LUAD is the most prevalent type of lung cancer and a prototype for precision oncology [17]. Accumulating evidence shows that lncRNAs regulate the pathological processes of lung cancer. Several studies have also identified some lncRNAs that affect the prognosis of non-small cell lung cancer; however, to date, there is no systematic method to identify lncRNA signature sets to predict survival for LUAD patients. Herein, we identified a novel and efficient autophagy-related lncRNA signature based on a TCGA dataset and validated its efficiency using an independent EGA dataset. Our signature was able to effectively stratify patients according to OS. We further determined the efficacy of our signature in the training set and verification set, indicating that the signature has high prognostic value.
To our knowledge, this is the first autophagy-related lncRNA predictive model developed based on RNA-sequencing data using a cohort of more than 100 LUAD patients. In the first step, we identified 1651 lncRNAs based on the lncRNA–autophagy gene co-expression network. The expression profiles of these lncRNAs in 439 LUAD patients were then analyzed by univariate and stepwise multiple Cox proportional hazards regression analyses. Fifteen lncRNA signatures related to autophagy were identified and used to establish a predictive model based on the linear combination of these lncRNAs. Using this predictive model, distinctive separation was observed in the survival curve between the high-risk and low-risk patient groups. The AUC obtained by ROC analysis was 0.728, demonstrating the high sensitivity and specificity of this lncRNA signature model. When considering other clinical factors together, multivariate Cox regression analysis indicated that the autophagy-associated lncRNAs could independently predict survival outcomes from these conventional clinical-pathological factors, including stage, sex, age, and TNM. Further stratified analysis showed that when predicting different survival rates of patients in the EGA cohort, the autophagy‐related lncRNA signature has good discriminatory power.
To date, only a few lncRNAs have been well characterized with respect to their biological functions. Recent studies have found that the level of MIR22HG, one of our prognostic lncRNAs, is significantly reduced compared to that in normal lung tissue. Further, silencing MIR22HG triggers the signaling for cell survival and apoptosis via aberrant expression of the oncogenes p2, YBX1, and MET [18]. Other studies suggested that MET and p21 are important proteins involved in the autophagy pathway [19, 20]. In line with these previous results, our analysis based on the TCGA cohort showed that HSPC324 expression levels are distinctly higher in lung tumor tissues than those in normal tissues [21]. Further, abnormal expression of the lncRNA HSPC324 inhibits proliferation, and increases apoptosis and reactive oxygen species (ROS) accumulation in lung adenocarcinoma cells, as an important member of the ROS-dependent autophagy pathway [21, 22]. However, the biological functions of most of our prognostic lncRNAs are not clear.
LncRNAs participate in various biological activities by either promoting or suppressing gene expression at the transcriptional and post-transcriptional levels [23]. Thus, it is possible to deduce the biological role of an lncRNA from the functional perspective of co-expressed protein-coding genes (PCGs) [24]. Therefore, we studied the pattern of co-expression with regard to mRNAs and lncRNAs, and conducted functional enrichment analysis of the co-expressed PCGs to determine the biological function of prognosis-associated lncRNAs in the signature [25, 26]. Our results revealed that PCGs for which the expression value was negatively or positively related to the prognosis of lncRNA and also related to autophagy were enriched in several functional groups of recognized pathways related to autophagy, such as the lysosome pathway, proteasome pathway, oxidative phosphorylation, and phosphatidylinositol signaling system, which are also strongly related to the pathogenesis of LUAD. Therefore, it is reasonable to speculate that autophagy-related lncRNA markers might participate in LUAD by regulating PCGs in recognized autophagy-related biological pathways and activities.