Identification of Shared Hub Genes between Endometriosis and Ovarian Clear Cell Carcinoma Using Integrative Bioinformatic Analysis

Endometriosis is characterized by the presence of tissue resembling the uterine lining outside the uterus ( 1 ). Although the precise cause of endometriosis remains elusive, its high prevalence among women worldwide underscores the need for further research into its pathogenesis, contributing factors, and the identification of additional biomarkers ( 2 ). Although various factors have been associated with susceptibility to endometriosis, the relationship between women's fertility status and this condition remains unclear. While endometriosis itself does not directly induce cancer, it has been associated with an increased risk of ovarian clear cell carcinoma (OCCC) ( 3 ). OCCC is a specific subtype of epithelial ovarian cancer that has been linked to endometriosis in epidemiological and molecular studies. Since infertility frequently accompanies advanced endometriosis, we hypothesized that comparing gene expression profiles between fertile and infertile women with endometriosis, followed by intersection with OCCC -associated genes, might reveal shared molecular features. This approach aims to generate hypotheses regarding biomarkers that may be associated with both infertility -related endometriosis and OCCC, potentially informing future risk assessment and early detection strategies. We acknowledge that direct comparisons between fertile and infertile women with endometriosis would be more appropriate for isolating fertility-specific effects, and such analyses are suggested for future studies ( 4 ). The identification of biomarkers linking women’s fertility status with endometriosis, and subsequently represent a pivotal advancement in the early detection of this cancer and in distinguishing it from endometriosis. This bioinformatic study seeks to address this knowledge gap. Using bioinformatics tools, we aimed to identify potential biomarkers that distinguish between fertile and infertile women with stage IV endometriosis. Furthermore, we evaluated these biomarkers to investigate their potential association with OCCC. By exploring the molecular landscape of endometriosis in relation to fertility status, we aimed to provide insights that may improve risk assessment and potentially contribute to the development of novel therapeutic approaches for OCCC, particularly in women with endometriosis.

Our analysis identified distinct sets of DEGs among the study groups. In the GSE120103 dataset, which included tissue samples from 9 disease-free fertile women, 9 disease-free infertile women, 9 fertile women with stage IV endometriosis, and 9 infertile women with stage IV endometriosis, a significantly higher number of DEGs was identified in infertile women with stage IV endometriosis (4,223 genes) than in fertile women with stage IV endometriosis (1390 genes) ( Table 1 ). In the GSE29450 dataset, which included tissue samples from 10 primary OCCC tumors and 10 normal ovarian surface epithelia, comparison of OCCC tissues with normal ovarian surface epitheliaidentified2,762 DEGs ( Table 1 ). To further examine the relationship among these gene sets, a Venn diagram was generated to illustrate the overlapping and unique DEGs identified in each comparison. In addition, volcano plots, heatmaps, and comprehensive lists of DEGs and hub genes are provided in Tables S1, S2 , and Figures S1-S6 (See Supplementary Online Information at www.ijfs. ir ) confirming the distinct expression profiles observed among the study groups. Characteristics of the GSE dataset included in this study and the numbers of differentially expressed genes (DEGs) The enriched KEGG pathways included Wnt, Notch, MAPK, PI3K/Akt, and TGFB signaling pathways. GO enrichment analysis showed that these DEGs were significantly associated with immune response, cell proliferation, angiogenesis, cell adhesion, and cell migration (Figes. 1 - 3 ). Functional enrichment analysis of DEGs in fertile women with stage IV endometriosis. DEGs identified infertile women with stage IV endometriosis from the GSE120103 dataset (n=9) were subjected to functional enrichment analysis. A. KEGG pathway enrichment analysis. B. GO biological process enrichment analysis. C. GO molecular function enrichment analysis. DEGs; Differentially expressed genes, KEGG; Kyoto Encyclopedia of Genes and Genomes, and GO; Gene Ontology. Functional enrichment analysis of DEGs in infertile women with stage IV endometriosis. DEGs identified in infertile women with stage IV endometriosis from the GSE120103 dataset (n=9) were subjected to functional enrichment analysis. A. KEGG pathway enrichment analysis. B. GO biological process enrichment analysis. C. GO molecular function enrichment analysis. DEGs; Differentially expressed genes, KEGG; Kyoto Encyclopedia of Genes and Genomes, and GO; Gene Ontology. Functional enrichment analysis of DEGs in ovarian clear cell carcinoma. DEGs identified between ovarian clear cell carcinoma tissues (n=10) and normal ovarian surface epithelium samples (n=10) from the GSE29450 dataset were subjected to functional enrichment analysis. A. KEGG