Expression Analysis, Diagnostic and Prognostic Accuracy of microRNAs in Oral Cancer: An Overview With Meta-Analysis

BACKGROUND: Oral squamous cell carcinoma remains a significant global health burden, largely due to delayed diagnosis and limited prognostic tools. MicroRNAs (miRNAs), as post-transcriptional gene regulators, have shown promise as biomarkers for both early detection and outcome prediction. However, variability in study design, analytical approaches, and biospecimen sources has hindered clinical translation. This meta-analysis aimed to synthesize current evidence on the diagnostic accuracy, prognostic significance, and differential expression of miRNAs in oral cancer. METHODS: A comprehensive review was conducted, including both systematic reviews and original studies. Data was extracted from primary research articles of the reviews to ensure consistency in effect size derivation. Prognostic outcomes were evaluated using seven independent hazard ratios (HRs) reflecting miRNA expression and survival association. Diagnostic accuracy was assessed from 12 studies reporting or allowing for approximation of the area under curve value. Differential expression analysis was based on 11 entries, using standardized mean differences (SMDs), either directly reported or estimated from fold-change values. Random-effects models were applied for all pooled analyses. Heterogeneity was assessed using I² and τ² statistics, and publication bias was tested via Egger's test. Subgroup analyses were conducted by sample type, effect type and direction of miRNA dysregulation. RESULTS: Prognostic meta-analysis demonstrated a strong and statistically significant association between upregulated miRNA expression and poor survival outcomes in oral cancer. The pooled hazard ratio from seven independent entries was 2.31 (95% confidence interval CI: 1.96-2.73), indicating more than a twofold increased risk of adverse outcomes associated with miRNA overexpression. No heterogeneity was observed (I² = 0%), and there was no evidence of publication bias or outlier influence. Diagnostic meta-analysis yielded a pooled area under curve of 0.787 (95% CI: 0.721-0.841), indicating moderate-to-high diagnostic performance of miRNA-based tests. While diagnostic heterogeneity was low in some entries, results from serum and saliva samples were highly variable, reflecting inconsistency in noninvasive miRNA biomarker performance. Expression analysis across 11 entries resulted in a pooled SMD of -0.05 (95% CI: -0.19 to 0.09), suggesting no overall net change. However, subgroup analyses revealed that miRNAs classified as oncogenic were significantly upregulated (SMD = +0.29), while tumour suppressors were consistently downregulated (SMD = -0.52), trends that were otherwise masked in the aggregated analysis. CONCLUSION: This meta-analysis confirms that dysregulated miRNA expression is significantly associated with both poorer prognosis and moderate-to-high diagnostic accuracy in oral cancer. The consistent directionality of expression changes highlights the biological relevance of specific miRNAs as either oncogenic or tumour-suppressive agents. Despite promising results, substantial variability in noninvasive sample data underscores the need for methodological standardization. Further prospective validation, harmonization of analytic workflows, and robust multicentre studies are essential to support the clinical integration of miRNA biomarkers in oral cancer detection and management.