Barros-Filho Mateus Camargo, Marchi Fabio Albuquerque, Pinto Clóvis Antônio, Rogatto Silvia Regina, Kowalski Luiz Paulo
International Research Center/AC Camargo Cancer Center (M.C.B.F., F.A.M., C.A.P., S.R.R., S.R.R.), Sao Paulo 01509-010, SP, Brazil; and Faculty of Medicine (S.R.R.), Sao Paulo State University, Botucatu 18618-970, SP, Brazil.
J Clin Endocrinol Metab. 2015 Jun;100(6):E890-9. doi: 10.1210/jc.2014-4053. Epub 2015 Apr 13.
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.
This study aimed to identify molecular markers to improve the diagnosis of thyroid lesions.
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). A diagnostic 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.
GABRB2 was ranked as the most frequently up-regulated gene in PTC (cross-study validation). Altered genes in PTC suggested a loss of T4 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.
A promising tool with high performance for PTC diagnosis based on three transcripts was designed with the potential to predict lymph node metastasis risk.
甲状腺结节在成人中很常见，甲状腺乳头状癌（PTC）是最常见的恶性病变。PTC的自然病史仍知之甚少，目前的诊断方法存在局限性，导致大量本可避免的手术。
本研究旨在识别分子标志物以改善甲状腺病变的诊断。
使用微阵列对61例PTC和13例周围正常组织（NT）进行基因表达谱分析。通过跨研究验证（来自外部数据库的138对匹配的PTC/NT）建立了可靠的基因列表。通过计算机分析对结果进行综合解读。通过RT-qPCR评估了一组28个转录本，包括良性甲状腺病变（BTL）和其他滤泡细胞源性甲状腺癌（OFDTC）。开发了一种诊断算法（训练集：23例NT、8例BTL和86例PTC），进行了验证（独立集：10例NT、140例BTL、120例PTC和12例OFDTC），并与临床特征相关联。
GABRB2被列为PTC中上调最频繁的基因（跨研究验证）。PTC中基因的改变提示T4反应性丧失和视黄酸代谢失调，突出了EZH2和组蛋白脱乙酰酶的假定激活（计算机预测）。一种结合CLDN10、HMGA2和LAMB3转录本的算法能够区分肿瘤与BTL样本（验证集中灵敏度为94%，特异性为96%）。高算法分数与区域淋巴结转移相关。
基于三个转录本设计了一种对PTC诊断具有高性能的有前景的工具，具有预测淋巴结转移风险的潜力。
