Effectiveness evaluation of the Thyroid Imaging Report and Data System proposed by American Radiological Society (2017) (ACR-TIRADS) for differential diagnosis in thyroid nodules / 中华超声影像学杂志

ABSTRACT

Objective@#To evaluate the effect of the Thyroid Imaging Report and Data System proposed by American Radiological Society (ACR-TIRADS) for differential diagnosis in thyroid nodules, and compare ACR-TIRADS to the TIRADS proposed by Kwak et al.(K-TIRADS) and the ultrasound-based risk stratification system evaluated by American Thyroid Association (ATA-Risk Stratification).@*Methods@#The clinical data of 1 760 patients with 1 912 thyroid nodules from 8 hospitals in Jiangsu province were retrospectively analysed. All of them were categorized based on ultrasound-based risk stratification systems. The ROC curve was established to assess and compare the diagnostic value of the systems.@*Results@#The area under the ROC curve (AUC) of ACR-TIRADS was 0.830, with high sensitivity and negative predictive value (86.9% and 87.5%, respectively), and relatively low specificity and positive predictive value (64.1% and 62.9%, respectively). The sensitivity and specificity of K-TIRADS were up to 84.9% and 76.1%, respectively. The AUC of ATA-Risk Stratification was 0.852, with relatively high specificity (83.4%), and low sensitivity (79.4%). There were significant differences in the AUC among the three ultrasound-based risk stratification systems, of which K-TIRADS was the highest (P<0.001). There was no significant difference in sensitivity of ACR-TIRADS and K-TIRADS (P=0.137), but significantly higher than that of ATA-Risk Stratification (P<0.001). There were significant differences in the specificity among the three systems, of which ATA-Risk Stratification was the highest (P<0.001). In addition, there were 109 nodules (5.7%) couldn′t be classified based on ATA-Risk Stratification, with high malignancy rate of 31.2%.@*Conclusions@#The diagnostic efficiency of ACR-TIRADS is good, but lower than K-TIRADS and ATA-Risk Stratification. ACR-TIRADS has the highest sensitivity, and ATA-Risk Stratification has the highest specificity, while the overall diagnostic efficiency of K-TIRADS is the best.