Performance of the ACR TI-RADS and EU TI-RADS scoring systems in the diagnostic work-up of thyroid nodules in a real-life series using histology as reference standard

Abstract

Objective: The ultrasonographic scores EU TI-RADS and ACR TI-RADS were introduced to give the clinicians indications for fine needle aspiration cytology (FNAC). The predictive role of these scores was never evaluated and compared in a surgical series of patients. The aim of this study was to evaluate the ex post diagnostic accuracy of EU TI-RADS and ACR TI-RADS in a real-life series of thyroidectomized patients and to evaluate the ‘missing’ thyroid cancer following the operational indications of these scores. Design: Retrospective monocentric cohort study. Methods: In total, 255 patients (harboring 304 nodules) undergoing thyroidectomy for benign and malignant thyroid conditions were enrolled. The prevalence of thyroid malignancy for each class of ACR TI-RADS and EU TI-RADS, their diagnostic accuracy, the number of ‘unnecessary’ FNAC and the number of ‘missed’ cancers were evaluated. Results: ACR TI-RADS and EU TI-RADS score had similar and satisfactory accuracy values for predicting thyroid malignancy (AUC: 0.835 for ACR TI-RADS vs 0.827 for EU TI-RADS). The ACR TI-RADS and EU TI-RADS categories (suspicious vs non-suspicious), age, sex and presence of a single nodule significantly and independently predicted the presence of malignancy in a logistic regression model. An ex post analysis according to the indications for FNAC for each score indicated that 31 and 16 cases of cancer would have been missed by ACR TI-RADS and EU TI-RADS scores, respectively. Conclusions: ACR TI-RADS and EU TI-RADS display a good performance in predicting thyroid cancer when histology is taken as reference standard, but additional clinical judgement is required to decide the indication for FNAC.