A New Model Based on Fatty Acid Metabolism-Related Genes to Predict the Prognosis of Esophageal Carcinoma

Abstract

Backgrounds: Esophageal carcinoma (EC), a type of malignant tumor originating from the esophageal mucosa, is frequently encountered in clinical practice. Fatty acids contribute to the polarization of tumor-associated macrophage to the immunosuppressive M2 phenotype. Fatty acid metabolism is an essential factor in tumor development and progression. Methods: EC expression and clinical data were downloaded from the Cancer Genome Atlas database and GSE53625 dataset. Fatty acid metabolism gene sets were downloaded from the Msigdb database. Differentially expressed fatty acid-related genes (DFAGs) in tumor and normal samples were screened using a t-test, and their prognostic value was investigated. Optimal prognostic DFAGs were identified to establish the fatty-acid-risk score (FARS) model. A nomogram was then established based on the independent prognostic factors. Associations between the FARS and clinical features, immune infiltration, tumor mutation burden (TMB), drug sensitivity, and tumor immune dysfunction and exclusion (TIDE) were analyzed using the prophetic package. Results: Six prognostic DFAGs were identified from 161 DFAGs and 13 prognostic genes. Three genes, PDHA1, PTGES3, and CPT2, were identified as optimal prognostic DFAGs to establish the FARS model. Patients with high FARS tended to present shorter survival times in the Cancer Genome Atlas (TCGA) database and GSE53625 cohorts. The tumor stage and FARS were identified as independent prognostic factors for establishing the nomogram. The nomogram showed high conformance with actual survival in predicting 1-, 3-, and 5-year survival rates. High-FARS groups had a high abundance of infiltrating M2 macrophages and higher TMB. Patients with high TMB tended to have low survival. High FARS also correlated with higher sensitivity to cisplatin, gemcitabine, and paclitaxel and a high TIDE score, indicating a poor response to immune checkpoint blockade (ICB) therapy. Conclusions: Predicting model including DHA1, PTGES3 and CPT2 was established. These genes could predict overall EC survival, chemosensitivity, and immunotherapy response and were correlated with aggressive clinicopathologic features.