Construction and validation of a prognostic model based on ten signature cell cycle-related genes for early-stage lung squamous cell carcinoma

Abstract

BACKGROUND: We performed a bioinformatics analysis to screen for cell cycle-related differentially expressed genes (DEGs) and constructed a model for the prognostic prediction of patients with early-stage lung squamous cell carcinoma (LSCC). METHODS: From a gene expression omnibus (GEO) database, the GSE157011 dataset was randomly divided into an internal training group and an internal testing group at a 1:1 ratio, and the GSE30219, GSE37745, GSE42127, and GSE73403 datasets were merged as the external validation group. We performed single-sample gene set enrichment analysis (ssGSEA), univariate Cox analysis, and difference analysis, and identified 372 cell cycle-related genes. Additionally, we combined LASSO/Cox regression analysis to construct a prognostic model. Then, patients were divided into high-risk and low-risk groups according to risk scores. The internal testing group, discovery set, and external verification set were used to assess model reliability. We used a nomogram to predict patient prognoses based on clinical features and risk values. Clinical relevance analysis and the Human Protein Atlas (HPA) database were used to verify signature gene expression. RESULTS: Ten cell cycle-related DEGs (EIF2B1, FSD1L, FSTL3, ORC3, HMMR, SETD6, PRELP, PIGW, HSD17B6, and GNG7) were identified and a model based on the internal training group constructed. From this, patients in the low-risk group had a higher survival rate when compared with the high-risk group. Time-dependent receiver operating characteristic (tROC) and Cox regression analyses showed the model was efficient and accurate. Clinical relevance analysis and the HPA database showed that DEGs were significantly dysregulated in LSCC tissue. CONCLUSION: Our model predicted the prognosis of early-stage LSCC patients and demonstrated potential applications for clinical decision-making and individualized therapy.