Among 114 patients with HNs who exhibited “hot” and “hyperfunction” results on thyroid scan imaging between 2008 and 2017, 73 were included in the analysis. Forty-one patients were excluded for the following reasons: no thyroid US was conducted within 3 months before or after the scan; the location of the nodule seen on the scan and the US did not agree (left/right and upper/mid/low); when multiple nodules did not have clear boundaries, or when the nodule was not clear on US; a history of thyroid surgeries; and patients taking anti-thyroid drugs or positive for anti-TSH receptor antibody. Additionally, 13 patients with multi-nodular goiter were excluded in a further analysis investigating the effect of TSH suppression, because multiple nodules can influence TSH levels (Fig. 1). As a control group, 188 patients with “cold” nodules on thyroid scan imaging were included in the analysis to compare differences in clinical and US findings. Electronic medical records and US findings were retrospectively reviewed. Institutional Review Board approval was obtained for this study (2018-04-009).

Using a single-head gamma camera (LEM Plus, Siemens, Erlangen, Germany) equipped with a pinhole collimator, thyroid scans in the anterior view were acquired 20 min after intravenous administration of 185 MBq of Tc-99m-pertechnetate. For imaging, patients were placed in the supine position with their neck extended; marking sources for size measurements were located on both lateral sides of the neck. HNs were defined as nodules that demonstrated greater tracer uptake than the surrounding normal thyroid tissue, and could be differentiated from the thyroid parenchyma. In cases of multi-nodular goiter on scanning, representative nodules with the clearest boundary were selected and analyzed using US.

In all cases, the sonographic images included both transverse and longitudinal real-time imaging of thyroid nodules using PACS (Picture Archiving Communications System, TechHeim, Seoul, Korea). All US images were obtained using an HDI 5000 system (Philips Medical Systems, Bothell, WA, USA) equipped with 7.5 to 12 or 8 to 15 MHz linear transducers. One endocrinologist, with 10 years' experience in thyroid US, reviewed thyroid nodule characteristics including location, size, volume (width×length×height×0.52), shape, margin, content, echogenicity, calcification, presence of halo, and vascularity according to previous recommended terminology.8) Depending on the nodule contents, the nodules were classified as solid, predominantly solid (>50% solid for a mixed nodule), predominantly cyst (>50% cyst for a mixed nodule), and pure cyst. Nodular shapes were classified as ovoid, round, TDW (when the anteroposterior diameter of the nodule was longer than its transverse diameter on a transverse or longitudinal plane), or irregular (when a nodule was neither ovoid to round nor TDW). Nodular margins were categorized as smooth, spiculated (obviously discernible, but not-smooth edge showing speculation, microlobulation), ill-defined margin (poorly demarcated margin which cannot be obviously differentiated form adjacent thyroid tissue). Echogenicity of the nodules was classified as hypoechoic (hypoechoic relative to thyroid parenchyma), isoechoic (same echogenicity as that of thyroid parenchyma), or hyperechoic (hyperechoic relative to thyroid parenchyma). Each nodule was also classified as showing microcalcification (echogenic foci of 1 mm or less), macrocalcification (when punctuate echogenic foci were larger than 1 mm in size), or rim calcification (peripheral curvilinear echogenic rim). Nodular echotexture was determined to be homogeneous or heterogeneous. Nodular vascularity was classified as none (absence of intranodular or perinodular vascularity), internal (intranodular vascularity without perinodular vascularity), peripheral (presence of circumferential vascularity at margin of nodule), or both (intranodular vascularity with perinodular vascularity) by assessed Doppler. The presence of a halo (thin or thick hypoechoic rim surrounding nodule) was also noted.8)

The malignancy risk for each thyroid nodule was classified according to the malignancy risk criteria published by the American Thyroid Association (ATA).1) When fine-needle aspiration (FNA) or surgeries were performed, the results of aspiration and the final surgical outcomes were verified.

Serum TSH concentrations (reference range, 0.25–4 uIU/mL [lower detection limit, 0.03 uIU/mL]) were measured using an immunoradiometric assay (RIA-gnost rhTSH, cisbio bioassays, France). TSH suppression was defined as HNs with lower TSH levels than the low-normal reference level (TSH <0.25 mIU/mL). The serum levels of free T4 (normal range, 0.78–1.94 ng/dL) were measured using an fT4 radioimmunoassay (RIA) kit (fT4-CTX, DiaSorin S.p.A.). The serum levels of total T3 (reference range, 80–100 ng/dL) were measured using an RIA kit (T3-CTK, DiaSorin S.p.A.). TSH receptor antibody levels were determined using the B.R.A.H.M.S. TRAK human RIA kit (B.R.A.H.M.S. GmbH, Hennigsdorf/Berlin, Germany) according to manufacturer's instructions. TSH receptor antibody levels ≥1.5 IU/L were considered to indicate positive results.

For subgroup analysis according to TSH suppression, two group were classified based on TSH <0.25 mIU/mL. We compared US finding, age, sex, thyroid function test between the two groups. On the assumption that TSH decreases as the size of nodule increases in HNs, receiver operating characteristic (ROC) curve analysis was performed to measure the cut-off values for the size and volume of nodules to predict TSH suppressed HNs.

