A group of long noncoding RNAs identified by data mining can predict the prognosis of lung adenocarcinoma

Abstract

Long noncoding RNAs (lncRNA) are reported to be potential cancer biomarkers. This study aims to find new lncRNA biomarker relevant to lung adenocarcinoma. Gene expression profile and clinical data of lung adenocarcinoma and lung squamous cell carcinoma patients were downloaded from the UCSC Xena database. These data were analyzed to identify potential lncRNA prognostic biomarkers, and the candidate lncRNAs were analyzed and verified with association analysis, meta-analysis, survival analysis, gene ontology analysis, gene set enrichment analysis, and other statistical methods. A group of 5 lncRNAs was identified from the 1965 differentially expressed (fold-change >2) genes. Four of these 5 lncRNAs were expressed at a lower level in lung adenocarcinoma tissues and the other one at a higher level (P < .0001). A risk score model was constructed using a linear combination of the expression levels of these lncRNAs. High-risk patients showed poorer overall survival (hazard ratio [HR] = 2.14; 95% confidence interval [CI], 1.67-3.06, P < .0001), disease-free survival (HR = 1.84; 95% CI, 1.26-2.35, P = .0007), and recurrence-free survival (HR = 1.51; 95% CI, 1.02-2.40, P = .04). The 5-fold cross-validation and subsequent meta-analysis further verified that patients in the low-risk group had better survival (95% CI, 0.74-1.79, Z = 4.72, P < .00001). Furthermore, both univariate and multivariate Cox regression analyses revealed that the prognostic value of these 5 lncRNAs was independent of other clinical prognostic factors. Further analysis indicated that these 5 lncRNAs might be associated with tumor metastasis. Taken together, our study suggests new prognostic lncRNA biomarkers for lung adenocarcinoma.

Keywords: adenocarcinoma of lung; long noncoding RNA; prognosis; survival; tumor biomarker.