Integration of Scrna-Seq and Bulk Rna-Seq to Validate a Prognostic Associated Signature in Esophageal Squamous Cell Carcinoma

Abstract Background Considerable evidence has identified important targets for esophageal squamous cell carcinoma (ESCC), but its efficacy and the heterogeneity of ESCC are the main causes of treatment failure. Single-cell RNA-sequencing (scRNA-seq) is a powerful tool for analyzing tumor heterogeneity at the single-cell level and helps to better identify potential therapeutic targets. Methods ESCC scRNA-seq data were extracted from the Gene Expression Omnibus (GEO) database. Single-sample gene set enrichment analysis (ssGSEA) was performed using the limma package to analyze ESCC-associated prognostic marker genes from ESCC transcriptome data downloaded from the TCGA database and the results were validated with transcriptomic data from 56 ESCC patients in our hospital. KEGG (Kyoto Encyclopedia of Genes and Genomes) and GO (Gene Ontology) analyses of prognostic marker genes were performed by GSEA. Then, univariate and multivariate Cox and lasso regressions were performed to further build signatures. Finally, qPCR was used to assess the expression of prognostic marker genes in 5 pairs ESCC samples. Results Integrate scRNA-seq (GSE53624) datasets to obtain 13 cell clusters. We characterized single-cell expression profiles of 24076 cell samples and identified 4 marker genes associated with prognosis. Four key prognostic marker genes were trained and validated on the TCGA-ESCA dataset and in 55 patients from our in-hospital cohort to evaluate predictive overall survival (OS) models. Correlation analysis showed that prognostic marker genes were independent of risk clinical characteristics. In addition, we performed Cox regression analysis, suggesting that the 4 prognostic marker genes screened were independent risk factors related to the prognosis of ESCC patients. In addition, qPCR showed that the expression of IER3 and SAA1 was decreased and the expression of CDKN3 was increased in ESCC tissues, while the expression of TINP3 was not significantly different. Conclusion We constructed prognostic marker genes by scRNA-seq and validated the model in a large patient population as well as in ESCC tissues, which provides reasonable evidence and valuable resources for prognostic stratification and the study of potential targets for ESCC.