken et al. 2023). Ongoing advancements in single‐cell analysis in terms of multiplexing, throughput, resolution, and accuracy have allowed us to comprehensively map the genetic landscape of cells and revolutionized our ability to identify cellular heterogeneity and interactions (Baysoy et al. 2023). These approaches are fundamentally transforming molecular cell biology research. Therefore, the integration of network toxicology, molecular docking techniques, machine learning, and single‐cell analysis presents a promising analytical strategy.

This study aims to combine advanced methodologies to uncover the molecular mechanisms by which MEHP affects the onset and progression of PCa. This research not only provides scientific evidence for evaluating the health risks of environmental pollutants but also contributes to developing novel strategies for the prevention and treatment of PCa. Furthermore, it offers theoretical support for the creation of more effective environmental pollutant management policies.

2. Methods

2.1. Collection and Processing of Data

In this study, we utilized four bulk RNA‐seq datasets of PCa (PCa) from different sources, along with their corresponding clinical data, all sourced from the PCaDB database (http://bioinfo.jialab‐ucr.org/PCaDB/). These raw datasets primarily originated from The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), and ArrayExpress. Specifically, the study included data from the TCGA‐PCa, Taylor, CIT, and Stockholm cohorts. These datasets provide gene expression profiles of PCa in different patient groups, helping us to gain a comprehensive understanding of PCa's molecular features and clinical manifestations. All gene expression values in PCaDB were normalized and log2‐transformed. We used the “limma” package to perform differential expression analysis on tumor and adjacent normal tissue from TCGA patients, identifying differentially expressed genes (DEGs) with a threshold set at p < 0.05 and |log2(fold change, FC)| > 1. Additionally, to explore the potential effects of environmental pollutants on PCa, we obtained data on MEHP (2‐ethylhexyl phthalate) and its associated genes from the Comparative Toxicogenomics Database (CTD, http://ctdbase.org) (see Table S1).

2.2. Protein–Protein Interaction Analysis

To further investigate the potential biological functions and to gain deeper insights into the potential biological functions and interactions of DEGs in PCa, we constructed a protein–protein interaction (PPI) network. The network was constructed via the STRING online database (https://string‐db.org). During the construction, we established filtering criteria that required interaction scores above 0.4, indicative of interactions with moderate confidence. These filtering criteria help ensure that the selected interactions are both reliable and biologically significant, thereby minimizing the occurrence of false positives. After obtaining the PPI network, we conducted in‐depth visualization and analysis on the Cytoscape platform. In Cytoscape, we employed the CytoHubba plugin for network analysis, which offers several algorithms for identifying key nodes and functional modules within the PPI network. Specifically, we applied the Degree algorithm, which assesses the degree of each node (i.e., the number of connections each gene has with others) to evaluate gene association, thereby identifying core genes that play a pivotal role in the network.

2.3. Functional Enrichment

To further investigate the potential signaling pathways of DEGs in PCa, we conducted functional enrichment analysis using the “clusterProfiler” package. Using this tool, we comprehensively annotated the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways to identify key signaling pathways involved in disease progression. During functional enrichment analysis, we identified significantly enriched KEGG pathways and visualized the results with a p‐value < 0.05. This approach highlighted the key biological features and signaling mechanisms of common genes, shedding light on the potential molecular mechanisms of MEHP in PCa development and progression.

2.4. Construction of Prognostic Signatures Using 117 Machine Learning Models and Algorithms

To identify genes with potential prognostic value, we first performed univariate Cox regression analysis to select genes from the intersection of DEGs and genes associated with MEHP (2‐ethylhexyl phthalate). This approach aimed to identify genes closely related to the survival and disease prognosis of PCa patients. To ensure the reliability of the results, we used the TCGA‐PCa dataset as both the training and internal validation sets, while the Taylor, CIT, and Stockholm datasets were used as external validation sets. This approach allowed us to validate the results and assess the generalizability of the findings across different data sources. We applied 10 machine learning algorithms, including lasso, ridge, stepwise Cox, CoxBoost, random survival forest (RSF), elastic net (Enet), partial least squares regression Cox (plsRcox), supervised principal component analysis (SuperPC), generalized boosted regression modeling (GBM), and survival support vector machine (Survival‐SVM). A combination of 117 algorithms was applied on the TCGA‐PCa training set for variable selection and model construction based on a 10‐fold cross‐validation framework. To comprehensively evaluate the performance of each model, we computed the concordance index (C‐index) for each model on the training, internal validation, and external validation sets. Based on these methods, we developed a new feature model, named MEHP‐related feature (MEHPRS), to predict the prognosis of PCa patients. Furthermore, to evaluate the predictive performance and prognostic value of these four cohorts, we plotted ROC curves.

2.5. Collection and Processing of Single‐Cell RNA Sequencing Data

We obtained single‐cell RNA sequencing data from six PCa patients in the GSE137829 dataset and analyzed them using the “Seurat” package. First, we performed quality control (QC) by retaining cells with mitochondrial gene content below 10% and genes expressed in at least three cells within an expression range of 500–7000. This criterion ensured the selection of high‐quality cells and genes, minimizing the impact of data noise on the analysis. Next, we selected the top 3000 highly variable genes for further analysis. To eliminate batch effects between samples, we applied the “Harmony” package for batch effect correction and visualized the data using the “UMAP” method. Finally, cells were annotated based on marker genes of different cell types, with reference to previous studies (Chan et al. 2022; Chen et al. ²⁰²¹; D. Wang et al. ²⁰²³) (see Table S2). In subsequent analyses, we assessed copy number variation (CNV) in epithelial cells. To this end, we utilized the R package “inferCNV,” employing immune cells as a reference. CNV analysis was conducted with a “denoising” procedure, utilizing the default hidden Markov model (HMM) settings, with a threshold of 0.1. To minimize the detection of false‐positive CNVs, we additionally employed the default Bayesian latent mixture model, setting the posterior probability threshold at 0.5, thereby enhancing the accuracy and reliability of the results. Furthermore, we conducted pseudotime analysis using the R package “monocle2” to elucidate the differentiation trajectory of the cells. Initially, the UMI matrix was extracted from the Seurat object, and the analysis object was constructed using the newCellDataSet function. In the trajectory analysis, genes with an average expression above 0.1 were selected as the foundation for the analysis. Dimensionality reduction was performed using the DDRTree method, and the cells were ordered using the orderCells function.

