Identification and Validation of Co-Expressed Immune-Related Gene Signature Affecting the Pattern of Immune Infiltrating in Esophageal Cancer

Objective: Esophageal squamous cell carcinoma (ESCC) is a common malignant tumor of the digestive tract, and its molecular mechanisms have not been fully clarified. This study aimed to evaluate the immune infiltration pattern of esophageal cancer through a gene co-expression network, and to provide biomarkers for immunotherapy of esophageal cancer. Methods: We downloaded RNA-seq data of ESCC samples from GSE53625 and GSE66258 datasets, then assessed the immune score and tumor purity through the ESTIMATE algorithm. Next, a co-expression network was constructed by the weighted gene co-expression network analysis, and the key co-expressed immune- related genes were identified on the basis of existing human immune-related genes. Afterward, we utilized bioinformatics algorithms including GSVA, CIBERSORT, and ssGSEA to clarify the relationship between hub genes and immune infiltration patterns. Finally, these hub genes were used to evaluate the sensitivity to immunotherapy by the subclass mapping algorithm, which were further validated by digital pathology through the Hover- Net algorithm. Results: Sixteen immune-related genes with robust expression characteristics were identified and used to build gene signatures. The expression of gene signature was significantly related to the immune infiltration pattern and immunotherapy sensitivity prediction in patients with esophageal cancer. Consistent with previous studies, genetic changes at the level of somatic mutations such as NFE2L2 were revealed. Conclusion: A total of 16 immune-related genes with the total expression gene signature can be used as biomarkers for immunotherapy of esophageal squamous cell carcinoma. Its molecular mechanisms deserve further study to guide clinical treatment in the future.