pathway enrichment analysis. B. GO biological process enrichment analysis. C. GO molecular function enrichment analysis. DEGs; Differentially expressed genes, KEGG; Kyoto Encyclopedia of Genes and Genomes, and GO; Gene Ontology. For further analysis, two gene groups were selected. Group A consisted of 481 genes shared between the DEGs from infertile women with endometriosis in GSE120103 and the DEGs from GSE29450 , whereas Group B consisted of 123 genes shared between the DEGs from fertile women with endometriosis in GSE120103 and the DEGs from GSE29450 . KEGG and GO enrichment analyses revealed distinct functional characteristics for the two groups. Group A genes were enriched in cancer-related, inflammatory and metabolic pathways, including the PI3K-Akt signaling pathway. Group B genes were primarily enriched in immune response, stem cell signaling, and cancer -related pathways (Figes .4 , 5 ). Functional enrichment analysis of shared DEGs between infertile endometriosis and ovarian clear cell carcinoma. Common DEGs between infertile women with stage IV endometriosis from GSE120103 (n=9) and ovarian clear cell carcinoma samples from GSE29450 (n=10) were subjected to functional enrichment analysis. A. KEGG pathway enrichment analysis. B. GO biological process enrichment analysis. C. GO molecular function enrichment analysis. DEGs; Differentially expressed genes, KEGG; Kyoto Encyclopedia of Genes and Genomes, and GO; Gene Ontology. Functional enrichment analysis of shared DEGs between fertile endometriosis and ovarian clear cell carcinoma. Common DEGs between fertile women with stage IV endometriosis from GSE120103 (n=9) and ovarian clear cell carcinoma samples from GSE29450 (n=10) were subjected to functional enrichment analysis. A. KEGG pathway enrichment analysis. B. GO biological process enrichment analysis. C. GO molecular function enrichment analysis. DEGs; Differentially expressed genes, KEGG; Kyoto Encyclopedia of Genes and Genomes, and GO; Gene Ontology. PPI network analysis revealed distinct module organizations between the two groups. In group A, eight significant modules were identified, including five modules containing three nodes each, 2 two modules containing four nodes each, and one module containing five nodes. Group B exhibited six significant modules, consisting of one module with three nodes, two modules with four nodes each, one module with 12 nodes, and one module with 10 nodes. These findings indicate differences in network organization between the two groups, suggesting potential variations in the underlying biological processes and regulatory mechanisms ( Figes.S3, S4 , See Supplementary Online Information at www.ijfs.ir ). Hub gene analysis identified 17 hub genes in group A, including IGF1, FGF1, FGFR3, EPHX2, INA, BIRC5, PPIG, KIF20A, CYP2C9, NES, TLR4, RHO, SOX2, LY96, FCGR2A, S100B, and e In group B, 12 hub genes were identified, including SLC17A8, HMGA2, CCND2, MSRB3, NPY, SFRP1, MDM2, CYP19A1, NCOA3, CD86, CDC25A, and CXCL13 ( Tables S1 and S2 , See Supplementary Online Information at www.ijfs.ir ). Survival analysis revealed that two of the 17 hub genes in Group A ( BIRC5 and EPHX2 ) and one of 12 hub genes in Group B ( NCOA3 ) were significantly associated with overall survival in the OV cohort ( Figes.S5,6 , See Supplementary Online Information at www.ijfs.ir ). These findings suggest that these hub genes may have potential prognostic relevance and warrant further validation in independent cohorts and experimental studies

Endometriosis is chronic gynecological disease characterized by the presence of endometrial-like tissue outside the uterus ( 1 ). It affects approximately 10-15% of women of reproductive age and up to 30-50% of women with pelvic pain or infertility ( 8 ). Clinical presentation is highly variable, and some affected women remain asymptomatic, contributing to delays in diagnosis ( 8 ). Given the absence of accurate noninvasive diagnostic methods, the identification of reliable biomarkers remains an important research priority. OCCC represents a distinct subtype of epithelial ovarian cancer, accounting for only approximately 5-10% of all cases ( 4 ). Unlike the more common high-grade serous subtype, OCCC is thought to arise from a different cell of origin in the ovary and exhibits distinct clinicopathological characteristics ( 9 ). Although less common, OCCC can be an aggressive form of ovarian cancer and is often diagnosed at an advanced stage due to its relatively subtle clinical presentation. The exact etiology of OCCC is unclear. However, several studies suggest a possible association with endometriosis ( 10 ). However, given that OCCC can be more resistant to standard chemotherapy than other subtypes of ovarian cancer, early detection is critical for improving treatment outcomes ( 11 ). Enrichment analysis of DEGs identified in fertile women with stage IV endometriosis showed significant associations with various biological processes and molecular functions. KEGG pathway analysis revealed several enriched pathways, including the Wnt signaling pathway and steroid hormone biosynthesis. Dysregulation of these pathways may influence cell