An RNA Modification–Associated Gene-Based Prognostic Model and Its Relevance to the Immune Microenvironment and Therapeutic Response in Lung Adenocarcinoma

Objective The most frequent subtype of non-small cell lung cancer, lung adenocarcinoma (LUAD), has extraordinary molecular heterogeneity and poor survival. While studies on individual RNA modifications like m⁶A or m⁵C are extensive, how multiple types of RNA alterations collectively shape LUAD prognosis, immune contexture, and therapeutic response is still not well understood. Methods TCGA provided LUAD patients' transcriptome and clinical data. We identified DEGs associated to RNA modification and created a predictive model using LASSO and Cox regression. Ten genes were used to create a RiskScore to separate patients into high- and low-risk groups in the final model. A total of thirty-seven DEGs associated with this stratification were further analyzed through GO/KEGG enrichment, GSEA, PPI network construction, immune infiltration, and drug-sensitivity prediction. External validation was performed using the GSE31210 dataset, and functional assays were conducted in LUAD cell lines to confirm key gene effects. Results The model showed moderate prognostic accuracy in the TCGA cohort (1-year AUC = 0.618) and a similar but non-significant trend in the validation dataset (GSE31210; HR = 1.33, p = 0.405). Differentiating genes between high- and low-risk groups were focused on epithelial-mesenchymal transition (EMT, NES = 2.65), NF-κB signaling (NES = 2.44), and KRAS signaling (NES = 2.10). The high-risk group showed decreased CD8⁺ T-cell infiltration and increased inflammatory activity, indicating immunological dysregulation. Drug-sensitivity research showed that Dasatinib responsiveness increased with TNS4 expression. In vitro, silencing of KLK6 or TNS4 markedly suppressed LUAD cell proliferation and migration. Conclusion This work developed a ten-gene prognostic model based on RNA-modification-related signatures and extended it to risk stratification, functional pathway analysis, and experimental validation of KLK6 and TNS4. The model may assist in early prognostic assessment and individualized therapeutic planning for LUAD, and highlights KLK6/TNS4 as potential molecular targets for precision treatment.