We conducted a retrospective cohort study at the Thyroid Cancer Center of Samsung Medical Center, a tertiary referral center in Korea, by analyzing data collected between January 1997 and June 2015. A total of 1,323 patients who met the following inclusion criteria were enrolled. The inclusion criteria were: pathologically proven N1b PTC, total thyroidectomy with central neck dissection and lateral neck dissection, and preoperative results of both US and CT. The exclusion criteria were: a history of the previous thyroidectomy, non-PTC carcinomas (follicular/medullary/anaplastic), mixed type PTC, lobectomy, or distant metastasis.

US scanners (HDI 5000 or IU22; Philips Medical Systems; Bothell, WT, USA) equipped with commercially available 7 to 12 MHz linear array transducers were used for the primary assessment of cervical LNs. The US criteria for cervical LN metastasis were as follows: focal or diffuse hyperechogenicity, microcalcifications, cystic changes, loss of fatty hilum, round shape (long/transverse diameter ratio <1.5), and abnormal vascular patterns (chaotic or peripheral) (678). CT scans were performed preoperatively on patients who showed aggressive features on US scans. Aggressive features included tumor size >1 cm, upper pole-located tumor, multifocality, extrathyroidal extension, and clinically suspicious lymphadenopathy. CT scans were also performed according to the individual clinician's preference at the time of planning a surgery. CT scans were obtained using a 16-detector-row CT scanner (MX8000 Infinite Detector Technology; Philips, Haifa, Israel) with a reconstructed slice thickness of 3 mm for both axial and coronal images. The CT criteria for cervical LN metastasis were as follows: calcification, cystic or necrotic changes, heterogeneous cortical enhancement, and strong enhancement without hilar vessel enhancement (678). Two or more experienced radiologists interpreted each US and CT image. Suspicious LNs were defined as LNs meeting at least one of the US and CT criteria listed above. US and CT results were reported on a “per level” basis (7). For patients with multiple follow-up examinations, we selected the result closest to the operation date for analysis.

At our institution, we perform total thyroidectomy and central neck dissection before lateral neck dissection by following the principles that have been previously described (9). Moreover, we only perform therapeutic lateral neck dissection in patients with clinically suspicious N1b after confirmation based on fine needle aspiration biopsy. We define lateral neck dissection as any excision of lateral neck LNs, including modified radical neck dissection and selective neck dissection (10). Modified radical neck dissection refers to the excision of lateral neck LNs, including levels II–V, with preservation of one or more non-lymphatic structures, such as the spinal accessory nerve, internal jugular vein, or sternocleidomastoid muscle. Level I dissection is not performed unless and otherwise indicated. Selective neck dissection refers to the excision of suspicious lateral neck LNs accompanied by preservation of one or more of the LN groups that are routinely removed in modified radical neck dissection. During surgery, the cervical LN levels present in the dissected specimens are classified individually with a labeling suture.

Surgical specimens were microscopically examined by two or more experienced pathologists. The following histopathologic parameters were assessed: cell type of the main lesion, primary tumor size (longest diameter of the largest lesion), location, multifocality, bilaterality, extrathyroidal extension, lymphovascular invasion, margin involvement, regional LN metastasis, and underlying thyroid conditions, such as chronic lymphocytic thyroiditis. To distinguish bilaterality from multifocality, multifocality was defined as having two or more lesions of conventional PTC in one lobe, regardless of the presence of bilaterality. Of particular note, histopathology results of LN status were reported on a “per level” basis (11) and were indicated by surgeons during operation. Thyroid cancer staging was determined in accordance with the American Joint Committee on Cancer (AJCC) recommendations (12).

Statistical analysis was performed using SPSS version 21.0 software (SPSS Inc., Chicago, IL, USA) and SAS version 9.4 software (SAS Institute, Cary, NC, USA). The statistically significant differences were defined as those with P values of <0.05. Continuous variables are presented as the mean±standard deviation, and categorical variables are presented as the number of cases and percentages. Preoperative US and CT results were compared with the postoperative histopathology results, and diagnostic values were analyzed based on the “per level” concept in accordance with the AJCC level system (12). McNemar's test was used to compare the statistical difference in sensitivity, specificity, and accuracy between preoperative US and CT. Bennett method was used to compare the statistical difference in positive predictive value and negative predictive value between preoperative US and CT.

Of the 1,323 N1b PTC patients, 386 (29.2%) were male and 937 (70.8%) were female (Table 1). The mean age of the patients was 44.2 years, and 603 (45.6%) were older than 45 years. Right-sided lateral neck dissection was performed in 573 (43.3%) patients, left-sided lateral neck dissection was performed in 526 (39.8%) patients, and both lateral neck dissection was performed in 224 (16.9%) patients. As BRAF mutation analysis was not performed until 2008 in our institution, BRAF status was available for only 576 patients. The BRAF positivity rate among these patients was 74.0%. The mean tumor size was 1.7 cm and 423 (32.0%) patients had tumors smaller than 1.0 cm. Multiplicity, bilaterality, microscopic extrathyroidal extension, gross extrathyroidal extension, chronic lymphocytic thyroiditis, unilateral central LN metastasis, and bilateral central LN metastasis were seen in 509 (38.5%), 461 (34.8%), 859 (64.9%), 228 (17.3%), 350 (26.5%), 712 (53.8%), and 438 (33.1%) patients, respectively. Skip metastasis, defined as LLNM without central LN metastasis, was found in 173 patients (13.1%).

Of the 1,323 N1b PTC patients, a total of 5,356 levels were dissected as follows: 30 in level I, 1,424 in level II, 1,492 in level III, 1,495 in level IV, and 915 in level V (Table 2). The overall incidences of level I, II, III, IV, and V metastasis were 46.7%, 56.0%, 75.5%, 77.6%, and 17.2%, respectively.

Diagnostic values were calculated on a “per level” basis by comparing the results of US, CT, and histopathology (Table 3). Between US and CT, there was no significant difference in the sensitivity (59.3% vs. 61.0%, P=0.073), specificity (83.0% vs. 81.1%, P=0.051), or accuracy (68.6% vs. 68.9%, P=0.660). However, US + CT showed significantly higher sensitivity (74.6% vs. 59.3%, P<0.001) and accuracy (74.0% vs. 68.6%, P<0.001) compared to US alone.