Lymph Node Metastasis in Differentiated Thyroid Cancers

Abstract

Lymph node metastasis is common in differentiated thyroid cancers. Therapeutic neck dissection removes macroscopic nodal metastasis, reduces local recurrence, and facilitates cancer surveillance. On the other hand, microscopic nodal metastasis is also increasingly recognized as a potential cause of persistent disease or early recurrences. Prophylactic neck dissection, by removing microscopic nodal metastasis, has been proposed to reduce recurrence and prevent future reoperation. When cancer recurs, regional nodal recurrence is most common, and the management should be individualized. We hereby present a narrative review on the management of nodal metastasis in differentiated thyroid cancers.

1. Introduction

Differentiated thyroid cancers (DTC) account for over 90% of all thyroid cancers and are further divided into papillary thyroid carcinomas (PTC), follicular thyroid carcinomas (FTC), and Hürthle cell carcinomas. PTC is by far the most common subtype, accounting for 85% of all DTC [1].

1.1 Incidence of lymph node metastasis in DTC

The incidence of lymph node metastasis is different between various types of DTC. PTC is associated with higher frequency of nodal metastasis [2]. On the contrary, FTC seldom metastasizes through the lymphatics [3]. The incidence of nodal metastasis in PTC depends on its detection method and definition. Preoperative imaging can detect nodal disease in up to 30% of patients [4] whereas pathological series reveal nodal metastasis in 20−50% of operated patients [5]. With the replacement of traditional hematoxylin and eosin staining by modern immunohistochemical or molecular genetic techniques, the incidence of nodal micro-metastasis (defined as tumor foci <2 mm) has been reported in up to 90% of patients [6]. Hence nodal metastasis is very common in DTC, particularly amongst patients with PTC and is often underestimated.

1.2 Pattern of lymphatic spread

It is important to understand the pattern of lymphatic spread in DTC as it carries important implications in treatment planning. The prelaryngeal (Delphian) nodes and paratracheal nodes are the most common first sites of nodal metastasis in PTC [7]. Cancer cells then spread laterally to the jugular nodes in a contiguous stepwise fashion prior to spreading to distant sites [8]. If the nodal disease is identified in the lateral neck, both central and lateral neck dissection should be performed. However, “skip metastases”, i.e. the presence of lateral nodal disease without central nodal involvement, have been reported in 7–22% of patients with PTC [9, 10]. Multivariate analyses found that skip metastases were most commonly found in patients with unifocal tumor ≤1 cm at the upper one-third of thyroid lobe. Hence, patients without central nodal disease cannot be assumed to be free of lateral nodal metastasis. The knowledge of this pattern of spread enables accurate preoperative assessment of the nodal areas and dictates the necessary extent of neck dissection.

1.3 Classification of cervical lymph nodes

The cervical levels system has been used to accurately communicate the location of nodal disease, standardize terminology in research and guidelines, and define the boundaries of areas where lymph node dissection should be performed. Lymph nodes in the lateral neck compartment are grouped into level I–V, whereas levels VI and VII refer to lymph nodes in the central neck compartment. Several refinements and addition of sublevels were summarized in Table 1 [11, 12, 13].

Nodal level	Anatomical boundaries
IA—submental	Triangular area bounded by the anterior belly of the digastric muscles and hyoid bone
IB—submandibular	Triangular area bounded by the digastric muscle, stylohyoid muscle, and the body of the mandible
IIA/B-upper jugular	Area bounded by stylohyoid muscle anteriorly, the posterior border of the SCM posteriorly, Skull base superiorly, and the hyoid bone inferiorly. The spinal accessory nerve further divides sublevel IIA (anterior) and IIB (posterior)
III—middle jugular	Area bound by hyoid bone superiorly, inferior border of the cricoid cartilage inferiorly, medial border of the SCM anteriorly, and lateral border of the SCM posteriorly.
IV—lower jugular	Area bounded by the inferior border of the cricoid cartilage superiorly, the clavicle inferiorly, medial border of the SCM anteriorly, and lateral border of the SCM posteriorly.
VA/B—posterior triangle	Triangular area bounded by the posterior border of the SCM, trapezius muscle, and the clavicle. The inferior border of the cricoid cartilage further divides sublevel VA (superior) and VB (inferior)
VI—anterior	Area bounded by the hyoid bone superiorly, the suprasternal notch inferiorly, and the anterior border of SCM posteriorly and midline anteriorly.
VII—superior mediastinal	Extension of level VI from suprasternal notch to the level of the innominate artery.

