RESUMO
BACKGROUND: In the ultrasound evaluation of masses, elastography measures stiffness, which may predict malignancy. Studies of small or selected subgroups suggest that elastography may be useful in the evaluation of thyroid nodules (TNs). We prospectively tested the hypothesis that TN stiffness, as measured by strain elastography (SE), is an independent predictor of thyroid cancer (TC) in unselected TNs. METHODS: In 706 unselected patients with 912 TNs meeting the ATA criteria for a fine-needle aspiration biopsy (FNAB), we first performed conventional thyroid ultrasound and SE. Nodule stiffness was graded from least to most stiff by an elastography score (ES) of ES 0 to ES 3. Surgical resection was recommended for FNAB results that were not clearly benign. Bivariate and multivariate regression analyses identified the independent predictors of TC. RESULTS: There were 86 malignant TNs. ES was a significant predictor of TC (p=0.0001). The prevalence of TC was 57 of the 158 TNs (36.1%) for the ES 3 group, 12 of the 158 TNs (7.7%) for the ES 2 group, 16 of the 565 TNs (2.8%) for the ES 1 group, and 1 of the 33 TNs (3%) for the ES 0 group. By multivariate regression analysis, the independent predictors of TC were ES, microcalcifications, hypoechogenicity, and isthmus location. The positive predictive value (PPV) of ES was 36.1%, which was similar to the PPV of microcalcifications (35.9%), but greater compared with hypoechogenicity (13.6%) and isthmus location (16.9%). The negative predictive value (NPV) of ES was 97.2%, which was better than any other predictor for malignancy. CONCLUSIONS: We conclude that TN stiffness measured by elastography is an independent predictor of TC with a PPV that is equal to or greater than that of conventional ultrasonographic characteristics. NPV was greater than any other predictor of malignancy.