Preoperative High-Resolution Ultrasound for the Assessment of Malignant Central Compartment Lymph Nodes in Papillary Thyroid Cancer

Introduction

Papillary thyroid cancer (PTC), which accounts for approximately 85% of differentiated thyroid cancer (1), has a propensity for early regional lymph node metastasis. Cervical lymph nodes are involved in 20–60% of PTC cases (2 –6), with malignant nodes more frequently located in the central compartment (7 –9). Although lymph node metastases are more common with increased tumor size and extrathyroidal extension (1), they may be present with small, intrathyroidal tumors (10). Malignant regional lymph nodes are the most common independent risk factor for disease persistence and recurrence (11). Involved cervical lymph nodes should therefore be identified and removed at the initial operation.

Ultrasound plays a prominent role in the evaluation and management of regional lymph nodes in patients with PTC. The 2009 American Thyroid Association (ATA) guidelines recommend the preoperative assessment of cervical lymph nodes by ultrasound in all patients undergoing thyroidectomy for biopsy-proven thyroid malignancy (1). While the accuracy of preoperative ultrasound for the assessment of lateral compartment node metastases is well recognized, its role in the identification of central compartment node metastases in patients with PTC is less established. This study delineates the utility of high-resolution ultrasound (HUS) for the assessment of malignant central compartment lymph nodes (MCLN) in patients with PTC.

Methods

The Rhode Island Hospital Institutional Review Board approved this study. A retrospective chart review was undertaken of all consecutive patients who underwent total thyroidectomy for biopsy-proven PTC by a single endocrine surgeon in an academic tertiary care center between 2004 and 2014. Two hundred and twenty-seven patients met the inclusion criteria of having a biopsy-proven PTC diagnosed with fine-needle aspiration and a subsequent preoperative HUS, including evaluation of the central and lateral compartments as defined by the American Joint Committee on Cancer (AJCC).

Preoperative ultrasound was performed using GE Logiq machines using high-frequency linear 9–14 MHz transducers, with all studies conducted by a specialized group of radiologists at Rhode Island Hospital. Prior to February 2010, GE Logiq 9 machines were used, and from February 2010, GE Logiq E9 ultrasounds were employed. A lymph node was deemed sonographically abnormal in the presence of at least one of the following findings: size >1 cm, loss of fatty hilum, rounded shape, cystic changes, microcalcifications, and peripheral vascularity.

All 227 patients underwent a careful central compartment node exploration. All lymph nodes deemed to be suspicious by the surgeon were removed, with a frozen section performed on at least one representative lymph node. A central compartment node dissection was undertaken if the frozen section was positive.

Pathologists at Rhode Island Hospital performed a gross examination and microscopic evaluation of all specimens. Information regarding the number, location, and size of nodes was ascertained. Patients were staged according to the AJCC system.

Results

The demographic characteristics of the 227 patients with FNA-proven PTC are summarized in Table 1. There were 56 (24.7%) male and 171 (77.5%) female patients included, with a median age of 51.0 years. Two hundred and twelve (93.4%) patients were white, 12 (5.3%) were black, and three (1.3%) were of unknown race/ethnicity.

All 227 patients underwent preoperative HUS evaluation of their thyroid and central and lateral compartments. Thyroid tumor size ranged from 0.4 to 5.6 cm, with a median tumor size of 1.8 cm by sonography (Table 2). The tumor was located in the right lobe in 87 (38.3%) cases, in the left lobe in 50 (22.0%) cases, in the isthmus in 7 (3.1%) cases, and was multicentric in 83 (36.6%) cases.

HUS identified central compartment lymph node abnormalities in 51 (22.5%) patients, with a median abnormal central compartment lymph node size of 1.0 cm (range 0.5–4.0 cm) by sonography (Table 2). Abnormal central compartment lymph nodes were right-sided in 14 (27.4%) cases, left-sided in 27 (52.9%) cases, midline in two (3.9%) cases, and bilateral in eight (15.7%) cases.

One hundred and four (45.8%) patients had MCLN identified by surgery. Ninety-seven (93.3%) of these patients had central compartment lymph nodes that the surgeon found abnormal (Table 3). The remaining 7 (6.7%) patients with MCLN had no suspicious central compartment lymph nodes identified by HUS or the surgeon. These microscopic perithyroidal lymph nodes were identified on final pathology.

Seventy-nine (34.8%) patients had a positive intraoperative frozen section of a central compartment lymph node. One patient had a negative intraoperative frozen section of a single central compartment lymph node, but had such suspicious central compartment findings that a central compartment node dissection was undertaken, with the final pathology confirming MCLN. Of these 80 patients who underwent a central compartment node dissection, 46 (57.5%) had a negative preoperative central compartment HUS.

Twenty-four (10.6%) patients who did not have a central compartment node dissection had MCLN. Seventeen had abnormal central compartment lymph nodes that were identified by the surgeon. Intraoperative frozen section of at least one representative lymph node was, however, negative, and therefore a central compartment node dissection was not performed. Seven patients had no abnormal lymph nodes identified on central compartment node exploration, but had at least one malignant microscopic, perithyroidal lymph node (range 1–3 lymph nodes) identified on final pathology. Nineteen (79.1%) of these 24 patients had a negative preoperative central compartment HUS.

