Identification and validation of an autophagy-related gene signature for predicting prognosis in patients with esophageal squamous cell carcinoma

Abstract

Esophageal squamous cell carcinoma (ESCC) is the main subtype of esophageal cancer. Since autophagy-related genes (ARGs) play a key role in the pathogenesis of many tumors, including ESCC, the purpose of this study is to establish an autophagy-related prognostic risk signature based on ARGs expression profile, and to provide a new method for improving prediction of clinical outcomes. We obtained the expression profiles of ESCC from public data (GSE53625) and extracted the portion of ARGs. Differential expression analysis and enrichment analysis were performed to confirm abnormal autophagy-related biological functions. Univariate and multivariate Cox regression analyses were performed on RNA microarray data (GSE53625) to construct a prognostic risk signature associated with autophagy. The performance of the model was evaluated by receiver operating characteristic (ROC) analysis, survival analysis and Brier score. The model was subjected to bootstrap internal validation. The potential molecular mechanism of gene signature was explored by gene set enrichment analysis (GSEA). Spearman correlation coefficient examined the correlation between risk score and immune status and ferroptosis. The expression levels of genes and proteins were validated by qRT-PCR and immunohistochemistry in ESCC cell lines and ESCC tissues. We constructed and validated an autophagy-related prognostic risk signature in 179 patients with ESCC. The long-term survival of patients in high-risk group was lower than that in low-risk group (log-rank, P value < 0.001). ROC analysis and Brier score confirmed the reliability of the signature. GSEA results showed significant enrichment of cancer- and autophagy-related signaling pathways in the high-risk ESCC patients and immunoregulatory signaling pathways in the low-risk ESCC patients. Correlation analysis showed that the risk signature can effectively predict the effect of immunotherapy. About 33.97% (71/209) ferroptosis-related genes were significantly correlated with risk scores. Finally, the results of qRT-PCR and immunohistochemistry experiments were consistent with bioinformatics analysis. In brief, we constructed a novel autophagy-related gene signature (VIM, UFM1, TSC2, SRC, MEFV, CTTN, CFTR and CDKN1A), which could improve the prediction of clinical outcomes in patients with ESCC.