Establishing a Prognostic Model in Prostate Adenocarcinoma through Comprehensive scRNA-Seq and Bulk RNA-Seq Analysis and Validation

Abstract

Abstract Background The prognostic management of Prostate adenocarcinoma (PRAD) presents a considerable challenge to healthcare professionals. However, it fails to accurately capture the fundamental cellular and molecular functions within tumor cells. Methods The data for PRAD scRNA-seq were retrieved from the Gene Expression Omnibus (GEO) database. The limma program was utilized to identify differentially expressed genes (DEGs) in PRAD patients that exert an influence on overall survival (OS). For the identification of key modules associated with PRAD, Weighted Gene Correlation Network Analysis (WGCNA) was employed. The intersection of core cell marker genes, PRAD key module genes, and DEGs was utilized to build a predictive model using univariate Cox and Least Absolute Shrinkage and Selection Operator (LASSO) analyses. Furthermore, we conducted experimental validation by collecting patient samples. Results Analysis of 162,897 scRNA-seq datasets and identified 7 central cell types. From the scRNA-seq dataset, 1805 marker genes were identified, while the bulk RNA-seq dataset yielded 1086 DEGs. Additionally, 2545 genes were linked to a key module identified through WGCNA. A predictive model was derived from the expression levels of 21 signature genes following intersection, univariate Cox, and LASSO analyses. And we confirmed the accuracy of our analysis through the patient samples we collected. Conclusion This study developed a unique prognosis-predictive model to predict the survival condition of individuals with PRAD through the integration of scRNA-seq and bulk RNA-seq data. The risk score emerges as a potential independent predictive indicator, demonstrating a strong relationship with the immunological microenvironment.