Characterization of thyroid 'follicular neoplasms' in fine-needle aspiration cytological specimens using a panel of immunohistochemical markers: a proposal for clinical application.

The distinction of benign from malignant follicular thyroid neoplasms remains a difficult task in diagnostic fine-needle aspiration cytology, and some discrepant results have been reported for the individual immunocytochemical markers of malignancy proposed so far. The aim of this study was to test if the combined use of a panel of markers could improve the diagnostic accuracy in the preoperative cytological evaluation of 'follicular neoplasms' in an attempt to reduce the number of thyroidectomies performed for benign lesions. The immunocytochemical expression of galectin-3, HBME-1, thyroperoxidase, cytokeratin-19 and keratan-sulfate was retrospectively analyzed in 125 consecutive fine-needle aspiration samples (cell blocks) of indeterminate diagnoses of 'follicular thyroid neoplasm', and compared with their corresponding surgical specimens, including 33 follicular carcinomas, 42 papillary carcinomas and 50 follicular adenomas. Statistical analysis on each marker confirmed that galectin-3 and HBME-1 were the most sensitive (92% and 80% respectively) and specific (94% and 96% respectively) molecules. The use of these two markers sequentially in non-oncocytic lesions (testing HBME-1 as a second marker whenever galectin-3 proved negative) increased the sensitivity and specificity up to 97% and 95% respectively. In oncocytic lesions, HBME-1 proved to be less sensitive, and the sequential combination of galectin-3 and cytokeratin-19 reached 100% of both specificity and sensitivity. Our data showed that, as compared with the use of single markers, the sequential combination of two markers represents the most accurate immunohistochemical panel in managing patients with a fine-needle aspiration biopsy diagnosis of 'follicular neoplasms', especially in otherwise controversial categories such as oncocytic tumours. The combination of three or more markers did not substantially improve the diagnostic accuracy of the test.