American Association of Endocrine Surgeons- Annual Meeting Abstract Papers

A QUANTITATIVE ASSAY FOR DIFFERENTIATING BENIGN FROM MALIGNANT THYROID TUMORS BASED ON GENE EXPRESSION.
Jennifer Rosen, MD, Mei He, MD, Christopher Umbricht, MD PhD, Richard Alexander, MD, Martha Zeiger, MD & Steven Libutti, MD,
National Institutes of Health & The Johns Hopkins Medical Institutions, Baltimore, MD

Background: Thyroid nodules are common, and although most nodules are benign, fine needle aspirations commonly yield indeterminate diagnoses necessitating surgical intervention to exclude the diagnosis of carcinoma. Recently we developed an array-based gene expression approach to diagnose benign versus malignant thyroid lesions resulting in a 6 gene predictor model. The purpose of this study was to verify whether quantitative PCR using this 6 gene model can reliably differentiate benign from malignant thyroid nodules.
Methods: Molecular profiles of 25 benign tumors (15 follicular adenomas, 10 hyperplastic nodules) and 16 malignant tumors (9 papillary thyroid carcinomas, 7 follicular variants of papillary thyroid carcinomas) were analyzed by quantitative reverse transcription PCR (qRT-PCR) analysis using primers and probes developed from our 6 gene diagnostic model. The diagnosis using standard pathologic criteria was used as the "gold standard" for comparison. Ten additional samples were then analyzed as unknowns. A diagnosis-predictor model was built using the 41 training samples, and this class prediction rule was applied to predict the class of the ten unknown samples.
Results: Our predictor model using 41 training samples produced 2 distinct groups. After cross-validation the model was 83% accurate in predicting benign versus malignant with an error rate of 17%. This model correctly predicted four of five of our unknowns to be malignant and four of five to be benign (80% sensitive, 80% specific, 80% positive predictive value, 80% negative predictive value). The cancer gene profiles contained both a known cancer-associated gene (kit) and previously unidentified genes (Hs.296031, Hs.24183, LSM7, SYNGR2, C21orf4).
Conclusions: Molecular diagnosis based on a 6 gene model can differentiate between benign and malignant thyroid tumors with high sensitivity and specificity. The combination of these genetic markers may be used as a reliable test to effectively diagnose the malignant potential of thyroid nodules preoperatively, allowing for fewer surgical procedures for the purpose of diagnosis alone.