2.6. Construction of the Environmental Pollutant–Key Gene–PCa Network

To investigate the relationship between environmental pollutants and MEHPRS signature genes in PCa, we analyzed chemical–gene interactions using the CTD, with a particular focus on MEHPRS signature genes. Through this analysis, we identified and evaluated chemicals associated with key genes to construct a network illustrating the interactions among chemicals, genes, and PCa.

2.7. Molecular Docking

To further validate the direct effects of MEHP on the five key genes—VGF, ASPN, FOXS1, APLN, and AMH—and their respective binding sites, we employed MEHP as a small‐molecule ligand and the five key genes as macromolecular receptors. The three‐dimensional structures of the five key genes were retrieved from the Protein Data Bank (PDB, http://www.rcsb.org/), and the PDB format file for MEHP was sourced from the PubChem database (http://pubchem.ncbi.nlm.nih.gov/). The structures were converted into Mol2 format using the PyMOL software, and dehydration and hydrogenation preprocessing of the key proteins were carried out using AutoDock Vina (Version 1.5.6), followed by molecular docking. A binding energy lower than −5.0 kcal/mol is considered to indicate a strong receptor–ligand binding activity. The pdbqt result file generated by AutoDock Vina was imported into the PyMOL software, converted into pdb format, and the key protein with the highest binding energy was visualized using PyMOL 1.8. This analysis elucidated the primary binding sites of MEHP on the genes VGF, ASPN, FOXS1, APLN, and AMH, along with their mechanisms of action.

2.8. Statistical Analysis

In this study, statistical analyses were conducted using the R software (Version 4.3.2) and relevant packages. For continuous variables following a normal distribution, Student's t‐test was used for intergroup comparisons; for those not following a normal distribution, the Wilcoxon rank‐sum test was applied. To assess the predictive performance of the model and genes, ROC curve analysis and area under the curve (AUC) calculations were conducted using the “timeROC” R package. Additionally, Kaplan–Meier survival analysis was conducted using the “Survminer” R package. The significance level for statistical analysis was set at p < 0.05.

3. Results

3.1. Identification of Potential Target Genes of MEHP in PCa and Enrichment Analysis

We identified 3200 target genes associated with PCa and visualized them using a volcano plot (Figure 1A), which highlighted the top 10 upregulated and 10 downregulated DEGs. Differential gene expression from GEO data was then analyzed, and by combining search and screening from the CTD database, a Venn diagram was used to intersect these genes with MEHP‐interacting target genes, leading to the identification of 476 MEHP‐related genes influencing PCa prognosis (Figure 1B). Using univariate Cox regression analysis, 152 MEHP target genes associated with biochemical recurrence (BCR) in PCa were selected for further investigation (Figure 1C), representing potential targets linked to MEHP‐induced BCR. The heatmap reveals significant differences in the expression of these genes between normal and tumor tissues (Figure 1D). Furthermore, we constructed the PPI network of MEHP‐induced PCa target genes using the Degree algorithm in Cytoscape (Figure 1E). The network consists of 82 nodes and 181 edges, with each node representing a gene and edges indicating gene interactions. The node color intensity and size are proportional to their Degree values. A higher Degree value indicates greater importance of a node in the network, implying that the gene may play a pivotal role in the PPI network. To explore the biological characteristics and signaling mechanisms of key genes, we conducted KEGG enrichment analysis. The results revealed that MEHP may play a significant role in the development of PCa through pathways such as focal adhesion, extracellular matrix–receptor interaction, cAMP signaling, cell cycle, and AMPK signaling (Figure 1F).

FIGURE 1.

Identification of potential target genes of MEHP in PCa and enrichment analysis. (A) Volcano plot showing differential gene expression between PCa and normal groups from the GSE137829 dataset, highlighting the top 10 upregulated and top 10 downregulated DEGs. (B) Venn diagram of differentially expressed genes from the TCGA and CTD datasets. (C) Forest plot illustrating potential targets associated with BCR of MEHP‐induced PCa. (D) Heatmap of differentially expressed genes associated with recurrence in MEHP‐induced PCa. (E) PPI network of target genes in MEHP‐induced PCa. (F) Enrichment analysis of differentially expressed genes.

