Comparison of computed tomography features between follicular neoplasm and nodular hyperplasia

Background: To date, appropriate management for Bethesda IV thyroid nodules is controversial, and no specific features of follicular neoplasm and nodular hyperplasia on ultrasonography, computed tomography (CT), or other imaging modalities have been reported. This study aimed to compare CT features of follicular neoplasm and nodular hyperplasia and to determine the specific CT features that could be used to distinguish follicular neoplasm from nodular hyperplasia.

Methods: In 122 patients who underwent preoperative CT of the neck and thyroid surgery, 59 follicular neoplasms and 65 nodular hyperplasias were included. In each case, non-enhanced and contrast-enhanced CT images were obtained, and a single radiologist retrospectively analyzed CT images, including degree and pattern of attenuation, nodular configuration, margin, shape, pattern of calcification, degree and pattern of nodular enhancement, and CT halo sign. A univariate and multivariate logistic regression analyses were used to evaluate the predictive power of each variable and CT features with a high predictive power, respectively.

Results: According to the univariate analysis, iso-attenuation, intraglandular configuration, smooth margin, ovoid shape, decreased enhancement, and absence of CT halo sign were more frequently observed in nodular hyperplasia (p < 0.05), whereas low attenuation, expansile configuration, lobulated margin, taller-than-wide shape, increased enhancement, and presence of computed tomography halo sign were more frequently observed in follicular neoplasm (p < 0.05). Multivariate analysis revealed significant differences in configuration (OR: 2.73, 1.13-6.57), degree of enhancement (OR: 2.14, 1.21-3.78), and presence of CT halo sign (OR: 7.97, 2.74-23.37) between follicular neoplasm and nodular hyperplasia (p < 0.05).

Conclusions: Neck CT may be helpful for distinguishing follicular neoplasm from nodular hyperplasia.

Trial registration: Rretrospectively registered.