Table 1.

The levels and sublevels of the neck.

SCM: sternocleidomastoid muscle.

1.4 Risk factors predicting nodal metastasis

Understanding the risk factors predicting nodal metastasis in DTC is imperative for identifying the group of patients most vulnerable of nodal metastasis and best indicated for neck dissection. The most consistently reported risk factor for central compartment nodal metastasis is PTC of size larger than 1 cm [14, 15]. The presence of ipsilateral central compartment nodal metastasis is also associated with greater risks of contralateral central compartment involvement. Other reported risk factors include multifocal tumors, extrathyroidal extension, and younger age [14, 15, 16, 17, 18, 19]. The only risk factor for lateral compartment nodal metastasis is the presence of central compartment nodal metastasis [19, 20, 21].

With advances in molecular techniques, researchers have explored the utility of biomarkers in addition to conventional clinicopathologic features to predict lymph node metastasis in DTC. BRAF is a proto-oncogene which has received frequent attention. Multiple studies have reported the presence of BRAF^V600E mutation as a factor associated with nodal metastasis in DTC [22, 23, 24]. Yet, other studies have refuted such associations [25, 26]. To date, only seven heterogenous prospective studies have been published but with conflicting conclusions [27, 28, 29, 30, 31, 32, 33]. Hence, the true utility of BRAF^V600E mutation in risk stratifying nodal metastasis in clinically node negative (cN0) patients remains controversial.

Another potential prognostic marker of differentiated thyroid cancer is the Telomerase Reverse Transcription (TERT) promoter gene which regulates chromosomal integrity. The somatic mutation TERT-C228T was found in up to 11% of DTC [34]. PTC with TERT-C228T was associated with more aggressive tumor behaviors and poorer clinical outcomes when compared with patients exhibiting BRAF mutation alone [35]. The presence of TERT-C228T mutation has been found to be associated with nodal metastasis with an OR of 1.5–1.8 but its prospective role in stratifying patients for neck dissection remains to be elucidated [36, 37].

1.5 Detection of lymph node metastasis in DTC

1.5.1 Incidence of nonpalpable nodal metastasis

Targeted physical examination of the neck is the important first step in detecting nodal metastasis in DTC but its sensitivity is variable. Palpable nodal disease has been reported in only around 5–10% of patients but the incidence of nonpalpable nodal metastasis is much higher. Mayo Clinic reported that ultrasonography (USG) detected nonpalpable nodal metastasis in up to 32% of their PTC cohort [4]. Particularly amongst patients with prior neck surgery, the nonpalpable nodal disease was up to 28 and 64% for the central and lateral compartment, respectively. This highlights the importance of preoperative radiological assessment for nodal metastasis.

1.5.2 Role of USG and computer tomography in the detection of nodal metastasis

USG and computed tomography (CT) are commonly used for preoperative detection of nodal metastasis in DTC. However, it is important to determine which individual imaging modality should be chosen alone or in combination. In general, USG is cheap, easily accessible, and allows diagnostic real-time fine needle aspiration cytology or biopsy. Conversely, CT is a standardized technique that is non-operator-dependent, provides greater anatomical details, and allows accurate evaluation of the retropharyngeal, retrosternal, and mediastinal areas not accessible by USG [38]. In addition, CT can be useful in evaluating extra-thyroidal tumor extension, better detecting multi-level nodal involvement or presence of extranodal extension, and better assessing the anatomical relationship with adjacent critical structures. Hence, both USG and CT are complementary modalities for the investigation of nodal metastasis.

In a retrospective study, USG had an overall sensitivity of 51% and specificity of 92% in preoperative detection of nodal metastasis [39]. The performance of CT was statistically similar at a sensitivity of 62% and a specificity of 93%. Combined USG/CT only improved sensitivity in patients with lateral compartment nodal disease and in patients with nodal involvement of more than one level. These findings were replicated and summarized in a meta-analysis of 1691 patients [40].

Based on the above evidence, the American Thyroid Association (ATA) recommends USG as the first line modality in all DTC while additional CT is considered in larger cancers for patients with higher chance of nodal metastasis and extrathyroidal spread [41]. Scenarios, where CT would be particularly useful, include patients presenting with pressure symptoms arising from thyroid mass, hoarseness of voice, clinically fixative thyroid mass (cT4), and retrosternal thyroid extension incompletely assessed by USG, and patients with palpable bulky lymph nodes.