3.2. Construction of a Comprehensive Consensus Feature Using Machine Learning and Validation of Its Predictive Performance and Prognostic Value Through ROC Curves

We combined 117 machine learning algorithm combinations to analyze the 152 genes associated with BCR identified through univariate Cox regression analysis (Figure 2A). These algorithms were applied to the TCGA‐PCa cohort and three external validation datasets (Taylor, CIT, and Stockholm) to determine the optimal model (based on the highest average C‐index across the four cohorts). Among the 117 models, algorithm combinations were ranked based on the average C‐index across the four cohorts, and the lasso + RSF algorithm with the highest average C‐index was selected. This led to the identification of 16 of the most valuable MEHPRS genes (Figure 2B), namely, AMH, ASPN, FOXS1, VGF, APLN, SCN4A, MKI67, COMP, FOSB, CTHRC1, CHD5, OPCML, ALDH2, ISYNA1, LAMA3, and GABRA3. Additionally, we performed Kaplan–Meier survival analysis for AMH, ASPN, FOXS1, APLN, and VGF (see Figure S2), which suggests that the levels of these markers may serve as valuable biomarkers for predicting survival prognosis in PCa patients. Analysis of PCa patients in the TCGA‐PCa, Taylor, CIT, and Stockholm datasets revealed a significant association between high MEHPRS expression and reduced survival time (Figure 2C). Additionally, to evaluate the predictive performance and prognostic value of the four cohorts, we plotted ROC curves. The AUC values for the TCGA‐PCa, Taylor, CIT, and Stockholm datasets were 0.956, 0.764, 0.642, and 0.713, respectively, suggesting that MEHPRS has a strong prognostic value.

FIGURE 2.

Development of a consensus feature using machine learning and assessment of its predictive and prognostic performance via ROC curve analysis. (A) Heatmap comparing the performance of 117 machine learning models in predicting PCa prognosis. (B) Evaluation plot for the RSF model and feature importance plot for MEHP‐related genes selected by the lasso + RSF survival model, showing the top 16 most important MEHPRS. (C) Kaplan–Meier survival curve for MEHP in PCa patients. (D) ROC curves for the predictive models across four different PCa patient cohorts (TCGA, Taylor, CIT, and Stockholm).

3.3. Identification of the Role of Key MEHPRS Molecules in the Malignant Progression of PCa at the Single‐Cell Level

To investigate the role of key molecules in MEHPRS in the malignant progression of PCa, we retrieved the GSE137829 single‐cell RNA sequencing (scRNA‐seq) dataset from the GEO database, which includes expression profiles of 20,174 cells from six PCa patients. We performed QC (see Figure S1A,B) and removed batch effects from this single‐cell dataset (see Figure S1C). Based on the marker genes of different cell types (Chan et al. 2022; Chen et al. ²⁰²¹; D. Wang et al. ²⁰²³) (see Table S2), we annotated these cells into eight major clusters (Figure 3A), including B cells, endothelial cells, epithelial cells, fibroblasts, mast cells, myeloid cells, myofibroblasts, and T cells. The bubble plot shows the key marker genes for each cell group (Figure 3B). Next, we extracted epithelial cells for further analysis and used principal component analysis (PCA) for dimensionality reduction and clustering, which divided the epithelial cells into 10 subgroups (Figure 3C). To differentiate malignant cells, we used inferCNV to calculate and identify large‐scale chromosomal CNV based on the transcriptomic data from each sample, highlighting the intratumor heterogeneity in PCa. The inferCNV clustering heatmap generated for each sample (Figure 3D) shows T cells (as normal reference cells) in the upper part and the normalized expression values of epithelial cells in the lower part. In the heatmap, gain regions are represented in red, and loss regions are shown in blue. We observed that the CNV levels in Clusters 0, 4, 5, 7, and 10 were relatively high, while those in Clusters 1, 3, 6, and 9 were lower. Therefore, we classified Clusters 0, 4, 5, 7, and 10 as malignant epithelial cells (epi_malignant), while Clusters 1, 3, 6, and 9 were classified as benign epithelial cells (epi_benign) (Figure 3E). Subsequently, we conducted trajectory analysis on the epi_benign and epi_malignant cells using monocle2. The results of the pseudotime trajectory plot indicated that the epi_benign cells were at an earlier developmental stage, while the epi_malignant cells were at a later stage (Figure 3F). Finally, the heatmap shows the expression changes of the top 100 most significantly altered genes in the malignant progression of PCa from epi_benign (left) to epi_malignant (right) (Figure 3G).

FIGURE 3.

Single‐cell analysis of key MEHPRS molecules in PCa malignant progression. (A) Single‐cell landscape of the GSE137829 dataset, with the UMAP plot showing eight distinct cell types. (B) Bubble plot illustrating key marker genes annotated for each cell cluster. (C) UMAP plot showing eight subclusters of epithelial cells. (D) CNV clustering heatmap for epithelial cells and T cells based on inferCNV, with red indicating gain regions and blue indicating loss regions. (E) UMAP clustering plot of epi_malignant and epi_benign. (F) Pseudotime trajectory plot for epithelial cells, showing developmental time points for epi_benign and epi_malignant cells. (G) Heatmap of the top 100 significant genes with expression changes during the transition from epi_benign to epi_malignant.

3.4. Construction of an Environmental Pollutant–Key Genes–PCa Network

After constructing the MEHPRS genes using machine learning algorithms, we selected the top five key genes that contributed most to the risk features, AMH, ASPN, FOXS1, APLN, and VGF (Figure 2B), for the construction of the environment pollutant–key gene–PCa network. Figure 4A–E illustrates the relationships between these five key genes and environmental pollutants. To further investigate the relationship between MEHP and PCa, we constructed an environment pollutant–key gene–PCa network, aiming to identify the key factors most related to MEHPRS by analyzing the interactions between environmental pollutants and key genes. Based on CTD data, the results revealed that 19 environmental pollutants were associated with these key genes (Figure 4F). Additionally, we identified pollutants that were associated with more than three shared target genes and compiled a comprehensive list of all environmental pollutants (see Tables 1 and S3). After constructing MEHPRS signature genes using machine learning algorithms, we selected the top five key genes that contributed the most to the risk profile: AMH, ASPN, FOXS1, APLN, and VGF (Figure 2B). To further investigate the relationship between MEHP and PCa, we constructed an environmental pollutant–key gene–PCa network. This network aimed to identify the key factors most relevant to MEHPRS by analyzing the interactions between environmental pollutants and key genes. Networks illustrate the relationships between these five key genes and environmental pollutants (Figure 4A–E). Additionally, we identified environmental pollutants associated with more than three shared target genes (Table 1) and compiled a comprehensive list of all environmental pollutants (Table S3).

