The majority of gene expression profiling studies in cancer research is based on comparison of normal against cancerous tissues. Although these comparisons identify general mechanisms underlying certain cancer types, the heterogeneity of tumors contributes to challenges in the interpretation of results. In this study, Agilent human miRNA microarrays were used to analyze 3 normal tongue tissues and 17 oral squamous cell carcinomas (OSCC) extracted from tongue tumors to identify deregulated miRNAs. Our results showed that miRNA expression pattern provided more precise information when cancers are subcategorized based on clinicopathological parameters. Combinatorial analyses using 4 different parameters (tumor primary site, tumor stage, histological subtype, and human papillomavirus 16 status) revealed OSCC subtypes with specific expressions of selected miRNAs. Nine significantly upregulated miRNAs (miR-424, miR-15b, miR-196a, miR-21, miR-155, miR-9*, miR-9, miR-7-1*, and miR-7) and one downregulated miRNA (miR-486-5p) were applied for this analysis. Our study demonstrated the combine use of several miRNAs with multiple clinicopathological parameters could subcategorize these oral cancers. Kaplan-Meier survival analyses showed that advanced tumor stage (hazard ratio = 3.36; 95% confidence interval, 1.08-10.49; p = 0.037), and/or keratinization status (hazard ratio = 3.34; 95% confidence interval, 1.10-10.12; p = 0.033) are related to poor prognosis of oral cancer patients. Interestingly, all the tumors highly expressing miR-21 were keratinized tumors and the corresponding patients had poor survival rate < 30 months after diagnosis ((hazard ratio = 12.46; 95% confidence interval, 2.57-60.32; p = 0.0017) indicating the prognostic importance of miR-21. The expression of the specific miRNAs (miR-21, miR-424, miR-7, miR-7-1*, and miR-155) overexpressed in the keratinizing tumors was significantly correlated to each other (0.438 < r < 0.615, p < 0.041). In silico target prediction analyses using multiple bioinformatics tools identified candidate genes containing multiple binding sites for these miRNAs. The miRNA signatures are indicators for cancer diagnostics, prognostics, and therapeutics and may lead eventually to individualized clinical treatment.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 191. doi:1538-7445.AM2012-191

©2012 American Association for Cancer Research
2012