Context: Ultrasound (US) elastography (USE) was recently been reported as a sensitive, noninvasive tool for identifying thyroid cancer. However, the accuracy of this technique is hampered by the intra- and interoperator variability, some US features of the nodule, and the coexistence of autoimmune thyroid disease (ATD).
Objectives: The purpose of this article was to assess the accuracy of USE findings in the differential diagnosis of thyroid nodules compared with other US features to evaluate its feasibility in the presence of ATDs and identify the strain index (SI) cutoff with the highest diagnostic performance.
Design: We evaluated 528 consecutive patients for a total of 661 thyroid nodules. All nodules underwent fine-needle aspiration cytology (FNAC) and USE evaluation. The SI was calculated as a ratio of the nodule strain divided by the strain of the softest part of the surrounding normal tissue.
Results: The median SI value was significantly higher in THY4 and THY5 than in THY2 nodules in ATD-positive, ATD-negative, and ATD-unknown patients. The cutoff of SI for malignancy was estimated to be 2.905 by receiver operating characteristic curve analysis in a screening set (379 FNAC results), and then tested in a replication set (252 FNAC results). In all cases, a SI value of ≥ 2.905 conferred to the nodule a significantly greater probability of being malignant. This SI cutoff had the greatest area under the curve, sensitivity, and negative predictive value, compared with the conventional US features of malignancy.
Conclusion: The elastographic SI has a high sensitivity, specificity, and negative predictive value for the diagnosis of thyroid malignancy both in the presence and in the absence of ATD. If our data on USE are also confirmed in THY3 nodules, FNAC could be avoided in a number of thyroid nodules with certain features.