FIGURE 4.

Construction of the environmental pollutants–key genes–PCa network. (A–E) Interaction analysis of EDCs and genes. (F) Prediction of the EDCs–gene–PCa network.

TABLE 1.

A list of EDCs associated with three or more key interaction genes related to PCa.

	Chemical name	Interacted key gene count	Interacted key gene
1	Benzo(a)pyrene	5	AMH, ASPN, FOXS1, APLN, VGF
2	Bisphenol A	5	AMH, ASPN, FOXS1, APLN, VGF
3	Cadmium chloride	5	AMH, ASPN, FOXS1, APLN, VGF
4	Estradiol	5	AMH, ASPN, FOXS1, APLN, VGF
5	Mono‐(2‐ethylhexyl) phthalate	5	AMH, ASPN, FOXS1, APLN, VGF
6	Bisphenol S	4	AMH, ASPN, FOXS1, VGF
7	Sodium arsenite	4	AMH, FOXS1, APLN, VGF
8	Resveratrol	3	AMH, FOXS1, APLN
9	Ethinyl estradiol	3	AMH, APLN, VGF

3.5. Molecular Docking of AMH, ASPLN, ASPN, FOXS1, and VGF With MEHP

Molecular docking was performed to evaluate the potential binding interactions between MEHP and the genes AMH, ASPN, FOXS1, APLN, and VGF. The structure, depicted in warm pink at the center of the binding plot, represents MEHP, while the rest corresponds to the protein binding sites (Figure 5A–E). Typically, binding energy is used to evaluate the binding stability between a ligand and its receptor. In general, a binding energy of < 0 kcal/mol indicates that the receptor and ligand can spontaneously bind without external energy, while a binding energy of less than −5 kcal/mol suggests a strong binding. The results show that MEHP exhibits a weaker binding affinity for specific amino acid residues of AMH (LYS‐534, GLU‐84, and ASN‐486) and VGF (GLN‐338), with hydrogen bond interactions facilitating the binding. The molecular docking energy of MEHP with AMH is −4.8 kcal/mol, while that with VGF is only −4.0 kcal/mol (see Table S4). MEHP shows a stronger binding affinity for specific amino acid residues of APLN (GLY‐122), FOXS1 (ARG‐112 and PHE‐114), and ASPN (ASN‐127 and HIS‐172). The molecular docking binding energies of MEHP with APLN and FOXS1 are both −6.1 kcal/mol, while those with ASPN are −5.1 kcal/mol (see Table S4).

FIGURE 5.

Docking results of AMH, ASPLN, ASPN, FOXS1, and VGF with MEHP. (A–E) Display of docking results between AMH, ASPLN, ASPN, FOXS1, VGF, and MEHP.

4. Discussion

Environmental pollutants, classified as potential carcinogens, have been the focus of extensive research in recent years. For example, long‐term exposure to environmental BPA is closely associated with a high risk of PCa (Prins et al. 2018). Similarly, exposure to other environmental pollutants, such as PFOS (Thomas et al. 2023), dust/debris (Gong et al. 2019), chlordan (Multigner et al. 2010), Agent Orange (Lui et al. 2023), and pesticides (Brureau et al. 2021), is also associated with the aggressive development of PCa. Recent studies have increasingly suggested that environmental pollutants, especially EDCs, may play a role in the initiation and progression of PCa by disrupting hormone signaling, altering gene expression profiles, and modulating epigenetic regulation mechanisms (Chernychenko et al. 2020).

PAEs, notably DEHP and its metabolite MEHP, are widespread EDCs that have been shown to disrupt androgen signaling pathways (Y. Zhang, Guo, et al. ²⁰²¹) and to induce oxidative stress while compromising antioxidant defense mechanisms, potentially contributing to the acceleration of PCa progression (Erkekoğlu et al. 2010; P. Song et al. ²⁰²⁴; Thomas et al. ²⁰²³).

DEHP is extensively utilized in the manufacturing of consumer products, including plastics, flooring, furniture, clothing, and packaging materials. These products may release DEHP into the environment during their production, usage, and disposal, resulting in extensive contamination of water, soil, and air (Y. Zhang, Guo, et al. ²⁰²¹). Research indicates that DEHP has a significant bioaccumulation potential, posing substantial risks to aquatic life and human health, particularly adversely affecting the reproductive and endocrine systems (Eales et al. 2022). Although the environmental contamination and toxicity of parent PAEs have been extensively documented, research on their metabolites remains relatively scarce.

MEHP, the principal metabolite of DEHP, has garnered significant attention because of its potential environmental and health hazards. MEHP has been identified in various environmental media, including medical PVC bags, plastic containers (Wowkonowicz 2023), water within PVC pipes (Sanjuan et al. 2023), and soil (A. Wang et al. ²⁰¹²). Research indicates that MEHP exhibits substantial environmental persistence and bioaccumulation, particularly within aquatic ecosystems. For instance, MEHP concentrations in the Tama River, Japan, display seasonal fluctuations, with higher levels observed in winter (Suzuki et al. 2001). The toxicological effects of MEHP on various organisms are well established, as it disrupts liver metabolism, alters cell proliferation, and induces genotoxicity (Park et al. 2020). It may also disrupt reproductive function in aquatic organisms by modulating endocrine pathways. MEHP degrades slowly in aquatic environments, and its metabolites, such as MEHHP and MEOHP, degrade at comparable rates (Suzuki et al. 2001). In soil, MEHP accumulation can significantly alter bacterial communities responsible for organic matter decomposition, particularly by inhibiting Tumebacillus species (F. Zhu, Zhu, et al. ²⁰¹⁸).

