Comprehensive analysis of circRNA expression pattern and circRNA-miRNA-mRNA network in oral squamous cell carcinoma

Objective: CircRNAs are critical gene modulators in tumor initiation and progression. However, the expression pattern and molecular pathogenesis of circRNAs in oral squamous cell carcinoma (OSCC) are still poorly characterized. Methods: RNA sequencing with CIRCexplorer2 pipeline was performed to identify circRNAs in 46 tumor-normal paired tissues from OSCC patients. Another set of 48 head and neck squamous cell carcinoma samples from the MiOncoCirc database were utilized as an independent validation. Results: Of the 1276 identified high-confidence circRNAs, 154 were differentially expressed between tumor and normal tissues (log(2)vertical bar Fold Change vertical bar >= 1 and false discovery rate < 0.05). CircRNAs expression was globally down-regulated in tumors compared to normal tissues (P = 9.44 x 10(-14)). Correlation analysis demonstrated that the global expression of circRNAs was positively related to tumor infiltrating lymphocyte (P = 1.10 x 10(-4)) and stromal signature (P = 2.70 x 10(-3)) whereas negatively associated with cell proliferation markers (P = 4.32 x 10(-2)). CircRNAs-miRNAs-mRNAs regulatory network revealed 6574 interactions, and the target genes were enriched in extracellular matrix and immune-related pathways. Survival analysis were performed on target genes in immune-related pathways, and 20 genes were significantly associated with the prognostic status of OSCC in The Cancer Genome Atlas cohort. The risk model constructed with above 20 genes was associated with the prognosis status of OSCC (HR = 3.28, P = 5.06 x 10(-11)), and the result was validated in an independent study (GSE41613) (HR = 2.06, P = 1.73 x 10(-2)). Conclusion: CircRNAs showed a global down-regulation pattern in OSCC tissues, and genes regulated by circRNAs primarily involved in immune and extracellular matrix pathways, which could also affect the OSCC prognosis, indicating that they may serve as potential prognostic biomarkers.