Integrated analysis identifies an aging-related prognostic signature for lung adenocarcinoma

Abstract

Lung adenocarcinoma (LUAD) is characterized by marked molecular heterogeneity, resulting in substantial variability in clinical outcomes. Aging is a major risk factor for LUAD and is closely associated with genomic instability, metabolic dysregulation, cellular senescence, and remodeling of the tumor microenvironment. However, the prognostic significance and biological relevance of aging-related genes in LUAD remain incompletely understood. This study aimed to systematically characterize aging-related molecular alterations in LUAD and develop a prognostic signature with potential biological and clinical relevance. Transcriptomic and clinical data for LUAD were obtained from The Cancer Genome Atlas (TCGA), and aging-related genes were collected from a public aging-associated gene set. Differentially expressed genes (DEGs) were identified by comparing tumor and normal tissues and intersected with genes from the GEO validation cohorts GSE37745 and GSE50081 to obtain common aging-related DEGs. Prognosis-associated genes were screened by univariate Cox regression, and molecular subtypes were identified by consensus clustering. An aging-related prognostic signature was then constructed using LASSO-Cox regression and validated in two independent GEO cohorts. Associations of the risk score with clinicopathological features, somatic mutation patterns, stemness, and the tumor immune microenvironment were further explored. In addition, single-cell RNA-seq data from GSE131907 were used to investigate the cellular distribution of a key model gene, while RT-qPCR analysis of clinical samples and in vitro experiments were performed for preliminary validation. A total of 274 aging-related DEGs were identified in LUAD, of which 52 were significantly associated with overall survival. Based on these genes, LUAD samples were classified into two molecular subtypes with distinct prognoses, immune infiltration patterns, and biological pathway enrichment characteristics. An eight-gene prognostic signature consisting of CD19, CYP17A1, TFAP2A, HMGA2, ABCC2, PCSK9, CLDN14, and INHA was subsequently established. This signature stratified patients into high-risk and low-risk groups with significantly different overall survival in the TCGA cohort, and its prognostic relevance was further evaluated in the GSE37745 and GSE50081 cohorts. The risk score was independently associated with prognosis and correlated with advanced clinicopathological features. High-risk tumors exhibited higher stemness, a higher somatic mutation frequency, and enrichment of proliferation-related pathways, whereas low-risk tumors displayed stronger immune-related characteristics and higher immune and stromal scores. Single-cell analysis showed that CLDN14 had relatively higher expression in B cells than in other annotated cell populations. RT-qPCR analysis in paired tumor and adjacent normal tissues from 10 LUAD patients further showed that CLDN14 was significantly upregulated in tumor tissues. In vitro assays suggested that CLDN14 knockdown impaired the proliferative, migratory, and invasive capacities of LUAD cell models. We established and validated an aging-related eight-gene prognostic signature for LUAD with independent prognostic value. This signature was associated with distinct clinical, molecular, stemness, mutational, and immune microenvironmental features. In addition, CLDN14 was identified as a candidate gene potentially associated with LUAD progression and was supported by preliminary experimental validation. These findings provide an initial aging-related framework for prognostic stratification and biological characterization in LUAD.