Although the toxicological effects of MEHP have been consistently demonstrated in animal models, the long‐term implications of human exposure, particularly its potential risks to prostate health, remain insufficiently characterized. Under experimental conditions, MEHP has been shown to significantly alter prostate cell physiology by disrupting hormone signaling and inducing oxidative stress, thereby exerting toxic effects on multiple tissues and organs. MEHP plays a pivotal role in the malignant progression of PCa (Cavalca et al. 2022; Erkekoğlu et al. ²⁰¹⁰), particularly by promoting the oncogenic transformation of prostate cells. Moreover, MEHP may exacerbate the progression of PCa by altering gene expression and epigenetic regulation (P. Song et al. ²⁰²⁴). Studies have demonstrated that MEHP can alter DNA methylation profiles in PCa cells, thereby influencing the expression of tumor‐related genes and promoting cancer cell growth and metastasis.

PCa is the most common malignancy of the male urogenital system and ranks second in terms of mortality (Siegel et al. 2025). The onset and progression of PCa are significantly regulated by hormones, with androgens acting as key drivers of its pathogenesis. Androgens bind to AR, thereby activating downstream signaling cascades that promote the proliferation, survival, and migration of PCa cells (Dai et al. 2017; Luo et al. ²⁰²⁰). In the early stages of PCa, the tumor's dependence on androgens is crucial for its growth. Furthermore, other hormones, such as estrogen, insulin‐like growth factor (IGF), and stress hormones like cortisol, also contribute to the initiation and malignant progression of PCa. Estrogen, which is converted from androgens by aromatase, regulates the expression of ERα during the malignant progression of PCa, thereby influencing the tumor's resistance to treatment (Mak et al. 2015; Ricke et al. ²⁰⁰⁸). IGF enhances PCa cell proliferation through interactions with AR (Matsushita et al. 2022), whereas hormones such as cortisol accelerate tumor progression under stress conditions by modulating cell proliferation and apoptosis (Fabre et al. 2016).

Studies have shown that DEHP and its metabolite MEHP may promote benign prostatic hyperplasia (BPH) by upregulating KIF11 and activating the Wnt/β‐catenin signaling pathway, which subsequently enhances H3K27ac modification, revealing the potential mechanism by which environmental pollutants affect prostate health (P. Song et al. ²⁰²⁴). Moreover, MEHP may promote the development of prostate disease by disrupting lipid metabolism, oxidative stress, inflammation, and other key biological processes in prostate stromal cells and metastatic prostate epithelial cells (Thomas et al. 2023). Furthermore, phthalates (DEHP and its metabolite MEHP) have been found to induce cytotoxicity, DNA damage, and an imbalance in antioxidant status in LNCaP PCa cells through oxidative stress (Erkekoğlu et al. 2010). These studies suggest that MEHP may promote PCa development by disrupting oxidative stress in PCa cells, which in turn impairs intracellular antioxidant enzyme activity (Erkekoğlu et al. 2010; Thomas et al. ²⁰²³). At the same time, phthalate metabolites have been found to promote tumor expansion and invasiveness by increasing cell turnover, oxidative stress, miRNA biosynthesis, and cell migration potential in LNCaP PCa cells (Cavalca et al. 2022). On the other hand, MEHP may further promote PCa progression by lowering the proportion of 5‐mC methylated cytosine in LNCaP cells, affecting genome‐wide DNA methylation (Wu et al. 2017). Finally, MEHP promotes the proliferation and progression of LNCaP PCa cells by activating the Hedgehog (Hh) signaling pathway, enhancing PTCH gene expression, and interacting with the androgen signaling pathway, revealing its potential mechanisms in PCa initiation and progression (Yong et al. 2016). In these studies (Cavalca et al. 2022; Erkekoğlu et al. ²⁰¹⁰, P. Song et al. ²⁰²⁴; Thomas et al. ²⁰²³; Wu et al. ²⁰¹⁷; Yong et al. ²⁰¹⁶), MEHP exposure led to alterations in multiple key signaling pathways, which may contribute to the onset and progression of prostate disease. In our study, KEGG enrichment analysis results suggest that MEHP may play a critical role in the onset and progression of PCa by affecting focal adhesion, extracellular matrix–receptor interactions, the cAMP signaling pathway, cell cycle, and AMPK signaling pathway. These results are consistent with previous studies and also identify several potential pathogenic pathways. Although previous studies have revealed the potential role of MEHP in mediating PCa progression through various molecular mechanisms, no study has yet systematically analyzed how MEHP promotes PCa progression through multiple layers and pathways. To fill this gap, we comprehensively evaluated the potential of MEHP in promoting PCa development by integrating cutting‐edge technologies such as network toxicology, molecular docking, machine learning, and single‐cell analysis, further revealing its multidimensional mechanisms in cancer initiation.

Initially, we retrieved 2863 target genes associated with MEHP in PCa from the CTD database. Next, we intersected the MEHP target genes with the pathogenic differential genes of PCa, resulting in 476 target genes. Using univariate COX analysis and PPI network analysis, we initially identified 152 target genes.

