Molecular portrait of thyroid carcinomas: microrna expression profile and its prognostic value

Abstract

Thyroid malignancies remain an important clinical challenge due to their increasing incidence and marked biological heterogeneity. Although differentiated thyroid carcinomas generally have a favorable prognosis, a subset of patients develops recurrent disease, regional or distant metastases, and progressive tumor growth. These observations highlight the need for reliable molecular biomarkers capable of improving risk stratification and prognostic assessment. Owing to their tissue-specific expression patterns, remarkable stability in biological fluids, and involvement in the regulation of key cellular pathways, microRNAs have emerged as promising candidates for molecular diagnosis, prognostication, and personalized management of thyroid carcinomas. Objective. To identify and characterise a panel of differentially expressed microRNAs in thyroid carcinomas and to evaluate their diagnostic and prognostic significance through analysis of associations with patient age, tumour stage, regional lymph node involvement, histological subtype, and overall survival. Materials and Methods. A retrospective in silico bioinformatic study was conducted to identify differentially expressed microRNAs in thyroid carcinomas and evaluate their clinical and prognostic significance. The study consisted of two sequential analytical phases. During the first phase, candidate microRNAs were screened using three independent Gene Expression Omnibus (GEO) datasets (GSE104006, GSE191117, and GSE62054), containing expression profiles of malignant and non-malignant thyroid tissues. During the second phase, the clinical associations and prognostic relevance of the selected microRNAs were validated using the UALCAN web platform based on The Cancer Genome Atlas (TCGA) dataset. Functional annotation and target-gene analysis were performed using TarBase v8.0. MicroRNA–target interaction networks were constructed with miRNet v2.0, whereas pathway and biological process enrichment analyses were conducted using Diana Tools miRPath v3. Potential associations between identified target genes and thyroid carcinogenesis were evaluated using DisGeNET v7.0. Results and Discussion. An integrated bioinformatic analysis identified a panel of six differentially expressed microRNAs associated with thyroid carcinomas. Compared with non-malignant thyroid tissue, significant overexpression of hsa-miR-15a-5p (1.40-fold), hsa-miR- 146b-5p (63.39-fold), hsa-miR-221-5p (9.27-fold), and hsa-miR-222-3p (7.67-fold) was observed (p<0.0001 for all comparisons). In contrast, the expression levels of hsa-miR-199b-5p and hsa-miR-484 were significantly reduced (3.29-fold each; p<0.0001). Analysis of clinical and prognostic associations demonstrated that relatively higher expression of hsa-miR-199b-5p within the carcinoma cohort was associated with a 35% reduction in 5-year overall survival (p<0.0001), identifying this microRNA as a potential independent predictor of adverse clinical outcome. Conclusions. A panel of six differentially expressed microRNAs associated with thyroid carcinomas was identified through integrated bioinformatic analysis. Several panel components demonstrated significant associations with clinicopathological characteristics and overall survival, supporting their diagnostic and prognostic relevance. These findings provide a scientific basis for the development and subsequent clinical validation of a molecular test for the preoperative differential diagnosis of thyroid nodules, particularly in patients with indeterminate Bethesda III–IV cytology.