proliferation, differentiation, and migration, thereby contributing to the abnormal endometrial growth. The Wnt signaling pathway, which plays a fundamental role in tissue development and homeostasis, has been implicated in endometriosis Similarly, steroid hormone biosynthesis is essential for regulating the menstrual cycle and endometrial function, and alterations in this pathway can contribute to disease development. GO enrichment analysis further indicated the involvement of these genes in epithelial structure maintenance, digestive epithelium maintenance, nervous system development, immune system regulation, steroid hormones biosynthesis, cell communication, response to stimuli, and growthrelated processes. Molecular function analysis also highlighted their involvement in hormone-related activities. Enrichment analysis of DEGs identified in infertile women with stage IV endometriosis revealed multiple biological pathways potentially associated with endometriosis, infertility, and OCCC. KEGG enrichment analysis revealed several enriched pathways, including cytokine-cytokine receptor interaction, chemokine signaling, PI3K-Akt, MAPK, and Notch signaling pathways. Cytokine-cytokine receptor interaction and chemokine signaling play essential roles in immune cell recruitment and activation and are closely associated with chronic inflammatory responses, a hallmark of both endometriosis and infertility. The PI3K-Akt and MAPK signaling pathways regulate cell growth, proliferation, and differentiation and have been implicated in the pathogenesis of both endometriosis and OCCC. GO enrichment further indicated the involvement of these genes in angiogenesis, immune response, cellular processes related to proliferation and adhesion. Molecular function analysis also highlighted enrichment in growth factor activity, cytokine receptor binding, hormone activity, and angiogenesis. KEGG enrichment analysis of DEGs between OCCC and normal ovarian surface epithelia identified several significantly enriched pathways, including pathways in cancer, thePI3K-Akt signaling pathway, the TGFβsignaling pathway, cell adhesion molecules, extracellular matrix–receptor (ECM-receptor) interaction, the cell cycle, the p53 signaling pathway, and focal adhesion. Genes involved in the PI3K-Akt and TGF-βsignaling pathways play important roles in cell growth, proliferation, survival, and immune regulation. In addition, genes involved in cell adhesion molecules, ECM-receptor interaction, and focal adhesion contribute to cell-cell and ECM interactions and may promote ovarian cancer cell migration and metastasis. GO biological process enrichment further indicated associations with cell cycle regulation, cell adhesion, and cell migration, response to estrogen, and immune response regulation. The BIRC5 gene encodes the survivin protein, which is involved in cell division and cell survival ( 12 ). It plays an essential role in mitosis, by ensuring accurate chromosome segregation during cell division ( 12 ). However, survivin exerts anti-apoptotic functions that may promote the survival of damaged or abnormal cells, thereby contributing to tumor development. ( 11 ). The role of BIRC5 in fertility remains under investigation. Dysregulated BIRC5 expression has been associated with issues related to oocyte maturation and embryo implantation, suggesting a potential role in female fertility ( 13 - 15 ). Nevertheless, the underlying mechanisms linking BIRC5 to human fertility have not yet been fully elucidated. Increased BIRC5 expression has been reported in ectopic endometrial tissues from patients with endometriosis, suggesting a potential role in the abnormal proliferation and maintenance of these displaced endometrial lesions ( 16 ). Furthermore, the anti-apoptotic function of BIRC5 may contribute to the resistance of these ectopic lesions to therapeutic interventions ( 16 ). Elevated BIRC5 expression has also been associated with a more aggressive OCCC phenotype and poorer prognosis ( 17 ). Its ability to promote cell survival and inhibit apoptosis may facilitate tumor progression and resistance to therapeutic interventions ( 17 , 18 ). Further investigation of the molecular mechanisms through which BIRC5 contributes to OCCC development and progression may provide insights into potential therapeutic strategies. The EPHX2 gene encodes the enzyme epoxide hydrolase 2, which plays a crucial role in the metabolism of epoxide compounds generated during cellular processes ( 19 ). Impaired metabolism of these compounds may lead to DNA damage and disruption of cellular functions ( 19 ). EPHX2 contributes to the detoxification of these metabolites, thereby helping to maintain cellular homeostasis. The association between EPHX2 and infertility remains incompletely understood. Genetic variations in the EPHX2 gene has been associated with an elevated risk of infertility in females ( 20 ). These variants may impair epoxide metabolism, potentially affecting oocyte quality or embryo development ( 21 ). However, further studies are needed to validate this association and clarify the underlying mechanisms. Altered EPHX2 expression has also been reported in ectopic