Background: The 2015 American Thyroid Association (ATA2015) and the American College of Radiology Thyroid Imaging and Reporting Data System (ACR TI-RADS) are two widely used thyroid sonographic systems.
Objectives: To compare the two systems for accuracy of cancer risk prediction.
Methods: Preoperative ultrasound images from 265 patients who underwent thyroidectomy at our hospital from January 2012 to March 2019 were retrospectively categorized by the ACR TI-RADS and ATA2015 systems. Diagnostic performances were compared.
Results: Of 238 nodules assessed, 115 were malignant. Malignancy risks for the five ACR TI-RADS categories were 0%, 7.5%, 11.4%, 59.6%, and 90.0%. Malignancy risks for the five ATA2015 categories were 0%, 6.8%, 17.0%, 55.5%, and 92.1%. The proportion of total nodules biopsied was higher with the ATA2015 system than the ACR TI-RADS system: 88.7% vs. 66.3%. Proportions of malignant nodules and benign nodules biopsied were higher with ATA2015 than with ACR TI-RADS: 93.3% vs. 87.8% and 84.4% vs. 46.3%, respectively. Specificity and sensitivity rates were 53.6% and 84.3%, respectively, for ACR TI-RADS, and 15.5% and 93.3%, respectively, for ATA2015. The two systems showed similarly accurate diagnostic performance (AUC > 0.88). False negative rates for ACR TI-RADS and ATA2015 were 15.6% and 6.6%, respectively. Rates of missed aggressive cancer were similar for the two systems: 3.4% and 3.7%, respectively.
Conclusions: ACR TI-RADS was superior to ATA2015 in specificity and avoiding unnecessary biopsies. ATA2015 yielded better sensitivity and a lower false negative rate. Identification of aggressive cancers was identical in the two systems.