Context: British Thyroid Association (BTA), American Thyroid Association (ATA), and American Association of Clinical Endocrinologists (AACE/ACE/AME) recommend for thyroid nodules an ultrasound (US)-based stratification of risk of malignancy. Aim of our study was to assess the diagnostic accuracy of US classification systems and their reliability for indication to fine-needle aspiration (FNA).
Design: Prospective study on 987 thyroid nodules consecutively referred for FNA. US images were independently reviewed by four experts for assignment of malignancy risk. Cytologically benign nodules had confirmation with a second FNA, whereas Bethesda class IV, V, and VI nodules were operated upon. Class III nodules had surgery or follow-up on the basis of clinical, immunocytochemical, and US features.
Results: BTA: Malignancy rate was 2.8% in benign, 10.0% in indeterminate, 51.3% in suspicion, and 80.9% in malignant US class. Sensitivity was 0.74, specificity was 0.92, and accuracy was 0.89. ATA: Malignancy rate was 0.0% in benign, 2.2% in very low suspicion, 3.0% in low suspicion, 5.8% in intermediate, and 55.0% in high suspicion US class. Sensitivity was 0.81, specificity was 0.87, and accuracy was 0.86. AACE/ACE/AME: Malignancy rate was 1.1% in low-risk, 4.4% in intermediate-risk, and 54.9% in high-risk US class. Sensitivity was 0.82, specificity was 0.87, and accuracy was 0.86. K correlation coefficient was 78.9%, 76.9%, and 82.0% for BTA, ATA, and AACE/ACE/AME classifications.
Conclusions: Classification systems had elevated predictive value of malignancy in high-risk classes. ATA and AACE/ACE/AME systems were effective for ruling out indication to FNA in low-US-risk nodules. A similar diagnostic accuracy and a substantial interobserver agreement was provided by the three- and the five-category classifications.