Digital gene expression analysis might aid in the diagnosis of thyroid cancer
Background Thyroid cancer represents approximately 90% of endocrine cancers. Difficulties in diagnosis and low inter-observer agreement are sometimes encountered, especially in the distinction between the follicular variant of papillary thyroid carcinoma (fvptc) and other follicular-patterned lesions, and can present significant challenges. In the present proof-of-concept study, we report a gene-expression assay using NanoString nCounter technology (NanoString Technologies, Seattle, WA, U.S.A.) that might aid in the differential diagnosis of thyroid neoplasms based on gene-expression signatures.
Methods Our cohort included 29 patients with classical papillary thyroid carcinoma (ptc), 13 patients with fvptc, 14 patients with follicular thyroid carcinoma (ftc), 14 patients with follicular adenoma (fa), and 14 patients without any abnormality. We developed a 3-step classifier that shows good correlation with the pathologic diagnosis of various thyroid neoplasms. Step 1 differentiates normal from abnormal thyroid tissue; step 2 differentiates benign from malignant lesions; and step 3 differentiates the common malignant entities ptc, ftc, and fvptc.
Results Using our 3-step classifier approach based on selected genes, we developed an algorithm that attempts to differentiate thyroid lesions with varying levels of sensitivity and specificity. Three genes—namely SDC4, PLCD3, and NECTIN4/PVRL4—were the most informative in distinguishing normal from abnormal tissue with a sensitivity and a specificity of 100%. One gene, SDC4, was important for differentiating benign from malignant lesions with a sensitivity of 89% and a specificity of 92%. Various combinations of genes were required to classify specific thyroid neoplasms.
Conclusions This preliminary proof-of-concept study suggests a role for nCounter technology, a digital gene expression analysis technique, as an adjunct assay for the molecular diagnosis of thyroid neoplasms.