Furthermore, KEGG enrichment analysis revealed that MEHP may drive the onset and progression of PCa through several key signaling pathways, primarily involving cell migration, proliferation, signal transduction, and metabolic regulation. Specifically, focal adhesion and extracellular matrix–receptor interactions may play a critical role in cancer cell invasion and metastasis (Figel and Gelman 2011; Kiss et al. ²⁰¹³), while cell cycle activation could accelerate cell proliferation (Brot and Mongan 2018). In addition, the cAMP signaling pathway may facilitate the malignant progression of PCa by enhancing AR activity (Merkle and Hoffmann 2011), while the AMPK signaling pathway may support cancer cell survival through energy metabolism regulation (White et al. 2018). The interplay of these pathways may constitute the key mechanism by which MEHP induces PCa development, offering a theoretical foundation for future studies on its molecular mechanisms and potential therapeutic strategies.

Next, we constructed a prognostic model related to MEHP pathogenesis using a combination of various machine learning techniques and validated its predictive performance in an external validation set through ROC analysis. We identified the key genes contributing most to the risk model, namely, AMH, ASPN, FOXS1, APLN, and VGF, which were defined as hub genes. Additionally, we employed single‐cell developmental trajectory and copy number analyses to identify changes in key molecules associated with the malignant progression of PCa.

Furthermore, we constructed an environmental pollutant–key gene–PCa network to systematically analyze the interactions between environmental pollutants and key genes and identify the factors most strongly associated with MEHP. The results revealed that 19 environmental pollutants were associated with the hub genes, with MEHP, benzo(a)pyrene (BaP), BPA, cadmium chloride (CdCl₂), and estradiol (E₂) identified as the most significant potential environmental pollutants. BaP, a crystalline PAH, exhibits high lipophilicity and hydrophobicity, facilitating its bioaccumulation in the food web (Fu et al. 2022). Studies suggest that BaP promotes the proliferation, mutation, and invasion of PCa cells by inducing DNA damage (Nwagbara et al. 2007), modulating the expression of oxidative stress–related proteins (Chaudhary et al. 2007) and through the activation of the JAK2/STAT3 signaling pathway (Gao et al. 2020). BPA is a synthetic organic compound produced from acetone and phenol, widely utilized in the manufacturing of epoxy resins and polycarbonate plastics (Kang et al. 2006). Studies indicate that BPA accelerates PCa progression and potentially influences treatment resistance by regulating AR signaling, promoting PCa cell proliferation and invasion (J. Zhang ²⁰¹⁸), and modulating key genes such as COL1A1 and COL1A2 (Liu et al. 2022). Cadmium (Cd), found in CdCl₂, is a nonessential element with potential toxicity and carcinogenic properties (Wright and Welbourn 1994). CdCl₂ promotes PCa initiation and progression by inducing prostate cell proliferation and inhibiting apoptosis (Arriazu et al. 2005). Estradiol (E₂) is a primary estrogen, with contamination mainly originating from sewage treatment plants, wastewater facilities, and endocrine‐disrupting compounds in fecal discharge (Adeel et al. 2017). Studies suggest that E₂ levels are closely correlated with PCa severity and may serve as a potential biomarker for predicting its progression (Salonia et al. 2011). Through the construction of the environmental pollutant–key gene–PCa network, we identified that MEHP, BaP, BPA, CdCl₂, and E₂ synergistically regulate hub genes, promoting the onset and progression of PCa.

To elucidate the mechanism underlying MEHP‐induced PCa, we conducted molecular docking analysis on five hub genes: AMH, ASPN, FOXS1, APLN, and VGF. The results revealed that these key genes bind effectively with MEHP, with the binding energies for APLN, FOXS1, and ASPN genes being lower than −5 kcal/mol. APLN, FOXS1, and ASPN genes all demonstrate favorable binding energies and high binding affinities with MEHP, indicating that these genes are primary targets of MEHP. Furthermore, the formation of multiple hydrogen bonds between MEHP and the specific amino acid residue HIS‐172 in ASPN may account for the higher binding energy observed in the molecular docking, likely contributing to increased stability of the docking interaction. However, the precise mechanism by which MEHP regulates the expression of hub genes to drive PCa remains to be further explored, in order to elucidate its role in the onset and progression of PCa.

Anti‐Müllerian hormone (AMH) is a glycoprotein hormone secreted by granulosa cells in the ovaries. In the study by Ma et al. (2022), the research team developed a gene signature model incorporating four chromatin regulatory genes (CRGs) to predict relapse‐free survival (RFS) in PCa patients. Among these, AMH is one of the four CRGs, suggesting its significant potential in predicting RFS in PCa patients. This finding also suggests that AMH could serve as a potential biomarker for personalized treatment in PCa patients (Ma et al. 2022), offering new directions for precision medicine strategies. Furthermore, AMH has been recognized as an effective predictor of BCR. Thus, AMH could serve as a promising prognostic biomarker for disease invasiveness, providing novel reference points for risk stratification and personalized treatment strategies in PCa patients (Kontogiannis et al. 2024). A study identified AMH as a high‐risk gene, with its expression significantly elevated in PCa tissues, which was closely linked to poor survival prognosis (Fei and Chen 2022). Other studies have indicated that PCa patients with low AMH expression levels have longer survival and lower recurrence rates, while higher AMH expression promotes PCa progression (L. Wang et al. ²⁰²¹). Furthermore, AMH is closely linked to female ovarian hormones, such as estrogen and plays a pivotal role in regulating the male endocrine system, particularly during embryonic development (Rey et al. 2003). Collectively, these findings suggest that AMH could serve as a potential biomarker for elucidating the mechanism through which MEHP regulates androgen secretion, thereby facilitating the onset and progression of PCa.

