High Diagnostic Accuracy Based on CLDN10, HMGA2, and LAMB3 Transcripts in Papillary Thyroid Carcinoma

Abstract

Context: Thyroid nodules are common in adult population and papillary thyroid carcinoma (PTC) is the most frequent malignant finding. The natural history of PTC remains poorly understood and current diagnostic methods limitations are responsible for a significant number of potentially avoidable surgeries. Objective: This study aimed to identify molecular markers to improve the diagnosis of thyroid lesions. Design: Gene expression profiling was performed using microarray in 61 PTC and 13 surrounding normal tissues (NT). A reliable gene list was established using cross-study validation (138 matched PTC/NT from external databases). Results were collectively interpreted by in silico analysis. A panel of 28 transcripts was evaluated by RT-qPCR, including benign thyroid lesions (BTL) and other follicular cell-derived thyroid carcinomas (OFDTC). Adiagnostic algorithm was developed (training set: 23 NT, 8 BTL, and 86 PTC), validated (independent set: 10 NT, 140 BTL, 120 PTC, and 12 OFDTC) and associated with clinical features. Results: GABRB2 was ranked as the most frequently up-regulated gene in PTC (cross-study validation). Altered genes in PTC suggested a loss of T-4 responsiveness and dysregulation of retinoic acid metabolism, highlighting the putative activation of EZH2 and histone deacetylases (predicted in silico). An algorithm combining CLDN10, HMGA2, and LAMB3 transcripts was able to discriminate tumors from BTL samples (94% sensitivity and 96% specificity in validation set). High algorithm scores were associated with regional lymph node metastases. Conclusions: A promising tool with high performance for PTC diagnosis based on three transcripts was designed with the potential to predict lymph node metastasis risk.