Diagnostic performance of the EU TI-RADS and ACR TI-RADS scoring systems in predicting thyroid malignancy.

Abstract

Several ultrasound scoring systems have been developed to stratify the risk of malignancy of thyroid nodules, including ACR (American College of Radiology) and EU (European) TI-RADS. This study aimed to assess the diagnostic performance of these two classifications using histology as a reference standard. It was a single-centre, retrospective study including 156 patients who underwent thyroidectomy. Ultrasound data of 198 nodules (99 malignant nodules and 99 benign nodules) were analysed. Both classifications were applied for all nodules. Ultrasound criteria associated with malignancy were solid composition (OR=7.81; p < 10-3 ), hypoechoic character (OR=16.42; p < 10-3 ), irregular contours (OR=7.47; p < 10-3 ), taller-than-wide shape (OR=3.58; p = 0.02), microcalcifications (OR=3.02; p = .006) and the presence of cervical adenopathy (OR=3.89; p = .006). The prevalence of malignancy was 15.5%, 69% and 76.9% for EU TI-RADS categories 3, 4 and 5, respectively. It was 33.3%, 57% and 91.1% for ACR TI-RADS categories 3, 4 and 5, respectively. For category 5, EU TI-RADS and ACR TI-RADS had sensitivities of 60% and 41%, specificities of 82% and 96%, respectively. For categories 4 and 5 combined, the diagnostic performance of these two classification systems became comparable with a sensitivity of 89% and 86% for EU-TIRADS and ACR-TIRADS, respectively. The area under the ROC curve was 0.81 for the EU TI-RADS classification and 0.82 for the ACR TI-RADS classification. EU TI-RADS and ACR TI-RADS scoring systems seem to be comparable in predicting malignancy in thyroid nodules.