New genetic testing developed in Italy may boost the accuracy of thyroid cancer detection while reducing unnecessary diagnostic thyroidectomies by nearly 50%.
Investigators at the University of Pisa developed models to measure gene expression levels in 93 thyroid tumor cell samples obtained with fine-needle aspiration. Examining aspirated cells is considered the most accurate and cost-effective way to rule out a thyroid cancer diagnosis. But these analyses don’t always give clear-cut results. In about 30% of samples, a definitive diagnosis can’t be made and patients have part of the thyroid gland removed so pathologists can examine a larger tissue sample to determine whether the tumor is malignant. However, malignancies are diagnosed in only 20% of these surgeries, indicating that the procedure often doesn’t benefit the patient.
In evaluating whether gene expression levels could accurately predict thyroid malignancies, the researchers in Pisa narrowed their search to 8 genes that have been linked with thyroid cancer. Their analysis showed that expression levels of the 8 genes correctly predicted malignancies 89% of the time.
The study, released online today in BMC Cancer, recommends 3 ways that gene expression testing can be used in a stepwise fashion with fine-needle aspiration to preoperatively diagnose thyroid cancer.
Of the cell samples analyzed in the study, more than half of malignancies had a specific mutation in the BRAF gene. The researchers suggested that if a clear diagnosis can’t be made with initial tests of aspirated cells, the cells then be evaluated for the BRAF mutation. If it’s present, the cells should be considered malignant. But if the diagnosis still is uncertain, low activity of another gene, KIT, also is a strong predictor of malignancy. If uncertainty persists after BRAF and KIT testing, the researchers said the 8-gene assay should be used. If all the tests fail to give definitive results, then patients could be referred for diagnostic thyroidectomy.
“In this study we developed a molecular approach that is able to correctly classify as certain benign 46% of [indeterminate] lesions,” the researchers wrote. “The model here presented is highly accurate and may provide a tool to overcome the difficulties in today’s preoperative diagnosis of thyroid malignancies.”