The Impact of Metastatic Lymph Nodes on Risk Stratification in Differentiated Thyroid Cancer: Have We Reached a Higher Level of Understanding?

Clinically evident lymph nodes

The 2015 ATA guidelines estimate the risk of recurrence for patients with a clinically evident lymph node at approximately 20% (2). The review of the literature by Randolph et al. identified the range for risk of recurrence to be 10–42% for clinically N1 disease (3). Clinically apparent or evident lymph nodes are identified on imaging or palpation during the initial workup or intraoperatively. While this characterization clearly influences risk of recurrence, it is important to understand the limitations and variability of the term “clinically evident.”

High-resolution ultrasonography continues to evolve in its ability to detect suspicious nodal metastases. However, it remains operator dependent. Reproducibility of ultrasonographic risk characterization of a lymph node relates not only to the size and the ultrasound features, but also to the skill of the ultrasonographer and comprehensiveness of the examination—factors that are intangible and difficult to standardize.

Preoperative palpation of the neck of a patient yields variable results, depending on the body habitus, tactile perceptiveness, and experience of the examiner. Intraoperative determination of what constitutes a clinically evident lymph node is similarly subjective (4). The experience of the surgeon is also likely to be a determining factor in distinguishing clinically evident from non-clinically evident adenopathy (5). The location of the lymph node can play a role in the surgeon's ability to detect it on palpation. A small node that is located near the trachea, and felt by the surgeon to be firm against the trachea, may be more readily identified than a larger node in the lateral aspect of the central neck or a positive node located adjacent to the esophagus.

Hence, while “clinical evidence” is identified in the ATA guidelines as a variable to stratify patients between the low and the intermediate recurrence risk groups, this parameter is prone to significant inconsistency.

Number of lymph nodes

The number of lymph nodes influences the risk profile of the patient, with five or fewer micrometastatic nodes assigned to the low risk of recurrence group, and more than five positive nodes <3 cm designated to the intermediate risk category (2). Reliance on the number of positive nodes should account for the variability in pathology reporting of the surgical specimen. Numerous factors and variables related to both the surgeon and the pathologist can impact pathologic reporting. First, the surgical completeness of a lymph node dissection in clearing a given compartment during surgery can impact the reporting of the number of positive lymph nodes. Second, several possible errors can occur during the pathologic assessment of lymph nodes. Harvesting nodes from a specimen is often performed by a resident or pathology assistant and undoubtedly introduces an error of underreporting. A pathologist may fail to identify small metastatic deposits in a lymph node because of the way a node is sectioned prior to staining or because he or she may overlook micrometastatic deposits that are entirely composed of a single metastatic thyroid follicle.

ENE

Randolph et al. identified that ENE conferred a range of risk of between 15% and 32% (3). A review of the literature of studies that evaluated the prognostic significance of ENE was performed. A Pubmed search of “differentiated thyroid carcinoma” and “extranodal extension/invasion” was undertaken. Studies were included if they specified treatment, length of follow-up, and number of patients, and if they evaluated the prognostic significance of ENE. Table 1 shows the expanding list of clinical studies that have reported the impact of ENE on disease recurrence, disease-specific survival (DSS), and disease-free survival (DFS). The review by Wu et al. in 2015 revealed a significant overall impact of ENE that was magnified in patients >45 years of age (6). Patients with ENE had a significantly lower DSS and DFS. In the entire cohort, 10-year DSS was 99% in the patients without ENE, and this markedly diminished to 73% in patients with ENE. In patients >45 years of age, the impact of ENE was even more dramatic, with a 100% DSS when ENE was absent compared with 63% for patients with ENE (p < 0.0001). DFS in this older cohort was even more discrepant, with a 74% DFS in patients who were ENE negative compared with 11% for ENE-positive patients. Lango et al. reported that patients with ENE had higher post-treatment levels of thyroglobulin, as well as a greater risk of both persistence of nodal disease and progression of systemic disease. The presence of ENE was associated with a 20% risk of nodal persistence (7).