All statistical analyses were performed using SPSS version 16.0 (IBM Corporation, Chicago, IL, USA) for Windows (Microsoft Corporation, Redmond, WA, USA). The Student's t test was used for comparisons involving continuous variables, and chi-squared tests and Fisher's exact tests were performed for categorical variables. For subgroup analysis according to TSH suppression, the Student's t-test was used for comparison of continuous variable, and chi-squared test for categorical variables between groups. p<0.05 was considered to be statistically significant. ROC curve was drawn to measure the cut-off values for the size and volume of nodules to predict TSH suppressed among HNs.

There were no significant differences in age and female proportion between patients with HNs and those with cold nodules (Table 1). The mean TSH level was lower in patients with HNs than in those with cold nodules (0.82±0.85 uIU/mL vs. 2.15±4.94 uIU/mL, respectively; p=0.025). However, there were no differences in free T4 and total T3 levels between the groups.

Both nodule size and volume were not different between the two groups; there were also no differences in the location of thyroid nodules (Table 1). HNs exhibited partially solid or cystic features more frequently than cold nodules (50.7% vs. 33.0%, respectively; p= 0.004). Suspicious features, including nodule shape, “irregular,” and “taller than wide” were more frequently observed in cold nodules than in HNs (17.6% vs. 5.5%, respectively; p=0.010). In particular, the “taller than wide” shape was not observed in HNs. The proportion of nodules with hypo-echogenicity (p< 0.001) and internal calcification (p=0.019) were higher in cold nodules than in HNs, with a difference that was statistically significant. Internal or peripheral hypervascularity was observed in 95.8% of HNs compared to 44.8% of cold nodules (p<0.001). Peripheral halo was observed more frequently in HNs than in cold nodules (37.0% vs. 21.3%, respectively; p=0.009). However, there were no significant differences in nodule margin and echotexture between the groups. According to the malignancy risk stratification from the ATA guideline, cold nodules exhibited a higher frequency of cases with “highly suspicious” or “intermediate suspicious” categories than HNs. In particular, there were no cases of the “highly suspicious” category in HNs.

FNA was performed in 66 of 73 HNs, and evaluated according to the Bethesda classification.9) Of the 66 HNs that underwent FNA, 6 (9.2%) were nondiagnostic or unsatisfactory, 58 (87.9%) were benign, 1 (1.5%) was atypia of undetermined significance (AUS) or follicular lesion of undetermined significance (FLUS), and 1 (1.5%) was suspicious for malignancy (Table 2). After subsequent FNA, 11 nodules underwent surgery in the authors' hospital. After surgery, 10 benign (3 follicular adenoma and 7 nodular hyperplasia) and 1 malignant (papillary thyroid carcinoma) were confirmed. One nodule was diagnosed as minimally invasive follicular carcinoma by surgery without FNA.

FNA was performed in 156 of the 188 cold nodules; of these, 14 (9.0%) were nondiagnostic or unsatisfactory, 111 (71.2%) were benign, 13 (8.3%) were AUS or FLUS, 4 (2.6%) were suspicious for malignancy, and 12 (7.7%) were malignant. The other two nodules were lymphoma and anaplastic carcinoma. After subsequent FNA, 36 nodules underwent surgery in the authors' hospital. After surgery, 23 benign and 13 malignant nodules were confirmed. The 5 subjects who were diagnosed with suspicious for malignancy or malignant in the FNA result wished to transfer to another hospital for thyroid surgery.

The number of patients with malignant thyroid nodules was 2 (2.7%) in the HN group, and 18 (9.6%) in the cold nodule group.

Data from a total of 60 subjects, excluding 13 subjects with multiple nodular goiters, are shown in Table 3. According to the definition of TSH suppression (TSH <0.25 mIU/mL), 18 (30%) patients exhibited suppressed TSH levels. TSH suppressed nodules demonstrated significantly lower TSH levels, and significantly higher levels of free T4 and total T3. There were no differences in age and female proportion between the groups. TSH suppressed nodules were significantly larger (2.93±1.22 vs. 1.89±1.09 cm; p=0.002) and had a higher volume than normal nodules accompanied by normal TSH levels (9.28±9.94 vs 3.25±5.59 cm³; p=0.004) in HNs. However, there were no differences in location, content, shape, margin, echogenicity, and vascularity between HNs accompanied by suppressed or normal TSH levels.

The area under the ROC curve, according to size and volume, were 0.736 (95% CI=0.664–0.879, p< 0.001) and 0.761 (95% CI=0.698–0.906, p<0.001), respectively (Fig. 2). To predict TSH suppressed HNs, the optimal cut-off value for nodule size was 2.6 cm, with sensitivity of 72.2%, specificity of 76.2%, 56.5% of positive predictive values (PPV) and 86.5 of negative predictive value (NPV). The optimal volume to predict TSH suppressed HNs was 1.13 cm³, with sensitivity of 88.9%, specificity 61.9%, 50.0% of PPV and 92.9% of NPV.