1.5.3 Role of positron emission tomography in the evaluation of nodal metastasis

Recently there is an increasing interest in the role of positron emission tomography (PET) with 18-fluoro-2-deoxy-d-glucose (¹⁸FDG) for evaluating nodal metastasis in DTC. In the last two decades, ¹⁸FDG-PET has emerged as a growing method for detecting DTC recurrences, particularly in non-iodine avid diseases such as Hürthle cell carcinomas [42, 43]. In a multicentre study, the sensitivity of ¹⁸FDG-PET to detect recurrent disease was shown to be greater than ¹³¹I whole body scan (WBS) [44]. However, PET provides low spatial resolution and suboptimal anatomical details. Co-registering PET with CT overcomes this limitation such that ¹⁸FDG -PET/CT has become an adjunctive tool in the detection of recurrent DTC, particularly in patients with elevated serum thyroglobulin and negative WBS.

The potential role of PET/CT in the detection of nodal metastasis prior to initial surgery has growing research attention. Jeong et al. evaluated the utility of preoperative PET/CT in their retrospective cohort and reported that PET/CT had a diagnostic accuracy of 92.3% with a sensitivity of 30% and a specificity of 96% in detecting nodal metastasis which was comparable to USG and CT [45]. However, PET/CT did not provide any superior diagnostic accuracy when indirectly compared with USG and CT in a network meta-analysis [46]. Further high-quality direct comparative studies are required to further elucidate the role of PET/CT particularly in the preoperative assessment for nodal metastasis. Current evidence only supports the role of PET/CT in the postoperative detection of recurrent DTC.

Table 2 summarizes the diagnostic performance of various radiological modalities in the preoperative detection of cervical nodal metastasis across all cervical levels.

Level I–VI	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	Accuracy (%)
USG	51	92	81	76	77
CECT	62	93	84	80	81
USG + CECT	66	88	77	81	79
^18FDG-PETCT	30	96	61	89	87

Table 2.

Imaging modalities in preoperative detection of cervical nodal metastasis.

PPV: positive predictive value; NPV: negative predictive value; and CECT: contrast enhanced CT.

1.6 Prognostic implication of nodal metastasis

It is important to understand how nodal metastasis in DTC affect prognosis in order to justify its preoperative detection and neck dissection. Although survival of DTC is generally very good, certain specific populations may suffer from greater chances of recurrence and mortality.

1.6.1 Lymph node metastasis and cancer-specific survival

Regional nodal metastasis had been implicated as a prognostic factor for survival in earlier studies involving large retrospective cohorts [47, 48, 49]. For example, Podnos et al. demonstrated a significantly lower overall survival at 14 years (79% vs. 82%) for node-positive patients in their cohort of 9904 patients. However, these studies have been challenged by newer reports that nodal metastasis only adversely affected cancer-specific survival in a subset of patients >45 years of age but not in those <45 years [50].

1.6.2 Lymph node metastasis and disease recurrence

Regional nodal metastasis also had been implicated as a predictor of disease recurrence. Macroscopic lymph node involvement, i.e. clinically palpable or radiologically detectible metastatic lymph node, is associated with a high rate of local recurrence (10–42 percent) [51]. Furthermore, patients with >5 positive lymph nodes, higher lymph node ratio, and the presence of extranodal extension were associated with even higher risks of local disease recurrence [51, 52]. However, the association of microscopic (radiologically undetectable and nonpalpable) nodal disease and recurrence was not well demonstrated [53, 54, 55, 56]. A randomized controlled trial of clinically node negative (cN0) patients showed that microscopic nodal disease detected by prophylactic neck dissection did not affect disease-free survival [57].

1.6.3 Lymph node metastasis in staging systems

Based on how nodal metastasis affects survival and recurrence, the presence of nodal metastasis upstages patients over the age of 45 years from American Joint Committee on Cancers (AJCC 6th and 7th edition) from stage I to stage III disease [58]. Similarly, the proposed modifications of the American Thyroid Association risk stratification system classified <5 microscopic nodal metastasis as a component of low-risk disease, macroscopic nodal metastasis or >5 microscopic nodal metastasis as intermediate-risk disease, and any nodal metastasis > = 3 cm in greatest dimension as high-risk disease [59]. These staging systems influence the degree of postoperative thyroid-stimulating hormone suppression therapy and determine whether radioiodine ablation (RAI) should be given. Hence, accurate nodal status has an important clinical significance.

