Diagnostic Performance of American College of Radiology TI-RADS: A Systematic Review and Meta-Analysis.

OBJECTIVE. The objective of our study was to investigate the diagnostic performance of the American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TI-RADS) in the risk stratification of thyroid nodules. MATERIALS AND METHODS. A literature search of MEDLINE, Embase, Cochrane Library, Web of Science, and Google Scholar was performed for articles published before July 31, 2019. We included studies using ACR TI-RADS for stratification of thyroid nodules and cytology results from ultrasound-guided fine-needle aspiration biopsy (FNAB), pathology results from surgical resection of the thyroid, or both FNAB cytology and surgical pathology results as the reference standard. Summary estimates of sensitivity and specificity were calculated with the bivariate random-effects modeling and are visually presented in forest plots. We performed multiple subgroup analyses and meta-regression to explore the effects of various clinical settings. We compared ACR TI-RADS with the American Thyroid Association (ATA) guidelines and Korean Thyroid Imaging Reporting and Data System (TIRADS) in studies providing head-to-head comparison. RESULTS. Sixteen studies with 18,614 patients involving a total of 21,882 nodules, were included. The pooled sensitivity and specificity of ACR TI-RADS were 0.89 (95% CI, 0.81-0.93) and 0.70 (95% CI, 0.60-0.78), respectively. The calculated area under summary ROC curve was 0.86 (95% CI, 0.83-0.89), with a diagnostic odds ratio of 18.46 (95% CI, 9.77-34.88). Meta-regression revealed that patient number was a significant factor for heterogeneity (p = 0.02). Ten studies compared the performance of ACR TI-RADS and ATA guidelines: The pooled sensitivity was 0.83 versus 0.87 (p = 0.5), respectively, and the pooled specificity was 0.69 versus 0.50 (p = 0.1). In six studies providing direct comparison of ACR TI-RADS and Korean TIRADS, the pooled sensitivity was 0.85 versus 0.91 (p = 0.13), and the pooled specificity was 0.57 versus 0.24 (p < 0.001). CONCLUSION. ACR TI-RADS showed favorable sensitivity and moderate specificity in risk stratification of thyroid nodules. The use of ACR TI-RADS could avoid a large number of unnecessary biopsies, although at the cost of a slight decline in sensitivity.