Both Yamashita et al. and Ito et al. reported a worse DFS and cause-specific survival for patients with ENE (8 –10). It should be noted that there is an inherent challenge in comparing U.S. studies with Japanese studies with respect to the significant discrepancy in the use of radioactive iodine (RAI), which is not routinely administered in Japan and is much more commonly used in the United States. In addition, the studies by Ito et al. use different criteria from those of other authors. ENE was determined on the basis of intraoperative findings and not on the basis of histology. Ito et al. classified lymph nodes into LN ex0 when there was no clinical evidence of extension of disease to adjacent organs, LN ex1 when disease extended to other organs but the node(s) could be surgically separated from the other organs, and LN ex2 when there was massive extension of disease to other organs and excision required removal of the involved organ (11). It is imperative that a careful review of the literature takes into account such important differences in classification of ENE in order to draw conclusions regarding the findings in those studies and especially to compare them with other studies that use different criteria for ENE.

In the study by Ricarte-Filho et al., the presence of both BRAF positivity and ENE had an additive effect on conferring a worse DSS (12). The authors speculated that the BRAF^V600E mutation alone was not sufficient to worsen DSS, but that BRAF^V600E in combination with the oncogenic event leading to ENE may lead to a higher metastatic potential. In addition, they hypothesized that the biology of cancers that manifest ENE was associated with more disseminated disease that cannot be adequately treated with RAI due to the RAI resistance of BRAF-positive cancers (12).

These publications are far from uniform in the population of patients with differentiated thyroid cancer that were studied, as well as the treatments that were utilized. It is also important to recognize at the outset that different criteria are used for reporting ENE. In some reports, it is defined as cancer cells extending outside the lymph node capsule (Table 1, A). In others, the criteria of intraoperative identification of lymph node invasion of surrounding structures was used, a finding that likely represents a different cohort of patients altogether (Table 1, B). Still others defined ENE as tumor cells extending into an adjacent organ requiring its excision (Table 1, C), while others did not report the criteria for classifying a node as having ENE (Table 1, D).

The definition and the interpretation of ENE by histopathologists are not necessarily consistent. This has been evaluated in a concordance study of 11 senior pathologists, recognized for their expertise in thyroid pathology (13). Following an evaluation of 61 permanent section slides of lymph nodes with metastatic papillary thyroid cancer, there was a kappa coefficient of 0.35, indicating that there was only a fair level of agreement. The proportion of agreement was 0.68 in the criteria for reporting the pathologic designation of ENE (13). The risk of inter-observer variability for determining ENE was noted by Ricarte-Filho et al., who mentioned that there was a lack of stringent criteria for defining ENE (12). An important initiative of the College of American Pathologists would be to standardize reporting of ENE.

While overwhelming evidence points to the significant impact of ENE in conferring a negative impact on DFS and DSS, there are many unanswered questions. Lango et al. determined that ENE was unlikely with microscopic nodal disease and could only be identified in macroscopic tumor deposits (7). Similarly, Yamashita et al. noted a significant size discrepancy of lymph nodes with ENE (13.9 ± 7.1 mm) compared with lymph nodes without ENE that were on average 3.3 ± 3.2 mm (8). However, a study by Alpert et al. noted that smaller lymph nodes can indeed manifest ENE (14). In that study, 81 lymph nodes with ENE were evaluated for size, and 47% were ≤10 mm, with 20.5% of that cohort measuring between 0 and 5 mm. The range in lymph node size was 1.6–41 mm. That study also highlighted the fact that the metastatic focus in a lymph node does not have to consume the entire node before breaking out of the capsule into the extranodal environment (14). Some nodes manifested ENE with only 50% of the cross-sectional area of the node effaced by tumor.

