Long noncoding RNAs (lncRNAs) are aberrantly expressed in various cancers types and can function as competing endogenous RNAs (ceRNAs), which promote and maintain tumor initiation and progression. In this study, we explored the functional roles and regulatory mechanisms of lncRNAs as ceRNAs in colorectal cancer and their clinical potential as biomarkers. The RNA sequencing profiles of patients with colorectal cancer were downloaded from TCGA database, and 62 lncRNAs, 30miRNAs, and 59 mRNAs were identified to comprise the ceRNA network (fold change > 2, P < 0.01). Functional enrichment analysis suggested that the target genes of the ceRNA network may be involved in the pathways related to cancer, including the signaling pathway that regulates the pluripotency of stem cells, wnt signaling pathway, hippo signaling pathway, basal cell carcinoma, and colorectal cancer. Univariate and multivariate Cox's proportional hazard regression model revealed that five (H19, MIR31HG, HOTAIR, WT1‐AS, and LINC00488) out of 62 lncRNAs were closely related to the overall survival (OS) (P < 0.05). Furthermore, the five‐lncRNA model could be an independent prognostic model in colorectal cancer. We computed for the risk function and constructed a risk score based on the five lncRNAs. Results showed that patients with high‐risk scores have poor survival rates. Additionally, combing the risk score and other clinicopathological features, we can better predict the patient's survival probabilities. Furthermore, we validate our model in the GSE38832 dataset. Collectively, our study has provided a deeper understanding of the lncRNA‐related ceRNA regulatory mechanism in CRC and identified five‐lncRNA model, which could be considered as candidate prognostic biomarkers and therapeutic targets.

Colorectal cancer (CRC) is one of the most common and deadly malignancies in the world. In China, the morbidity rate of CRC has increased during the period 2000 to 2011. Biomarker detection for early CRC diagnosis can effectively reduce the mortality of patients with CRC. To explore the underlying mechanisms of effective biomarkers and identify more of them, we performed weighted correlation network analysis (WGCNA) on a GSE68468 dataset generated from 378 CRC tissue samples. We screened the gene set (module), which was significantly associated with CRC histology, and analyzed the hub genes. The key genes were identified by obtaining six colorectal raw data (i.e., GSE25070, GSE44076, GSE44861, GSE21510, GSE9348, and GSE21815) from the GEO database (https://www.ncbi.nlm.nih.gov/geo). The robust differentially expressed genes (DEGs) in all six datasets were calculated and obtained using the library “RobustRankAggreg” package in R 3.5.1. An integrated analysis of CRC based on the top 50 downregulated DEGs and hub genes in the red module from WGCNA was conducted, and the intersecting genes were screened. The Kaplan–Meier plot was further analyzed, and the genes associated with CRC prognosis based on patients from the TCGA database were determined. Finally, we validated the candidate gene in our clinical CRC specimens. We postulated that the candidate genes screened from the database and verified by our clinical pathological data may contribute to understanding the molecular mechanisms of tumorigenesis and may serve as potential biomarkers for CRC diagnosis and treatment.

Colorectal cancer (CRC) is characterized by DNA methylation, which is associated with genomic instability and tumor initiation. As an important epigenetic regulation, DNA methylation can be used as a potential therapeutic target for CRC. In our study, we downloaded DNA methylation profiles (GSE17648 and GSE29490) and RNA sequencing microarray data (GSE25070 and GSE32323) from the Gene Expression Omnibus (GEO) database. As a result, 14 aberrantly methylated differentially expressed genes (DEGs) were screened according to the different criteria. We further validated these DEGs in The Cancer Genome Atlas (TCGA) database and obtained Pearson's correlation coefficient (COR) for the relationship between gene expression and DNA methylation. Three candidate genes (SOX9, TCN1, and TGFBI) with COR greater than 0.3 were screened out as Hub genes. The receiver operating characteristic result indicated that SOX9 and TGFBI effectively serve as biomarkers for the early diagnosis of CRC. Furthermore, the potential prognosis of the Hub genes for CRC patients was evaluated. Only TGFBI, which is regulated by methylation, can predict patient disease-free survival. Additionally, we examined the methylation level of the Hub genes in CRC cells in the Cancer Cell Line Encyclopedia database. Considering that methylation status tends to be highly modified on CpG islands in tumorigenesis, we screened the CpG island methylation of TGFBI based on the TCGA database and verified its diagnostic value in the GEO database. Our result revealed two Hub genes (TCN1 and TGFBI) whose aberrant expressions were regulated by DNA methylation. Additionally, we uncovered the hypermethylation of TGFBI on CpG islands and its clinical value in the diagnosis of CRC. K E Y W O R D S biomarker, colorectal cancer, DNA methylation, TGFBI