Preoperative detection of malignancy in fine-needle aspiration cytology (FNAC) smears with indeterminate cytology (Bethesda III, IV) by a combined molecular classifier

Abstract

AIMS: Analysis of molecular markers in addition to cytological analysis of fine-needle aspiration (FNA) samples is a promising way to improve the preoperative diagnosis of thyroid nodules. Previously, we have developed an algorithm for the differential diagnosis of thyroid nodules by means of a small set of molecular markers. Here, we aimed to validate this approach using FNA cytology samples of Bethesda categories III and IV, in which preoperative detection of malignancy by cytological analysis is impossible. METHODS: mRNA, 3 microRNAs, and the mitochondrial/nuclear DNA ratio). The decision tree-based classifier was used to discriminate between benign and malignant tumours. RESULTS: The molecular testing detected malignancy in FNA smears of indeterminate cytology with 89.2% sensitivity, 84.6% positive predictive value, 92.9% specificity and 95.2% negative predictive value; these characteristics are comparable with those of more complicated commercial tests. Residual risk of malignancy for the thyroid nodules that were shown to be benign by this molecular method did not exceed the reported risk of malignancy for Bethesda II histological diagnosis. Analytical-accuracy assessment revealed required nucleic-acid input of ≥5 ng. CONCLUSIONS: The study shows feasibility of preoperative differential diagnosis of thyroid nodules of indeterminate cytology using a small panel of molecular markers of different types by a simple PCR-based method using stained FNA smears.