Over the last several decades, the incidence of differentiated thyroid cancer (DTC) has been increasing worldwide. In Korea, DTC affects approximately 43.3 per 100,000 individuals in 2014 [1
]. Metastasis to the lymph nodes (LNM) is common in DTC, particularly in a subtype of the disease called papillary thyroid carcinoma (PTC) which makes up approximately more than 90% of thyroid cancer [2
]. In a study analyzing US data from the Surveillance, Epidemiology, and End Results (SEER) cancer registry, metastasis to nodes in the central and lateral neck has been reported to occur in about 57% and 34% of DTC cases [3
]. Therefore, central neck lymph node dissection and lateral lymph node neck dissection are often performed depending on the extent of DTC [4
]. Local metastasis of DTC to the lymph nodes and the potential for further distant metastasis increases the risk of recurrence of disease and is associated with a poorer prognosis. Therefore, thorough removal of metastatic lymph nodes is essential for effective treatment of DTC [5
The conventional method for detecting metastasis of DTC to the lymph nodes is high-resolution ultrasonography. However, approximately one-third of patients with PTC may have LNM that is missed by preoperative ultrasonography [7
]. Furthermore, the accuracy of ultrasonography is highly dependent on radiologist experience [8
] which varies considerably among hospitals. False negative results of ultrasonography provide a misleading prognosis of DTC which may lead to conservative treatment that increases the chance of recurrence and disease-specific mortality. We propose that other indicators that can predict the burden and extent of DTC are needed.
In medullary thyroid cancer (MTC), preoperative serum calcitonin levels have been shown to predict the burden and extent of tumor and support decision-making regarding the extent of surgery required. Even when ultrasonography detects no LNM, high levels of serum calcitonin can predict the presence of LNM in MTC [9
Our own research recently showed that preoperative serum thyroglobulin levels can predict distant metastasis of DTC [10
]. We therefore hypothesized that preoperative serum thyroglobulin levels may also be related to tumor burden (primary tumor size, number of LNM) and extent of LNM (including distant metastasis). If so, preoperative thyroglobulin levels may be a useful clinical biomarker to help to determine the optimal surgical extent for DTC resection. Here, we retrospectively reviewed primary tumor size, number and location of LNM, and distant metastases of 4029 DTC cases to determine whether preoperative serum thyroglobulin levels are associated with burden and extent of DTC.
In this study, we examined the association between preoperative serum thyroglobulin levels, tumor burden, and extent of DTC. As tumor burden (i.e., primary tumor size and the number of LNM) and tumor extent (i.e., presence of lateral LNM and distant metastasis) increased, we found that preoperative serum thyroglobulin increased. Specifically, a preoperative thyroglobulin level above 13.15 ng/mL could predict the presence of an ipsilateral lateral LNM, above 30.05 ng/mL could predict the presence of contralateral lateral LNM, and above 62.9 ng/mL could predict the presence of distant metastasis.
Our findings are significant, because the incidence of thyroid cancer has increased substantially worldwide [11
], a trend that is attributed to the overdiagnosis of small PTC. As a result, the current treatment guideline of DTC has become more conservative [14
], and lobectomy has been performed more frequently in the treatment of low-risk DTC [4
]. However, even if a DTC is completely resected, it may recur later, resulting in a poorer prognosis and requiring additional operations that can adversely impact patient health-related quality of life [17
]. There is a significant unmet need to be able to more accurately predict, in the preoperative stage, the full tumor extent to help guide surgical planning and balance the need to reduce recurrence and ensure optimal recovery from treatment. Our results suggest that preoperative serum thyroglobulin levels may provide additional information to ultrasonography in the preoperative setting rather than definite diagnosis of metastasis. For example, in a patient who has high preoperative serum thyroglobulin levels during the preoperative laboratory work-up for a lobectomy for low-risk DTC, surgeons should order cross-sectional imaging and consider a total thyroidectomy with or without additional neck dissection. Furthermore, if the preoperative thyroglobulin level is unequivocally high, clinicians should consider advanced whole-body imaging studies in pursuit of distant metastases.
Of note, even in a small-sized primary DTC of less than 1 cm diameter, LNM was frequently detected [18
]. In our study, among patients with a primary tumor diameter less than 1 cm, 992 (40.6%) had LNM and the smallest primary tumor diameter in patients with LNM was 0.1 cm. In addition, 147 (32%) patients with lateral LNM had a primary tumor of less than 1 cm and the smallest primary tumor diameter that also presented with lateral LNM was 0.2 cm. If these patients were treated solely with lobectomy based on their small tumor size, there would be a high likelihood of recurrence after an incomplete initial surgery. Therefore, primary tumor size alone does not reflect tumor extent and cannot accurately guide surgical treatment planning. In such cases, preoperative thyroglobulin could better inform clinical decision making, providing information on the true tumor extent.