endometrial tissue compared with normal endometrium ( 22 ), suggesting potential role in the pathogenesis of endometriosis. However, the biological significance of this finding remains unclear and requires further investigation. A better understanding of the role of EPHX2 may provide insights into its potential diagnostic and therapeutic relevance in OCCC. The nuclear receptor coactivator 3 (NCOA3), also known as AIB1, SRC-3, and TRAM-1, is a transcriptional coactivator protein ( 23 ). Elevated levels of NCOA3 have been widely reported in hormone-dependent cancers such as breast cancer (BRCA), prostate adenocarcinoma (PRAD), and ovarian cancer (OVCA) ( 24 ). NCOA3 has been implicated in the proliferation of both hormonedependent and hormone -independent prostate and breast cancer cells ( 25 - 27 ). Increased expression of this coactivator has also been associated with recurrence following hormone-deprivation therapy through activation of the MAPK signaling pathway ( 28 ). These findings support the potential relevance of NCOA3 expression as a candidate biomarker and warrant further investigation. Collectively, the dysregulation of BIRC5, EPHX2 , and NCOA3 may represent candidate molecular features shared between infertility-associated endometriosis and OCCC. However, direct comparisons between fertile and infertile endometriosis samples would be more appropriate for isolating fertility-specific effects. Such analyses are recommended in future studies to better distinguish baseline fertility-related transcriptional differences from disease-specific alterations. The overlapping genes identified in this study should be regarded as hypothesis-generating findings that suggest potential shared molecular pathways, rather than evidence of causality or disease progression. Several limitations should be considered when interpreting these findings. The relatively small sample size limited clinical information, and lack of independent validation and experimental confirmation may have affected the robustness of the results. The identification of common DEGs across independently analyzed datasets from different conditions does not support inferences regarding biological progression or causality. In addition, because the TCGA-OV cohort includes mixed ovarian cancer histological subtypes and lacks fertility-related clinical information, the survival analysis should be interpreted only as an assessment of the general prognostic relevance of hub genes in ovarian cancer rather than OCCC-specific or fertility-related outcomes. Therefore, further validation in appropriate experimental models and independent clinical cohorts is required before any diagnostic, prognostic, or biological conclusions can be drawn.

This study provides preliminary bioinformatics-based identification of DEGs and hub genes potentially linking endometriosis and OCCC, with particular consideration of fertility status. Key genes such as NCOA3, BIRC5, and EPHX2 were highlighted as candidates for further investigation due to their potential roles in cell survival, hormone signaling, and epoxide metabolism. The findings should be considered hypothesis-generating and interpreted with appropriate caution. Future studies involving larger, multi-center cohorts, experimental functional validation, and direct comparisons between fertile and infertile women with endometriosis are warranted to validate and extend these findings.

This bioinformatics study was approved by the Ethics Committee of Islamic Azad University, Science and Research Branch (IR.IAU.SRB.REC.1402.008). Publicly available datasets were retrieved from the Gene Expression Omnibus (GEO) database. As only anonymized, publicly available data were analyzed, no additional informed consent was required. Two publicly available datasets, GSE122182 (miRNA expression) and GSE122183 (mRNA expression), were retrieved from the GEO ( https://www.ncbi.nlm.nih.gov/ geo/ ). The datasets were selected based on their public availability, relevance to OCCC, and the inclusion of clear clinical annotations, particularly fertility status. Samples with missing clinical information or poor-quality RNA were excluded. To minimize technical variability, batch effect correction and quantile normalization were applied. It should be noted that GSE122182 and GSE122183 were used only for exploratory miRNA–mRNA integration analyses and were not employed for differential expression, protein-protein interaction (PPI) network, hub gene, or survival analyses. The main discovery datasets were GSE120103 (endometriosis stratified by fertility status) and GSE29450 (OCCC versus normal epithelium). To establish an integrated view of molecular dysregulation, the following data linkage strategy was employed. First, differentially expressed miRNAs (DEmiRNAs) were identified from the GSE122182 dataset. Second, the multiMiR database was used to predict high-confidence target genes of these DEmiRNAs. Third, these predicted target genes were intersected with differentially expressed genes (DEGs) identified from the GSE122183 mRNA dataset. This intersection yielded a list of candidate genes potentially regulated by miRNAs associated with endometriosis