LAUSR

Background Lymph node metastasis is an important route of lung cancer metastasis and can significantly affect the survival of lung cancer. Methods All the analysis was conducted out in the R software. Expression profile and clinical information of lung adenocarcinoma (LUAD) patients were downloaded from The Cancer Genome Atlas database. Results In our study, we firstly identified the characteristic genes of lymph node metastasis in LUAD through two machine learning algorithms, least absolute shrinkage and selection operator (LASSO) logistic regression, and SVM-RFE algorithms. Ten characteristic genes were finally identified, including CRHR2, ITIH1, PRSS48, MAS1L, CYP4Z1, LMO1, TCP10L2, KRT78, IGFBP1, and PITX3. Next, we performed univariate Cox regression, LASSO regression, and multivariate Cox regression sequentially to construct a prognosis model based on MAS1L, TCP10L2, and CRHR2, which had a good prognosis prediction efficiency in both training and validation cohorts. Univariate and multivariate analysis indicated that our model is a risk factor independent of other clinical features. Pathway enrichment analysis showed that in the high-risk patients, the pathway of MYC target, unfolded protein response, interferon alpha response, DNA repair, reactive oxygen species pathway, and glycolysis were significantly enriched. Among three model genes, MAS1L aroused our interest and therefore was selected for further analysis. KM survival curves showed that the patients with higher MAS1L might have better disease-free survival and progression-free survival. Further, pathway enrichment, genomic instability, immune infiltration, and drug sensitivity analysis were performed to in-deep explore the role of MAS1L in LUAD. Conclusions Results showed that the signature based on MAS1L, TCP10L2, and CRHR2 is a useful tool to predict prognosis and lung cancer lymph node metastasis.