Thirty-nine of the 104 patients with MCLN were identified preoperatively by HUS, giving a sensitivity of 0.38 ([confidence interval (CI) 0.28–0.48]; Table 4). Preoperative HUS had a specificity of 0.90 [CI 0.84–0.95], with 111/123 MCLN negative patients having had a negative central compartment HUS. The predictive value of a positive preoperative central compartment HUS was 0.76 [CI 0.63–0.87]. With 65 false negatives and 111 true negatives, the negative predictive value of central compartment HUS was 0.63 [CI 0.55–0.70]. With 150 total true positives and true negatives, central compartment HUS had an accuracy of 0.66 [CI 0.60–0.70] for the identification of MCLN.

The pathologic characteristics of the cohort are summarized in Table 5. Median tumor size was 1.6 cm on pathology, with tumors ranging from 0.2 to 9.0 cm in size. One hundred and twelve (49.3%) patients had a T1 lesion, 50 (22%) a T2 lesion, 60 (26.4%) a T3 lesion, and five (2.2%) a T4 lesion. One hundred and four (45.8%) patients had MCLN, with an average number of MCLN of 2.1 (median 0). The 80 patients who underwent a unilateral central compartment neck dissection had on average 5.6 malignant lymph nodes (median 5, range 1–22 lymph nodes). In the 24 patients who did not have a central compartment neck dissection, the average number of malignant lymph nodes was 1.4 (median 1, range 1–3 lymph nodes). The median central compartment lymph node size was 1.0 cm (range 0.1–3.0 cm) on pathology. MCLN had a median size of 0.9 cm (range 0.6–2.0 cm).

Discussion

PTC has a marked propensity for early regional lymph node metastasis. Malignant cervical lymph nodes are present in 20–60% of patients with PTC (2 –6), and may occur even when the primary tumor is small and noninvasive (10). Central compartment lymph nodes are more commonly involved than those in the lateral compartment (7 –9).

Malignant lymph nodes represent the most common independent risk factor for disease persistence and recurrence (11). Central compartment node dissection, which carries an acceptable morbidity in the hands of an experienced surgeon (12 –14), reduces the risk of nodal recurrence and may improve disease-free survival (15,16). Patients with clinically involved central compartment lymph nodes, in accordance with 2009 ATA guidelines, should undergo a central compartment node dissection. The identification of pathologic central compartment lymph nodes is thus a pivotal branch point in the treatment algorithm.

Ultrasound plays a prominent role in the evaluation and management of cervical lymph nodes in patients with PTC. Preoperative ultrasound recognizes central compartment lymph node involvement in 20–30% of cases (17,18). HUS identified central compartment lymph node abnormalities in 22.5% patients in the present study. A recent study similarly found sonographically abnormal central compartment lymph nodes in 23.9% patients with PTC (19).

The preoperative identification of central compartment lymph nodes has a measurable impact on surgical management. HUS has the ability to detect pathologic nodes as small as 2–3 mm, metastatic lesions that may otherwise go undetected on intraoperative assessment (20). A positive central compartment HUS therefore warrants a scrupulous exploration of the central compartment in search of metastatic nodes. A positive preoperative ultrasound has been shown to alter the extent of operation in 23% of PTC cases (21).

The present study found that central compartment HUS has a low true positive rate but a high specificity for the assessment of MCLN. Another study comparably reported that preoperative ultrasound had 30% sensitivity and 87% specificity for the detection of central compartment lymph node metastases (22). A sensitivity of 23% and specificity of 97% is described by a more recent study (23).

HUS is less sensitive but more specific for the assessment of lymph node abnormalities in the central compartment than the lateral compartment. Preoperative ultrasound has a reported sensitivity of 70–94% and specificity of 80–84% for the detection of lateral compartment lymph node metastases (22,23). Overlying thyroid tissue likely leads to a higher number of false-negative central compartment ultrasounds, explaining why HUS is less sensitive for the assessment of MCLN.

In the present study, 57.5% of patients who underwent a central compartment node dissection for intraoperative biopsy-proven lymph node involvement had a negative preoperative central compartment HUS. It was the surgeon's identification of suspicious central compartment lymph nodes that prompted biopsy and subsequent intervention in these cases. A negative central compartment HUS therefore does not preclude the need for a careful central compartment node exploration.

There was a small group of patients who had central compartment node metastases that was not preoperatively or intraoperatively identified. Seven patients had a no abnormal central compartment lymph nodes noted by HUS or the surgeon, but had malignant lymph nodes identified by the pathologist. These were microscopic, involved perithyroidal lymph nodes. The significance of these subclinical lymph node metastases on recurrence and survival has been questioned by several retrospective studies (13).

Conclusion

Preoperative HUS is highly specific for the identification of MCLN in patients with PTC. The present findings emphasize, however, that a negative HUS does not obviate the need for careful exploration of the central compartment to determine the necessity for central compartment node dissection. Therapeutic central compartment node dissection, whether prompted by preoperative HUS or intraoperative suspicion, reduces the risk of nodal recurrence.