The inclusion of a thyroglobulin blood test at the preoperative stage may be particularly relevant in low-resource settings where there is a paucity of skilled radiologists to facilitate accurate preoperative clinical staging.
Although preoperative thyroglobulin was helpful for predicting tumor burden and extent in DTC, it is not as specific as calcitonin is for predicting MTC [9
]. This difference can be mostly attributed to the fact that thyroglobulin is also produced in normal thyroid tissue before thyroidectomy, while calcitonin is mainly produced in MTC cells or parafollicular (C) cells [19
]. One recent study has shown that preoperative thyroglobulin has negative results for predicting metastasis in DTC [20
]. For the same reason, it is not possible to determine the extent of surgery based on preoperative thyroglobulin results alone. However, with the recent trend of choosing lobectomy as a preferred surgical option, preoperative thyroglobulin levels can provide ancillary information about tumor extent before surgery, help to determine the extent of surgery, and flag cases for which a lobectomy may not be sufficient.
There were some limitations to this study. There is a bias inherent to the retrospective study setting and single center design. Because patients were recruited from a single teaching hospital, they were prone to a selection bias. External validation using a multicenter prospective study is needed in the future.
4. Materials and Methods
We used the institutional thyroid cancer database and retrospectively reviewed the clinical records of patients who were diagnosed and treated with DTC at Samsung Medical Center from 1994 to 2016. This study was approved by the institutional review board (IRB) of Samsung Medical Center (IRB File No. 2017-02-056) and performed in accordance with relevant guidelines and regulations. Informed consent was waived due to the retrospective design and the anonymization of all patient data.
Pathologic data on primary tumor size (diameter; cm) and the total number of LNM from patients’ anonymized medical records were used to determine tumor burden. To evaluate tumor extent, the location of LNM (i.e., central, ipsilateral lateral or contralateral lateral) and the presence of distant metastasis were ascertained from the clinical data for each patient. Distant metastasis was defined as a suspicious metastatic lesion detected before or within 6 months after initial surgery. Distant metastasis was found by pathological confirmation or imaging study such as computed tomography (CT), magnetic resonance imaging, whole-body scan (WBS), or positron emission tomography scans. Tumor extent was classified as (1) no metastasis; (2) presence of central LNM with no lateral LNM and no distant metastasis; (3) presence of ipsilateral lateral LNM with no contralateral lateral LNM and no distant metastasis, regardless of central LNM status; (4) presence of contralateral lateral LNM with no distant metastasis, regardless of central and ipsilateral lateral LNM; and (5) presence of distant metastasis, regardless of LNM. Central LNM did not distinguish between ipsilateral and contralateral LNM. Other primary tumor histology parameters, such as lymphatic invasion, blood vessel invasion, extrathyroidal extension (ETE), and resection margins, were also reviewed from clinical data.
Patients checked serum thyroglobulin levels from blood sampling preoperatively. Serum thyroglobulin was measured using immune-radiometric assays with a BRAHMS Thyroglobulin plus RIA assay (BRAHMS, Henningsdoft, Germany). The functional sensitivity of the Thyroglobulin plus RIA assay was 0.2 ng/mL, and the analytical sensitivity was 0.008 ng/mL. The intraassay coefficient of variation (CV) of the assay was 3.4%, and the interassay CV was 4.9% [21
For statistical analysis, continuous data are expressed in the format of mean ± standard deviation. Categorical data are expressed as percent values or absolute numbers. Preoperative serum thyroglobulin is presented as the median (range) in ng/mL. Pearson’s correlation analysis was used for comparison of continuous data and is represented by the correlation coefficient, r. Multivariate linear regression was used to analyze factors that affected preoperative thyroglobulin level elevation and is represented by the regression coefficient, β. The enter method was used to adjust other variables. Receiver operative characteristic (ROC) curve analysis was used to determine the threshold level for preoperative serum thyroglobulin to distinguish tumor extent. Sensitivity and specificity were calculated using an established method. A p-value of <0.05 was considered to be significant. Statistical analysis was performed using the Statistical Package for the Social Sciences software version 23 (IBM Corp., Armonk, NY, USA).