Circulating miR-21 as a Potential Biomarker for the Diagnosis of Oral Cancer: A Systematic Review with Meta-Analysis

Head and neck squamous cell carcinoma (HNSCC) is one of the main neoformations of the head–neck region and is characterized by the presence of squamous carcinomatous cells of the multi-layered epithelium lining the oral cavity, larynx, and pharynx. The annual incidence of squamous cell carcinoma of the head and neck (HNSCC) comprises approximately 600,000 new cases globally. Currently, the 5-year survival from HNSCC is less than 50%. Surgical, radiotherapy, and chemotherapy treatments strongly compromise patient quality of life. MicroRNAs (miRNAs) are a family of small noncoding endogenous RNAs that function in regulating gene expression by regulating several biological processes, including carcinogenesis. The main upregulated microRNAs associated with oral carcinoma are miR-21, miR-455-5p, miR-155-5p, miR-372, miR-373, miR-29b, miR-1246, miR-196a, and miR-181, while the main downregulated miRNAs are miR-204, miR-101, miR-32, miR-20a, miR-16, miR-17, and miR-125b. miR-21 represents one of the first oncomirs studied. The present systematic review work was performed based on the preferred reporting items for systematic review and meta-analysis (PRISMA) protocol. A search was carried out in the PubMed and Scopus databases with the use of keywords. This search produced 628 records which, after the elimination of duplicates and the application of the inclusion and exclusion criteria, led to 7 included articles. The heterogeneity of the studies according to the odds ratio was high, with a Q value of 26.616 (p < 0.001), and the I(2) was 77.457% for specificity. The heterogeneity was high, with a Q value of 25.243 (p < 0.001) and the I(2) was 76.231% for sensitivity. The heterogeneity of data showed a Q value of 27.815 (p < 0.001) and the I(2) was 78.429%. Therefore, the random-effects model was selected. The diagnostic odds ratio was 7.620 (95% CI 3.613–16.070). The results showed that the sensitivity was 0.771 (95% CI 0.680–0.842) (p < 0.001) while, for specificity, we found 0.663 (95% CI 0.538–0.770) (p < 0.001). The negative likelihood ratio (NLR) was 0.321 (95% CI 0.186–0.554), and the positive likelihood ratio (PLR) was 2.144 (95% CI 1.563–2.943). The summary ROC plot demonstrates that the diagnostic test presents good specificity and sensitivity, and the area under the curve (AUC), as calculated from the graph, was 0.79.