During the last few years, along with the emerging role of high-throughput sequencing technology, molecular characteristics associated with LUAD has been gradually recognized and are becoming the focus of various studies. In the present study, we selected 53 DELs, 21 DEGs, and 8 DEMs to build a ceRNA network that centered on three core miRNAs that were downregulated, including miR-195, miR-143, and miR-144. Then, a signature consisting of two mRNAs (HOXA10 and CBX2) and four lncRNAs (LINC00460, LINC00330, DGCR5, and C14orf132) was ultimately constructed. Our results showed that there were significant differences in the survival rates between the high and low-risk groups according to the risk scores of the identified signature. In addition, survival analysis revealed that the presence of HOXA10, CBX2, and DGCR5 as a risk factor led to a significant decrease in LUAD patient survival time, while C14orf132 was regarded as a protective factor, extending survival. Therefore, this signature is a strong predictor of overall survival.
Since miRNAs play a role in the molecular regulation of transcription , changes in these may impact prognosis. It has been widely recognized by scholars that miRNAs with high node degrees have shown an important inhibitory effect on LUAD and miR-195 can suppress the course of lung adenocarcinoma by regulating CD4 + T cell activation. In LUAD, oncogenic networks and oncogenes guided by miR-143 were identified and may help with tumor suppressor effects, establish new prognostic indicators, and find therapeutic targets. The propagation, migration, and invasion of LUAD cells are obstructed by overexpression of the miR-144 family target, EZH2. These studies all indicate that miRNAs in the ceRNA network may act as suppressors in LUAD. Meanwhile, these data provide strong clues that mRNAs and lncRNAs in the network are potential candidates for LUAD prognosis.
We relied on the intrinsic correlation between epigenetic characteristics and transcriptional regulation to identify an RNA-based signature as a reliable prognostic tool. We have also identified previous research that supports their importance in cancer. High expression of HOXA10 can promote tumor development in gastric cancer, hepatocellular carcinoma, and ovarian cancer[19–21]. More importantly, downregulated LINC00483 suppresses tumor cell invasion, migration, and epithelial to mesenchymal transition. It can also bind to miR-144 and encourage the radiosensitivity of LUAD by inhibiting the expression of HOXA10 , which promotes LUAD progression directly by enhancing Wnt/β-catenin signaling. Studies suggest that we can improve the clinical radiotherapy effect on LUAD by inhibiting the transcriptional regulation of HOXA10, thus prolonging survival. The other mRNA in the signature is CBX2, which is overexpressed in cancers, including breast cancer and hepatocellular carcinoma, and is significantly related to poor prognosis[24–25], indicating that its activity provides an advantages for the growth and development of cancer. In addition, targeted therapy studies in non-small cell lung cancer clearly show that SMARCE1 inhibits EGFR expression, in part by modulating the level of the polycomb repressive complex component CBX2. Specifically, SMARCE1 interacts directly with SWI/SNF and the EGFR oncogenic signal, as an important regulator of the drug response to MET and ALK inhibitors in non-small cell lung cancer cells. We speculate that CBX2 may have strong performance by regulating its own expression CBX2 and affecting target therapies. Moreover, survival analysis in the current study revealed that HOXA10 and CBX2 were risk factors for reduced survival and this was verified in the above studies.
Furthermore, lncRNAs act as ceRNA or miRNA sponges, representing an extensive form of gene expression regulation at the post-transcriptional level. Out of four lncRNAs (LINC00460, LINC00330, DGCR5, and C14orf132) in the signature, C14orf132 and LINC00330 were found to be related to the prognosis of colorectal cancer and bladder cancer, respectively[29–30]. In this study, high expression of C14orf132 could prolong the survival of LUAD patients, but its specific mechanism of action needs to be explored. Via targeting of the miR-302c-5p/FOXA1 axis, LINC00460 can facilitate tumor growth in LUAD. Similarly, LINC00460 has also been confirmed to promote the progression of multiple cancers through different target genes[32–34]. The last lncRNA, DGCR5, is slightly different because when it is highly expressed it exerts anticancer activity in various cancers, including gastric cancer, hepatocellular carcinoma, cervical cancer, and bladder cancer, suggesting a good prognosis[35–38]. Interestingly, DGCR5 promotes LUAD progression by inhibiting hsa-mir-22-3p, which means higher expression DGCR5 implies a worse prognosis for LUAD patients and this finding was consistent with our current survival analysis results. DGCR5 may be considered a unique prognostic indicator for LUAD. However, the underlying function remains unclear and is worth investigating.
By GO and KEGG functional enrichment analysis, the biological functions of the hub DEGs in the network were identified. Downregulated DEGs were mainly involved in the cell cycle, while upregulated DEGs were linked with protein digestion and absorption. A previous study showed that cell cycle pathway disruption can cause cell cycle arrest, which is related to the prognosis of human cancers. Moreover, the GO terms were related to protein binding, the nucleoplasm, and DNA replication. The first enrichment function, protein binding, is important during tumor metastasis when tumor cells interact with the microenvironment through binding of cell surface receptors to protein ligands. The transcriptionally active genes, especially those involved in developmental regulation and the cell cycle, interact in the nucleus with the nuclear porin. It is also known that DNA replication disorders cause genomic instability and confer genetic diversity during tumorigenesis, explaining how the above pathways may identify effective treatment strategies for LUAD.
For further verification of the significance of this study’s findings, we compared our risk score results with clinical risk factors (age, pathological stage, and TNM stage), and determined that our risk score showed a superior effect on survival, although we cannot exclude the possibility that clinical risk factors affect patient prognosis. Different from previous studies, we constructed a ceRNA network centered on prognosis-related miRNAs and used univariate and multivariate analysis to select an effective mRNA-lncRNA signature from the network. We then used clinical risk factors to validate the independent predictive significance of the signature. We believed that the use of the various statistical tests may make the results more reliable. However, this study is not without flaws. First, this study was based on an online database, which has limitations. Second, this study is somewhat simple, so additional in-depth functional analysis of the four lncRNAs in the signature is necessary. Even so, our research may provide guidance for future studies, which may help in the selection of clinical treatment targets.