Abstract
Background
Ovarian cancer (OV) is the fifth leading cause of cancer death among females. Growing evidence supports a key role of tumor microenvironment in growth, progress, and metastasis of OV. However, the impacts of gene expression signatures related with OV microenvironment on prognosis have not been well-established . This study aimed to apply ESTIMATE algorithm to extract genes related with tumor microenvironment that predicted poor outcomes in OV patients.
Methods
The gene expression profile of OV samples were downloaded from The Cancer Genome Atlas (TCGA) database. The immune scores and stromal scores of 469 OV samples were available based on the ESTIMATE algorithm. To better understand impacts of gene expression signatures related with OV microenvironment on prognosis, these samples were categorized based on their ESTIMATE scores into high and low score groups. A different OV cohort from the Gene Expression Omnibus (GEO) database was used for external validation.
Results
The molecular subtypes in OV patients were correlated with stromal scores, in which the mesenchymal subtype had the highest stromal scores (p < 0.0001). Poor prognosis were found in patients (especially for patients with overall survivals (OS) < 5 years) with higher stromal score (p = 0.0376). 449 differentially expressed genes (DEGs) in stromal scores group were identified and 26 DEGs were significantly associated with poor prognosis in OV patients (p < 0.05). Eventually, 6 genes have further validated to be significantly associated with poor outcomes in 40 patients from a different OV cohort of GEO database (p < 0.05).
Conclusion
In this study, several genes related with tumor microenvironment that predicted poor prognosis in OV patients were extracted. In addition, some previously overlooked genes could be potential prognostic biomarkers for OV.
A decomposition model to track gene expression signatures: preview on observer-independent classification of ovarian cancer