Objectives: The purpose of this study was to stratify the malignancy risk of US features, with an emphasis on nodule echogenicity.
Methods: A total of 1,058 nodules of 824 consecutive patients (236 malignant and 822 benign) were included in this study. Malignancy risk of each nodule was analyzed according to US features, with an emphasis on nodule echogenecity, and was stratified into 4-tier categories.
Results: In multivariate analysis, isoechogenicity, indistinct margin, non-solid internal content, and parallel orientation were predictive of benign nodules (P < 0.002), while hypoechogenicity, marked hypoechogenicity, spiculated/microlobulated margin, solid content, nonparallel orientation (taller than wide), microcalcification, and macrocalcification were predictive of malignancy (P ≤ 0.037). Although the presence of US features associated with malignancy was significantly predictive of malignancy in hypoechoic and markedly hypoechoic nodules (P ≤ 0.004), it was not associated with malignancy in isoechoic or hyperechoic nodules. Thyroid nodules could be stratified into four categories according to the malignancy risk: benign (risk 0%), probably benign (risk ≤ 5%), indeterminate (risk > 5 and < 50%), and suspicion of malignancy (risk > 50%).
Conclusions: The US-based four-tier categorization system will be useful for predicting the risk of malignancy and decisions regarding FNA for thyroid nodules.
Key points: • No US feature was predictive of malignancy in isoechoic nodules. • Isoechoic nodules without calcification can be included in the probably benign category. • We suggest a four-tier categorization stratified primarily by nodule echogenecity. • The four-tier categorization of thyroid nodules will be useful for FNA decisions.