Background: Prognosis is a main factor affecting the survival of patients with lung adenocarcinoma (LUAD), yet no robust prognostic model of high effectiveness has been developed. This study is aimed at constructing a stable and practicable gene signature-based model via bioinformatics methods for predicting the prognosis of LUAD sufferers.
Methods: The mRNA expression data were accessed from the TCGA-LUAD dataset, and paired clinical information was collected from the GDC website. R package "edgeR" was employed to select the differentially expressed genes (DEGs), which were then used for the construction of a gene signature-based model via univariate COX, Lasso, and multivariate COX regression analyses. Kaplan-Meier and ROC survival analyses were conducted to comprehensively evaluate the performance of the model in predicting LUAD prognosis, and an independent dataset GSE26939 was accessed for further validation.
Results: Totally, 1,655 DEGs were obtained, and a 7-gene signature-based risk score was developed and formulated as risk_score = 0.000245∗NTSR1 + (7.13E - 05)∗RHOV + 0.000505∗KLK8 + (7.01E - 05)∗TNS4 + 0.000288∗C1QTNF6 + 0.00044∗IVL + 0.000161∗B4GALNT2. Kaplan-Meier survival curves revealed that the survival rate of patients in the high-risk group was lower in both the TCGA-LUAD dataset and GSE26939 relative to that of patients in the low-risk group. The relationship between the risk score and clinical characteristics was further investigated, finding that the model was effective in prognosis prediction in the patients with different age (age > 65, age < 65) and TNM stage (N0&N1, T1&T2, and tumor stage I/II). In sum, our study provides a robust predictive model for LUAD prognosis, which boosts the clinical research on LUAD and helps to explore the mechanism underlying the occurrence and progression of LUAD.
Copyright © 2020 Yingqing Zhang et al.