VGF is a neurosecretory protein that plays a crucial role in the radioresistance of PCa. Studies have shown that high expression of VGF is strongly associated with radioresistance in PCa cells, especially in the DU145 and LNCaP cell lines, where changes in its expression levels can significantly impact the effectiveness of radiation therapy (Seifert et al. 2019). This finding implies that VGF could act as a potential regulatory factor of radioresistance, offering novel avenues for optimizing radiation therapy strategies for PCa (Seifert et al. 2019). In Roosa Kaarijärvi et al.'s (2021) study, VGF was found to be strongly associated with the neuroendocrine phenotype of PCa. Studies have shown that VGF interacts with neurodifferentiation factors such as SMARCA4, promoting the neuroendocrine transformation of cancer cells in PCa. Furthermore, high expression of SMARCA4 is strongly associated with increased invasiveness and shortened survival, indicating that VGF may play a crucial role in PCa progression by regulating this process (Kaarijärvi et al. 2021). These results indicate that VGF may play an important role in disease progression by regulating key signaling pathways associated with PCa and could serve as a potential therapeutic target or prognostic biomarker for PCa.

ASPN serves as a stromal marker in the tumor microenvironment and is strongly associated with the metastatic progression of PCa. In the study by Hurley et al. (2016), it was found that the ASPN D14 homozygous and D13/14 heterozygous genotypes were significantly associated with an increased risk of metastatic recurrence in PCa patients, while the D13 homozygous genotype was associated with a reduced risk, suggesting that the genetic polymorphisms of ASPN may play a critical role in PCa progression. Moreover, ASPN is highly expressed in mesenchymal stem cells (MSCs), where it inhibits MSC differentiation through interaction with BMP‐4. Studies have shown that loss of ASPN impairs the self‐renewal capacity of MSCs and diminishes essential cell populations within the tumor microenvironment, such as cancer stem cells and the tumor vasculature, thereby influencing tumor progression. Furthermore, elevated ASPN expression is strongly correlated with shorter BCR time, higher Gleason scores, and advanced clinical staging, particularly within reactive stroma, where its expression is more pronounced. This characteristic indicates that ASPN may play a crucial role in the stromal response of aggressive PCa subtypes (Rochette et al. 2017). In the study by P. Zhang, Qian, et al. (2021), ASPN was found to be significantly overexpressed in PCa, and higher ASPN expression was strongly associated with a poorer prognosis. Multivariate analysis further indicated that elevated ASPN expression serves as an independent prognostic factor affecting overall survival (OS) in patients. This discovery emphasizes the pivotal role of ASPN in the progression of PCa. The aforementioned studies suggest that ASPN expression in the tumor microenvironment may contribute to the metastasis and progression of PCa. Elevated ASPN expression is not only linked to metastatic recurrence and poor prognosis but also likely contributes to the remodeling of the tumor microenvironment. Thus, ASPN may act as a biomarker for predicting metastatic risk in PCa and serve as a potential therapeutic target for intervention.

FOXS1 is closely associated with the AR and other genes related to sex hormones. In the study conducted by Panneerdoss et al. ²⁰¹²), it was found that the testosterone‐responsive miR‐471 modulates the expression of Foxd1 and Dsc1, which are expressed in a testosterone‐dependent manner in mouse Sertoli cells. FOXA1 is widely considered a pioneer factor for the AR. It directly binds to target sites within repressive chromatin, initiating the genomic regulatory process ahead of other transcription factors, thereby facilitating chromatin remodeling and enhancing AR accessibility to its target genes (Copeland et al. 2019). FOXA1 is recognized as a key cofactor in the AR‐mediated progression of PCa. Whole‐genome analysis of PCa cell lines revealed approximately 70% overlap between FOXA1 and AR binding sites, indicating that FOXA1 plays a central role in AR‐mediated transcriptional regulation (Cirillo et al. 2002). Furthermore, FOXS1 is pivotal in the regulation of hormone‐related signaling pathways. Research indicates that FOXS1, functioning as an oncogene, upregulates HILPDA, thus activating the FAK/PI3K/AKT signaling cascade to promote epithelial–mesenchymal transition (EMT), thereby facilitating the invasion and metastasis of PCa (Ren et al. 2024). Additionally, FOXS1 further promotes PCa progression via the Hedgehog/Gli1 signaling pathway, suggesting that it may act as a critical regulator of the disease and serve as a new potential target for targeted therapy (M. Wang and Huang ²⁰²³). In summary, the aforementioned research underscores the indispensable role of FOXA1 in the transcriptional regulatory network of PCa, further elucidating the potential role of FOXS1 in the androgen regulatory network and offering a novel perspective on the molecular mechanisms of androgen regulation.

APLN is an endocrine peptide primarily secreted by adipose tissue, exerting a pivotal role in various physiological processes, particularly in energy metabolism, glucose homeostasis, and the pathophysiology of obesity. Recent studies have identified the APLN/APJ axis as a crucial mechanism underlying reproductive dysfunction in diabetic men. Research suggests that targeting the APLN/APJ axis holds substantial therapeutic promise, offering potential improvements in reproductive function among diabetic individuals and presenting novel therapeutic avenues for related diseases (K. Song et al. ²⁰²²). Concurrently, APLN‐F13A, a mutant ligand of APLNR, effectively suppresses tumor angiogenesis and inhibits the invasion of cancer cells. Research demonstrates that the combined targeting of VEGFR2 and APLNR significantly improves survival rates in murine models, further substantiating the synergistic interplay between the APLN/APLNR signaling axis and existing antiangiogenic therapies. Targeting this pathway not only mitigates treatment resistance but also enhances antitumor efficacy, thus offering promising avenues for the refinement of antiangiogenic strategies (Mastrella et al. 2019).

