Relapse-related long non-coding RNA signature to improve prognosis prediction of lung adenocarcinoma

// Meng Zhou 1, * , Wanying Xu 1, * , Xiaolong Yue 2, * , Hengqiang Zhao 1 , Zhenzhen Wang 1 , Hongbo Shi 1 , Liang Cheng 1 , Jie Sun 1 1 College of Bioinformatics Science and Technology, Harbin Medical University, Harbin 150081, PR China 2 Medical Oncology Department, Affiliated Tumor Hospital, Harbin Medical University, Harbin 150001, PR China * These authors have contributed equally to this work Correspondence to: Jie Sun, e-mail: suncarajie@hotmail.com Keywords: long non-coding RNA, lung adenocarcinoma, prognosis, relapse, recurrence-free survival Received: November 18, 2015     Accepted: March 28, 2016     Published: April 18, 2016 ABSTRACT Increasing evidence has highlighted the important roles of dysregulated long non-coding RNA (lncRNA) expression in tumorigenesis, tumor progression and metastasis. However, lncRNA expression patterns and their prognostic value for tumor relapse in lung adenocarcinoma (LUAD) patients have not been systematically elucidated. In this study, we evaluated lncRNA expression profiles by repurposing the publicly available microarray expression profiles from a large cohort of LUAD patients and identified specific lncRNA signature closely associated with tumor relapse in LUAD from significantly altered lncRNAs using the weighted voting algorithm and cross-validation strategy, which was able to discriminate between relapsed and non-relapsed LUAD patients with sensitivity of 90.9% and specificity of 81.8%. From the discovery dataset, we developed a risk score model represented by the nine relapse-related lncRNAs for prognosis prediction, which classified patients into high-risk and low-risk subgroups with significantly different recurrence-free survival (HR=45.728, 95% CI=6.241-335.1; p=1.69e-04). The prognostic value of this relapse-related lncRNA signature was confirmed in the testing dataset and other two independent datasets. Multivariable Cox regression analysis and stratified analysis showed that the relapse-related lncRNA signature was independent of other clinical variables. Integrative in silico functional analysis suggested that these nine relapse-related lncRNAs revealed biological relevance to disease relapse, such as cell cycle, DNA repair and damage and cell death. Our study demonstrated that the relapse-related lncRNA signature may not only help to identify LUAD patients at high risk of relapse benefiting from adjuvant therapy but also could provide novel insights into the understanding of molecular mechanism of recurrent disease.