Diagnostic Models for Predicting Follicular Thyroid Carcinomas Using Circulating Plasma MicroRNAs.

BACKGROUND: Follicular thyroid carcinoma (FC) accounts for 10-15% of all thyroid cancers. FC is challenging to diagnose using fine-needle aspiration (FNA), making diagnostic thyroidectomy the standard approach. Recent studies have explored the use of circulating microRNAs for thyroid cancer diagnosis. This study evaluated the diagnostic value of circulating miRNAs in plasma for FC to potentially reduce unnecessary thyroidectomies and repeat invasive procedures. METHODS: This study was a retrospective observational study, and included consecutively selected 90 patients who underwent thyroidectomy at Pusan National University Hospital between January 2013 and February 2024 and were diagnosed with FC (49 patients) or follicular thyroid adenoma (FA) (41 patients) on final histopathology. Of these, 58 patients were enrolled in the utility assessment and 32 patients were included in the validation test. Among the 58 patients included in the utility assessment, microarray analysis was conducted on 15 patients who were randomly selected to identify novel plasma miRNAs. Next, TaqMan qRT-PCR was performed to evaluate the diagnostic utility of five plasma miRNAs and to develop a predictive model capable of predicting FC from FA using logistic regression as the utility assessment on 58 patients. Finally, in the validation test, TaqMan qRT-PCR and statistical analysis were conducted again on 32 patients and the constructed predictive models, verifying the accuracy of the predictive model. RESULTS: Using microarray analysis, a novel miRNA, miR-6085, was identified for its distinguishing capability between FC and FA. In the utility assessment, miR-6085, miR-146b-5p, miR-221, and miR-222 were significantly upregulated in the FC group. A predictive model combining these four miRNAs showed strong diagnostic value for FC, with an AUC of 0.928 (0.843, 1.000), sensitivity of 94.7% (84.2, 100), specificity of 86.4% (68.2, 100). The accuracy of this model was 76.2% (52.8, 91.8) in the validation test. CONCLUSIONS: A model combining four miRNAs (miR-6085, miR-146b-5p, miR-221, and miR-222) demonstrated high sensitivity, specificity, and accuracy, suggesting that it could be a useful tool for differentiating FC from FA.