The impact of nodal size and ENE has not been studied. In addition, the degree of nodal involvement by the metastatic focus has also not been evaluated. While one might intuitively contend that macroscopic, clinically evident nodes with ENE are more likely to confer a greater risk of recurrence or more serious adverse events, the alternative, in fact, may be true. Smaller nodal metastases that manifest ENE may indeed have a greater degree of biologic aggressiveness than larger nodes. One could also surmise that the metastatic focus of disease in a node that extends to the extranodal environment, before it has grown to consume the entire node, may indicate a more virulent phenotype.

The literature on the impact of ENE on prognosis in differentiated thyroid cancer is by no means perfect. However, the overwhelming majority of the series (17/19 reviewed here) demonstrate a negative prognosis when ENE is identified, providing support for the fact that it should be incorporated into staging and the risk of recurrence profile.

The 2015 ATA guidelines do not incorporate ENE in the three-tiered risk stratification system, but include it as a factor in the spectrum of risk of recurrence. Patients with fewer than three nodes manifesting ENE are placed at the low-risk end of the spectrum (2%), while those with more than three nodes with ENE are assigned a higher risk of recurrence (40%) (3).

It appears that this decision was based on the work by Leboulleux et al. (15). This study, published in 2005, determined that the risk of recurrence is impacted by the presence of more than three nodes with ENE (15). If fewer than three nodes with ENE are identified, then the risk of recurrence is only 1–4%, but if more than three nodes are identified, then the risk escalates to 32%. However, it is important to evaluate the design of the study by Leboulleux et al. in order to determine its applicability to the management of patients in the United States. Over a 10-year period, 148 patients with pathologically positive nodes or extrathyroidal extension were treated at the Institut Gustave Roussy. All patients underwent a total thyroidectomy, central compartment lymph node dissection, and an ipsilateral lateral compartment node dissection. If there was bilateral thyroid cancer, then a contralateral lateral compartment node dissection was performed as well. All patients underwent postoperative remnant ablation. Detailed pathologic assessment of lymph nodes was performed with 2 mm sections on all nodes that did not have grossly evident involvement. The mean number of metastatic nodes per patient in this population was nine, with a range of 1–66 nodes. The criteria by which ENE was judged were not defined in the publication. However, an extraordinary number of nodes with ENE were reported in this series. The mean number of metastatic nodes with ENE per patient was eight, and the range per patient was 1–57. Forty-four percent of the study population was reported to have nodes with ENE. The presence of ENE was an independent risk factor for persistent disease, with the number of such nodes impacting on that risk. Only 12% of patients with no nodes manifesting ENE had persistent disease, while patients with one to three ENE nodes had an 18% risk. Forty-nine percent of patients with more than three ENE nodes had persistent disease. Interestingly, central compartment nodal disease on initial pathology, and not lateral compartment disease, was found to be a risk factor for persistent disease. However, there were only eight patients (7%) who were found to have recurrent disease after a mean follow-up of 4.7 years (15).

The challenges in using the study by Leboulleux et al. to extrapolate usable information for risk stratification are evident. The authors reported an extraordinary number of patients with ENE, and the average number of nodes with ENE per patient is very high. All patients were subjected to more lymph node surgery than is routinely performed in the United States, and even then, the number of patients with persistent disease is very high. In addition, the pathologic sectioning of lymph nodes in this study is far more comprehensive than is usually performed in the United States. Finally, the risk of recurrent disease, identified in only eight patients in this study, is much lower than the rate of 30% that is recognized in most series. In addition, the patients with persistent disease on post-ablation whole-body scan were excluded from analysis of recurrent disease, even in locations outside of those recognized on the initial post-ablation scan. The peculiarities of this study with respect to extent of surgery, the lack of details relating to pathologic analysis and criteria for designating ENE, the very high rate of nodes with ENE, and the rates of persistent and recurrent disease make it difficult to extrapolate relevant information for assessment of risk in other patient populations.