and infertility, providing an integrated perspective on miRNA-mRNA regulatory interactions. The endometriosis gene expression dataset ( GSE120103 ), which includes tissue samples from 9 disease-free fertile women, 9 disease-free infertile women, 9 fertile women with stage IV endometriosis, and 9 infertile women with stage IV endometriosis, was selected and used as the discovery dataset to identify DEGs. Diagnosis and staging of endometriosis in this dataset were confirmed according to the American Society for Reproductive Medicine (ASRM) criteria by experienced gynecologists and pathologists, as reported in the original study ( 5 ). This dataset was generated using the Agilent-014850 Whole Human Genome Microarray 4x44K G4112F platform. The OCCC gene expression dataset ( GSE29450 ), which consisted of tissue samples from 10 primary OCCC tumors and 10 normal ovarian surface epithelia, was downloaded to identify DEGs. This dataset was generated using the Affymetrix Human Genome U133 Plus 2.0 Array [HG-U133_Plus_2] platform. The raw microarray datasets were preprocessed to ensure data quality and compatibility for downstream analyses. First, quantile normalization was applied to adjust potential systematic biases between samples and ensure comparability across arrays. Quantile normalization is widely used to ensure that the distribution of probe intensities is similar across all arrays, thereby reducing technical variation. After quantile normalization, batch effect correction was applied to reduce technical variation across platforms. All downstream analyses were performed on the normalized expression matrices ( 6 ). This step was essential because the data were obtained from different experimental platforms, and correcting for batch effects was critical to avoid false-positive associations. In addition, genes with consistently low expression levels across all samples were excluded using a predefined threshold. This filtering step reduced background noise and focused the analysis on genes with potential biological significance, thereby improving the sensitivity of downstream analyses, including differential expression and pathway enrichment analyses. DEGs were identified using the limma package in R (version 4.4.0). Adjusted P values were used to reduce the occurrence of false-positive results. Genes with |log2 fold change (FC)| ≥ 2 and an adjusted P<0.01 were considered differentially expressed between the study groups. Multi ple testing correction was performed using the Benjami ni-Hochberg false discovery rate (FDR) method. A Venn diagram was used to visualize the identified DEGs ( http:// bioinformatics.psb.ugent.be/webtools/Venn/ ). Functional and pathway enrichment analyses were performed to identify biological processes and signaling pathways associated with DEGs. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the EnrichR tool ( https://maayanlab.cloud/Enrichr/ ), which integrates multiple gene set enrichment databases to characterize the biological functions and signaling pathways associated with the identified DEGs. GO terms provide insight into biological processes, molecular functions, and cellular components, allowing gene products to be categorized based on shared functions, whereas KEGG pathways provide a systems-level understanding of molecular interactions within biological pathways. Enrichment analyses were conducted using the default background gene set provided by EnrichR, and statistical significance was defined as P<0.05. The STRING database (http://string.embl.de/) was used to construct a PPI network of DEGs based on known and predicted interactions. A confidence score ≥0.7 was used for to select high-confidence interactions among the DEGs ( 6 ). Subsequently, the network was visualized using Cytoscape software (version 3.5.1, Cytoscape Consortium, San Diego, CA, USA). Furthermore, the Molecular Complex Detection (MCODE) plug-in was used to identify the significant modules in the PPI network, with the following parameters: degree cutoff=2, K-Core=2, and Nnode score cutoff=0.2. For hub gene identification, separate PPI networks were constructed for Group A (shared DEGs between infertile endometriosis and OCCC) and group B (shared DEGs between fertile endometriosis and OCCC). For each group, the complete list of shared DEGs, including both upregulated and downregulated genes, was used as input. The cytoHubba plug-in was then applied, and nine topological algorithms (DNMC, MNC, Degree, EPC, BottleNeck, Closeness, Radiality, Betweenness, and Stress) were used to rank the nodes. The top five genes identified by each algorithms were selected for hub gene identification. To evaluate the prognostic relevance of the identified hub genes, overall survival analysis was performed using the OncoLnc platform ( http://www.oncolnc.org ), which integrates The Cancer Genome Atlas (TCGA) survival data with gene expression profiles. The cancer type was set to "OV" (ovarian serous cystadenocarcinoma). Survival analysis was performed by comparing patients in the upper 20% (high-expression) and lower 20% (low- expression) groups for each hub gene. Statistical significance was defined as P<0.05.