Identification of cancer stem cell-related biomarkers in lung adenocarcinoma by stemness index and weighted correlation network analysis.

Abstract

Accounting for tumor heterogeneity, cancer stem cells (CSC) are involved in tumor metastasis, relapse, and drug resistance. Genes regulating CSC characteristics in lung adenocarcinoma (LUAD) were explored and validated in this study. The mRNA stemness index (mRNAsi) of more than 500 LUAD cases from The Cancer Genome Atlas database were calculated using a one-class logistic regression machine learning algorithm based on the mRNA expression of pluripotent stem cells and their differentiated progeny. mRNAsi-related key genes were identified by weighted correlation network analysis. The expression levels and prognostic roles of key genes were analyzed in Oncomine, PrognoScan, and Kaplan-Meier plotter databases, and validated using data from our center. The mRNAsi was significantly higher in LUAD compared with normal lung tissues. LUAD patients of advanced stage exhibited a higher mRNAsi and worse overall survival (OS). Eight key genes were identified: heat shock 70kDa protein 4 (HSPA4), cell division cycle associated 7 (CDCA7), cell division cycle 20 (CDC20), cyclin-dependent kinase 1 (CDK1), CAP-GLY domain containing linker protein 1 (CLIP1), cyclin B1 (CCNB1), H2A histone family, member X (H2AFX), and Bloom syndrome, RecQ helicase-like (BLM). These genes were differentially expressed in various types of malignancies and validated in the LUAD cases. LUAD patients with low expression of CDC20, CDK1, CCNB1, H2AFX, or BLM had a significantly better OS, whereas OS was reduced for patients with low expression of CLIP1. In addition, the expression of CDCA7 did not significantly impact the OS of LUAD patients. The protein-protein interaction networks evaluated by STRING demonstrated strong relationships between these key genes, which were validated in our cases. The mRNAsi was significantly higher in LUAD compared with normal samples. Eight mRNAsi-related key genes were associated with prognosis and the cell cycle, and were strongly correlated with each other and differentially expressed in tumor and normal samples. We provide a new strategy for exploring stemness-related genes in LUAD cases.