A Rho GTPase-related gene signature predicts prognosis and reveals an immunosuppressive microenvironment in lung adenocarcinoma: an integrated analysis of bulk and single-cell RNA sequencing data

Abstract

Lung adenocarcinoma (LUAD) is a malignancy characterized by significant heterogeneity and variable clinical outcomes. RhoGTPases play a critical role in cancer progression; however, their precise effects on LUAD prognosis remain inadequately understood. This study aimed to construct and validate a RhoGTPase-driven prognostic signature for LUAD, alongside an examination of its biological relevance. Transcriptomic and single-cell RNA sequencing (scRNA-seq) data were obtained from public databases. Candidate genes were identified by intersecting differentially expressed genes (DEGs) with a set of 733 RhoGTPase-related genes. Prognostic genes were selected through univariate Cox regression and least absolute shrinkage and selection operator (LASSO) analysis. A prognostic model was developed and subsequently validated. Analyses including functional enrichment, immune infiltration, and drug sensitivity were conducted. Additionally, scRNA-seq analysis was employed to define key cellular populations and explore the dynamic expression of the prognostic genes. A total of 5,420 DEGs were identified, yielding 233 candidate genes upon intersection with the RhoGTPase-related gene set. Four genes (CCT6A, PLK1, DSG2, PCDH7) were designated as prognostic. The risk model effectively stratified patients into the high-risk group (HRG) and the low-risk group (LRG), revealing significant differences in overall survival rates. The risk score served as an independent prognostic indicator. Functional enrichment analysis showed that HRG was associated with cell cycle pathways, while LRG was predominantly enriched in immune-related pathways, with HRG exhibiting a more immunosuppressive microenvironment. Drug sensitivity analysis indicated potential responsiveness to specific agents, such as doramapimod, in HRG. ScRNA-seq analysis identified epithelial cells as the primary population of interest. Pseudotime analysis uncovered a developmental trajectory in which CCT6A, DSG2, and PCDH7 were associated with late-stage, differentiated epithelial cell states. A robust prognostic model based on four RhoGTPase-related prognostic genes (CCT6A, PLK1, DSG2, PCDH7) was established, effectively stratifying LUAD patients. This model provides valuable insights into the heterogeneity of LUAD and has the potential to inform personalized therapeutic strategies.