TMPRSS2 Suppresses Metastasis in Endometriosis-Associated Ovarian Cancer: A Multi-Gene Prognostic Signature Construction and Functional Validation

Technical Summary: TMPRSS2 Suppresses Metastasis in Endometriosis-Associated Ovarian Cancer

Problem Statement
Endometriosis (EMS) is a chronic, refractory disease associated with an increased risk of developing endometriosis-associated ovarian cancer (EAOC), primarily comprising ovarian clear cell carcinoma (OCCC) and endometrioid ovarian carcinoma (EnOC). Despite the known risks, there is a critical lack of highly specific biomarkers for the early identification of high-risk EMS patients who may progress to EAOC. Furthermore, the complex molecular mechanisms driving the malignant transformation from EMS to EAOC, particularly regarding metastasis and the extracellular matrix (ECM), remain poorly understood. Current clinical tools are insufficient for early screening or accurate prognosis prediction in this specific patient population.

Methodology
The study employed a multi-step approach integrating bioinformatics analysis with experimental validation:

1. Data Acquisition and Differential Expression Analysis: The authors utilized two GEO datasets: GSE157153 (containing 10 adjacent EMS and 29 EAOC samples) and GSE73614 (containing 103 OCCC and EnOC samples with clinical data). Differentially expressed genes (DEGs) were identified between EMS and EAOC samples using the "limma" package.
2. Functional Annotation: Gene Ontology (GO), KEGG pathway, Gene Set Enrichment Analysis (GSEA), and Disease Ontology (DO) analyses were performed to elucidate the biological processes and pathways involved in EAOC progression.
3. Biomarker Identification: Machine learning algorithms, specifically LASSO logistic regression and Support Vector Machine-Recursive Feature Elimination (SVM-RFE), were applied to identify candidate diagnostic biomarkers. The intersection of genes from both algorithms was selected.
4. Prognostic Signature Construction: Using univariate Cox, LASSO, and bootstrap stepwise Cox regression on the GSE73614 cohort, a multi-gene risk signature was constructed. Patients were stratified into high- and low-risk groups based on median risk scores. A nomogram integrating the risk score with clinical factors (age, stage, grade) was developed to predict overall survival (OS).
5. Experimental Validation:
   • Cell Lines: Two EAOC cell lines (TOV-112D for OCCC and IGROV for EnOC) were used.
   • Manipulation: Cells were transfected with shRNA to knock down TMPRSS2 or lentiviral constructs to overexpress TMPRSS2.
   • Functional Assays: Cell proliferation (CCK-8), migration (wound-healing), invasion (Transwell), and adhesion (fibronectin/Matrigel coated plates) assays were conducted.
   • Molecular Mechanism: qRT-PCR and Western blotting were used to analyze Epithelial-Mesenchymal Transition (EMT) markers (E-cadherin, N-cadherin, Vimentin).
   • In Vivo Model: A peritoneal metastasis model was established in BALB/c nude mice using luciferase-labeled TOV-112D cells to monitor tumor growth and metastasis via bioluminescence imaging. Immunohistochemistry (IHC) was performed on metastatic tissues.

Key Contributions and Results

• Identification of Diagnostic Biomarkers: Through the intersection of LASSO and SVM-RFE algorithms, three genes—ESRP2, RAB3D, and SPINT1—were identified as candidate diagnostic biomarkers. These genes were significantly upregulated in EAOC samples, showing high diagnostic accuracy with Area Under the Curve (AUC) values of 1.0 for ESRP2 and RAB3D, and 0.997 for SPINT1.
• Construction of a Prognostic Signature: A six-gene risk signature consisting of GAP43, RBBP8, ERCC6L, CST7, CNPY2, and TMPRSS2 was constructed. The risk score formula was derived as: Risk Score = 3.931Exp(GAP43) + 3.301Exp(ERCC6L) - 3.542Exp(RBBP8) - 3.823Exp(CNPY2) - 4.390Exp(CST7) - 1.243Exp(TMPRSS2).
  • Patients in the high-risk group exhibited significantly poorer overall survival compared to the low-risk group.
  • The risk score was confirmed as an independent prognostic factor, outperforming clinical variables like age and tumor stage in predictive accuracy (C-index).
  • The constructed nomogram demonstrated good calibration for predicting 1-, 3-, and 5-year OS.
• Functional Role of TMPRSS2: Among the signature genes, TMPRSS2 was selected for deep functional investigation.
  • In Vitro: Overexpression of TMPRSS2 significantly inhibited cell proliferation, migration, and invasion, while promoting cell adhesion. Conversely, knockdown of TMPRSS2 enhanced these malignant phenotypes.
  • In Vivo: In the mouse peritoneal metastasis model, mice injected with TMPRSS2-overexpressing cells showed significantly weaker bioluminescence signals and fewer metastatic tumors in the omentum and liver compared to the control group.
  • Mechanism: The study demonstrated that TMPRSS2 suppresses metastasis by modulating the EMT process. TMPRSS2 overexpression increased the epithelial marker E-cadherin and decreased mesenchymal markers N-cadherin and Vimentin. Knockdown produced the opposite effect.
• ECM and Pathway Insights: Functional enrichment analyses highlighted that DEGs in EAOC are significantly involved in ECM remodeling, cell adhesion, and cytoskeleton organization. The study suggests that ECM dysregulation plays a pivotal role in the malignant transformation of EMS.

Significance and Claims
The authors claim that this study provides the first gene risk signature specifically constructed for EAOC patients, offering a reliable tool for prognosis prediction. Furthermore, the study identifies TMPRSS2 as a novel tumor metastasis suppressor in EAOC, a role that had not been previously confirmed in this specific cancer context. The findings elucidate a mechanism where TMPRSS2 inhibits metastasis by promoting cell adhesion and suppressing the EMT process, potentially mediated through interactions with SPINT1 (as suggested by PPI network analysis).

The authors position these findings as a novel theoretical basis for understanding EAOC pathogenesis and a potential foundation for developing targeted therapeutic strategies. However, they explicitly acknowledge limitations, including the relatively small sample sizes in public databases which limited external validation, and the need for future studies to validate TMPRSS2 expression in clinical EAOC samples and further explore its interacting molecules. The paper maintains a cautious stance regarding the diagnostic performance of the initial three biomarkers, noting that the perfect AUC values may reflect overfitting due to the limited sample size and lack of independent external validation.