1869P - Analytical validation of a thyroid cancer diagnostic method based on the relative quantification of CLDN10, HMGA2 and LAMB3 expression

Abstract

BackgroundFine-needle aspiration biopsy (FNAB) is the standard diagnostic method of thyroid nodules. However, a fraction of the cytology assessments generates indeterminate results. Our group developed a method based on the relative quantification of three transcripts to aid the diagnosis of thyroid lesions. In this study, we aimed to validate the performance and applicability of this diagnostic test in FNAB. MethodsSeventy-five prospectively collected FNAB were evaluated for CLDN10, HMGA2 and LAMB3 gene expression (EIF2B1 as reference gene) by RT-qPCR, using the leftover cells inside the needles. A mathematical model previously established was applied and the scores were compared with the Bethesda categories and the final postoperative report. ResultsWe demonstrated the feasibility of the molecular test, even with very small yields of RNA. The three genes tested showed higher expression in the conclusive malignant cases (15 Bethesda VI) in relation to the conclusive benign nodules (35 Bethesda II). The application of the diagnostic algorithm allowed the distinction of these samples with 93% sensitivity (14/15) and 100% specificity (35/35). Among the patient with indeterminate nodules (Bethesda III-V), the test was able to correctly discriminate malignant from benign nodules with 71% sensitivity (5/7 tumors) and 100% specificity (4/4 benign lesions). Fourteen indeterminate cases are being followed-up and were not submitted to surgery to date. Furthermore, we observed that larger tumors (>1cm), presenting lymph node metastasis and extrathyroidal extension were associated with higher algorithm scores (total of 22 malignant cases). ConclusionsA thyroid diagnostic method based on the analysis of only four genes by RT-qPCR (3 targets and 1 reference) revealed high accuracy and applicability even in very limited FNAB material, presenting also a potential prognostic role. Legal entity responsible for the studyThe authors. FundingFAPESP. DisclosureAll authors have declared no conflicts of interest.