ML221, a small‐molecule inhibitor targeting APLN, effectively attenuates PI3K/Akt signaling, thereby inhibiting the growth of hepatocellular carcinoma (HCC). Studies indicate that the upregulation of APLN in HCC contributes to its oncogenic potential, promoting tumor cell proliferation and survival (Mastrella et al. 2019). Thus, targeting APLN presents a promising therapeutic strategy for HCC. To date, no studies have definitively established a direct connection between APLN and PCa, leaving this domain largely unexplored. This highlights the urgent need for future research into the role of the APLN/APJ axis in PCa, marking it a critical area of investigation. A thorough exploration of APLN's potential oncogenic mechanisms in PCa, along with its clinical utility as a therapeutic target, will not only advance our understanding of PCa biology but also open new avenues for precision treatment strategies.

All five hub genes identified in our study play a critical role in the progression of PCa. Although current research has not yet directly demonstrated the potential mechanisms by which these hub genes induce PCa, their involvement in the initiation and progression of the disease highlights their significance as potential biomarkers, offering important implications for new targeted therapeutic strategies.

This study utilizes network toxicology and molecular docking methodologies to investigate the potential toxic targets and mechanisms of MEHP in PCa, while also advancing the application of network toxicology in elucidating the pathogenic mechanisms of environmental toxicants. The integration of network toxicology with molecular docking methodologies enhances the efficiency, depth, and predictive accuracy of toxicological screening, thereby facilitating the evaluation of numerous underexplored environmental toxins and establishing it as a promising approach for toxicological analysis. This study provides the theoretical foundation for future in‐depth investigations into the mechanisms by which MEHP induces PCa.

5. Conclusion

In this study, we first retrieved MEHP‐related target genes in PCa from the CTD database. After intersecting these genes with PCa‐related differential genes, we identified MEHP target genes associated with biochemical recurrence of PCa using univariate Cox analysis and PPI analysis. Furthermore, KEGG enrichment analysis revealed that MEHP may promote PCa through focal adhesion, extracellular matrix–receptor interactions, the cAMP signaling pathway, cell cycle regulation, and the AMPK signaling pathway. Next, we constructed a MEHP‐related prognostic model and validated its predictive performance using ROC analysis on an external validation set. From the model, we identified the most significant key genes (AMH, ASPN, FOXS1, APLN, and VGF) as hub genes. Furthermore, we employed single‐cell analysis, including developmental trajectory and copy number analysis, to identify key molecular changes involved in the malignant progression of PCa. Furthermore, by identifying key genes associated with PCa and potential environmental pollutants, we constructed a chemical–gene–PCa relationship network that revealed 19 environmental pollutants potentially linked to PCa and identified the key factors most closely associated with MEHP. Notably, environmental pollutants that significantly impact the progression of PCa include cadmium, BPA, and BaP, all of which are important etiological factors in the development of PCa. Machine learning algorithms identified the hub genes as crucial genes closely associated with PCa. Molecular docking analysis of the interactions between MEHP and these hub genes further validated their potential. Specifically, APLN, FOXS1, and ASPN genes exhibit strong binding energies and high affinities for MEHP, suggesting that these genes are primary targets of MEHP molecules. This study provides a theoretical foundation for future investigations into the effects of environmental pollutant MEHP exposure on PCa, offering valuable clinical strategies and guidance for environmental health research in PCa prevention and treatment. While our study offers new perspectives on the association between MEHP and PCa, it still has several limitations. First, the study relies on observational expression patterns of biomarkers, which do not allow for inferring causality. Secondly, the microarray data should be validated using independent methods, such as quantitative PCR, Western blotting, or immunohistochemistry. Additionally, the findings need to be validated in larger independent cohorts and further corroborated by epidemiological studies.

Author Contributions

Chenyu Liang: writing – original draft, visualization, methodology, data management. Weicheng Tian: writing – review and editing, writing – original draft. Hengxi Zeng: writing – original draft, data management. Ziyang Xia: visualization, software. Zijie Luo: visualization, software. Yue Zhuo: writing – original draft. Minlian Pan: writing – original draft. Kangbu Wu: data management. Siyu Xiong: data management. Xuejing Lin: writing – review and editing, supervision, conceptualization. Xinchun Li: writing – review and editing, supervision, conceptualization. Jiaxi Yu: writing – review and editing, supervision, conceptualization.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1. Quality control and batch effect removal in single‐cell datasets. (A) Before the quality control step for the GSE137829 dataset. (B) After the quality control step for the GSE137829 dataset. (C) Carry out batch effect adjustment on the samples post‐quality control.

Figure S2. Kaplan–Meier survival analysis of AMH, ASPN, FOXS1, APLN, and VGF.

Table S1. Data on genes associated with MEHP are presented.

Table S2. Marker genes for different cell types are presented.

Table S3. A comprehensive list of 19 environmental pollutants associated with key genes related to MEHP.

Table S4. The binding energy of MEHP docking with hub genes.

Liang, C. , Tian W., Zeng H., et al. 2025. “Revealing the Impact of Mono(2‐ethylhexyl) Phthalate (MEHP) on Prostate Cancer Based on Network Toxicology and Molecular Docking Approaches.” Journal of Applied Toxicology 45, no. 10: 2078–2094. 10.1002/jat.4826.

Data Availability Statement

Data will be made available on request.