A prognostic score model based on eight metabolism-associated genes predicts the survival of adult females with lung adenocarcinoma

Abstract Background As a non-small cell lung cancer, lung adenocarcinoma (LUAD) is common in women and non-smokers. This study is aimed to construct a prognostic score (PS) model for adult females with LUAD. Methods The gene expression data of adult females with LUAD from The Cancer Genome Atlas database was obtained as the training set, and GSE50081 and GSE37745 from Gene Expression Omnibus database were downloaded as the validation sets. The differentially expressed genes (DEGs) between LUAD and normal samples were screened by limma package. The metabolism-associated DEGs were selected by Gene Set Enrichment Analysis, and were conducted with enrichment analysis using DAVID tool. After the independent prognosis-associated genes were identified by survival package, the optimal gene combination was screened using penalized package to build the PS model. Besides, the nomogram survival model based on the independent prognostic clinical factors was constructed by rms package. Using HPAanalyze package, the protein expression levels of the optimal genes were mined. Results There were 2388 DEGs between LUAD samples and normal samples. Totally, 150 metabolism-associated DEGs were screened, for which PPAR signaling pathway was enriched. The optimal gene combination (involving CYP17A1, ASPG, DUOX1, CIDEC, TH, B4GALNT1, APOA2, and GCKR) was selected, based on which the PS model was built. Combined with pathologic stage and PS model status, the nomogram survival model was constructed. Moreover, CIDEC was a characteristic gene in lung cancer and other cancers. Conclusion The PS model and the nomogram survival model might be applied for the prognostic prediction of adult females with LUAD.