Background: Esophageal cancer (ESCA) is one of the most commonly diagnosed cancers in the world. Tumor immune microenvironment is closely related to tumor prognosis. The present study aimed at analyzing the competing endogenous RNA (ceRNA) network and tumor-infiltrating immune cells in ESCA.
Methods: The expression profiles of mRNAs, lncRNAs, and miRNAs were downloaded from the Cancer Genome Atlas database. A ceRNA network was established based on the differentially expressed RNAs by Cytoscape. CIBERSORT was applied to estimate the proportion of immune cells in ESCA. Prognosis-associated genes and immune cells were applied to establish prognostic models basing on Lasso and multivariate Cox analyses. The survival curves were constructed with Kaplan-Meier method. The predictive efficacy of the prognostic models was evaluated by the receiver operating characteristic (ROC) curves.
Results: The differentially expressed mRNAs, lncRNAs, and miRNAs were identified. We constructed the ceRNA network including 23 lncRNAs, 19 miRNAs, and 147 mRNAs. Five key molecules (HMGB3, HOXC8, HSPA1B, KLHL15, and RUNX3) were identified from the ceRNA network and five significant immune cells (plasma cells, T cells follicular helper, monocytes, dendritic cells activated, and neutrophils) were selected via CIBERSORT. The ROC curves based on key genes and significant immune cells all showed good sensitivity (AUC of 3-year survival: 0.739, AUC of 5-year survival: 0.899, AUC of 3-year survival: 0.824, AUC of 5-year survival: 0.876). There was certain correlation between five immune cells and five key molecules.
Conclusion: The present study provides an effective bioinformatics basis for exploring the potential biomarkers of ESCA and predicting its prognosis.
Keywords: Esophageal cancer; Immune infiltration; Prognostic model; TCGA; ceRNA network.
© 2021 The Author(s).