The deregulation of fatty acid metabolism plays a pivotal role in cancer. Our objective is to construct a prognostic model for patients with endometrial carcinoma (EC) based on genes related to fatty acid metabolism-related genes (FAMGs). RNA sequencing and clinical data for EC were obtained from The Cancer Genome Atlas (TCGA). Lasso-Penalized Cox regression was employed to derive the risk formula for the model, the score = esum(corresponding coefficient × each gene's expression). Gene set enrichment analysis (GSEA) was utilized to examine the enrichment of KEGG and GO pathways within this model. Correlation analysis of immune function was conducted using Single-sample GSEA (ssGSEA). The "ESTIMATE" package in R was utilized to evaluate the tumor microenvironment. The support vector machine recursive feature elimination (SVM-RFE) and randomforest maps were employed to identify key genes. The effects of PTGIS on the malignant biological behavior of EC were assessed through CCK-8 assay, transwell invasion assay, cell cycle analysis, apoptosis assay, and tumor xenografts in nude mice. A novel prognostic signature comprising 10 FAMGs (INMT, ACACB, ACOT4, ACOXL, CYP4F3, FAAH, GPX1, HPGDS, PON3, PTGIS) was developed. This risk score serves as an independent prognostic marker validated for EC. According to ssGSEA analysis, the low- and high-risk groups exhibited distinct immune enrichments. The key gene PTGIS was screened by SVM-RFE and randomforest method. Furthermore, we validated the expression of PTGIS through qRT-PCR. In vitro and in vivo experiments also confirmed the effect of PTGIS on the malignant biological behavior of EC.
Keywords: endometrial carcinoma (EC); fatty acid metabolism-related genes (FAMGs); prognostic model; prostacyclin synthase (PTGIS); tumor microenvironment (TME).