Abstract
BACKGROUND: Non-small cell lung cancer (NSCLC) is a fatal disease, and radioresistance is an important factor leading to treatment failure and disease progression. The objective of this research was to detect radioresistance-related genes (RRRGs) with prognostic value in NSCLC. METHODS: The weighted gene coexpression network analysis (WGCNA) and differentially expressed genes (DEGs) analysis were performed to identify RRRGs using expression profiles from TCGA and GEO databases. The least absolute shrinkage and selection operator (LASSO) regression and random survival forest (RSF) were used to screen for prognostically relevant RRRGs. Multivariate Cox regression was used to construct a risk score model. Then, Immune landscape and drug sensitivity were evaluated. The biological functions exerted by the key gene LBH were verified by in vitro experiments. RESULTS: Ninety-nine RRRGs were screened by intersecting the results of DEGs and WGCNA, then 11 hub RRRGs associated with survival were identified using machine learning algorithms (LASSO and RSF). Subsequently, an eight-gene (APOBEC3B, DOCK4, IER5L, LBH, LY6K, RERG, RMDN2 and TSPAN2) risk score model was established and demonstrated to be an independent prognostic factor in NSCLC on the basis of Cox regression analysis. The immune landscape and sensitivity to anti-tumour drugs showed significant disparities between patients categorized into different risk score subgroups. In vitro experiments indicated that overexpression of LBH enhanced the radiosensitivity of A549 cells, and knockdown LBH reversed the cytotoxicity induced by X-rays. CONCLUSION: Our study developed an eight-gene risk score model with potential clinical value that can be adopted for choice of drug treatment and prognostic prediction. Its clinical routine use may assist clinicians in selecting more rational practices for individuals, which is important for improving the prognosis of NSCLC patients. These findings also provide references for the development of potential therapeutic targets.