The gold standard of thyroid nodule examination? Prospective validation of the ACR TI-RADS in a secondary referral center

Abstract

The aim of this prospective study was the validation of the risk stratification of thyroid nodules using ultrasonography with the American College of Radiology Thyroid Imaging, Reporting and Data System (ACR TI-RADS) and partly in comparison to American Thyroid Association (ATA) guidelines in a secondary referral center. Fine needle aspiration biopsy (FNA) (n=605) and histological examinations (n=63) were the reference standards for the statistical analysis. ACR TI-RADS cut-off value: TR4 with sensitivity 85.7 %, specificity 54.1 %, PPV 58.5 %, accuracy 67.7 % (AUC 0.738; p<0.001). ATA cut-off value: "high suspicion" with sensitivity 80 %, specificity 83.3 %, PPV 80 %, accuracy 81.8 % (AUC 0.800; p=0.0025). 18.4 % nodules (3 malignant) could not be assigned to a proper ATA US pattern group (p<0.0001). Both ACR TI-RADS and ATA have allowed fair selection of nodules requiring FNA with superiority of ACR TI-RADS according to classification of all thyroid nodules to the proper group. According to ACR TI-RADS almost one third of the patients were incorrectly classified with 17.9 % missed thyroid carcinomas, exclusively micropapillary carcinomas, even though, the amount of FNA would be reduced to 48 %.

Fig. 1

ROC curve analysis of ACR TI-RADS with the best fit curve for TR4. Sensitivity=85.7 %, Specificity=54.1 %, PPV=58.5 %, NPV 83.3 %, AUC=0.738 (95 % CI 0.596–0.834), Youden Index 0.39; p<0.001 (n=65). PPV – positive predictive value, NPV – negative predictive value.