2. Management of lymph node metastasis in initial surgery for DTC

2.1 Classification and types of neck dissection

The management of nodal metastasis must be discussed with clear definitions of various types of lymph node involvement:

1. Clinically node positive/negative (cN+/0) denotes patients with/without clinically or radiologically detected preoperative lymph node involvement

2. Pathological node positive/negative (pN+/0) denotes patients with/without histoanatomical findings of malignant cell foci (>2 mm in macrometastasis, 200 μm–2 mm in micrometastasis, <2 μm in isolated tumor cells)

3. Occult nodal metastasis (cN0-pN+) denotes patients with nonpalpable radiologically non-detectable nodal involvement. This is synonymously referred to as microscopic metastasis in literature.

In cN+ patients, lymphadenectomy is referred to as therapeutic neck dissection. Whereas in cN0 patients, lymphadenectomy is referred to as prophylactic neck dissection. The distinction between prophylactic and therapeutic neck dissection cannot be overemphasized as the impact of clinically detectible macroscopic nodal involvement is different from microscopic metastasis in cN0-pN+ disease.

Neck dissection is also classified by the lymphadenectomy extent (Table 3) [13]. Selective neck dissection is increasingly performed as the growing knowledge of the pattern of spread of various cancers enables surgeons to spare the lymph node levels not considered at risk of nodal metastasis. Selective neck dissection has been traditionally further classified into supraomohyoid neck dissection (level I–III), central neck dissection (level VI), lateral neck dissection (level II–IV), and posterolateral neck dissection (level II–V). However, these selective neck dissection terminologies were confusing and obsolete with variable inclusion or exclusion of certain levels and sublevels. Hence, selective neck dissection is better described by specifying the levels and sublevels included [13].

Radical neck dissection	The basic standard procedure for cervical lymphadenectomy. All other types of neck dissection are alterations of this procedure. Level I–V lymph nodes are removed together with the IJV, SAN, and the SCM.
Modified radical neck dissection	Preservation of one or more non-lymphatic structure including the IJV, SAN, SCM, or the combination.
Selective neck dissection	Preservation of all of IJV, SAN, and SCM and one or more lymph node levels based on the pattern of spread of the malignancy.

Table 3.

Classification of neck dissections.

IJV: internal jugular vein; SAN: spinal accessory nerve; and SCM: sternocleidomastoid muscle.

2.2 Therapeutic neck dissection

Therapeutic neck dissection is performed when patients have preoperatively confirmed nodal metastasis (cN+). It can be performed during initial thyroidectomy or as a staged secondary procedure after initial thyroidectomy.

2.2.1 When nodal metastasis was detected in the central compartment preoperatively

For DTC with preoperatively confirmed central compartment nodal metastasis, therapeutic central neck dissection should be performed bilaterally. This should extend superiorly from the hyoid bone to the innominate artery inferiorly. It is laterally bound by bilateral carotid arteries, anteriorly bound by the superficial layer of the deep cervical fascia, and posteriorly bound by the deep layer of the deep cervical fascia [12]. A compartment-oriented lymphadenectomy with the removal of all fibroadipose tissues within the compartment while identifying and preserving the ipsilateral recurrent laryngeal nerve (RLN), superior, and inferior parathyroid glands should be performed. In the past, non-anatomical nodal dissection (“berry picking”) was performed but studies have shown a significantly greater recurrences rate (100% vs. 9%) with non-anatomical dissection versus compartment-oriented dissection, and similar surgical morbidities [60].

The most common morbidities of central neck dissection are transient hypocalcaemia (4–60%) and transient RLN injury (0–5%) but permanent hypocalcaemia and permanent RLN injury can occur in up to 15 and 12% of patients, respectively [61]. Although isolated studies demonstrated no additional complication risk from extra central neck dissection, most series had reported higher rates of complication when central neck dissection was performed concurrently with thyroidectomy and the outcomes were associated with surgeon’s experience [62, 63, 64, 65].

2.2.2 When nodal metastasis was detected in the lateral compartment preoperatively

For DTC with preoperatively confirmed lateral compartment nodal disease, the chance of concomitant central compartment disease is high. Therefore, patients should undergo both therapeutic lateral and concomitant central neck dissection.

The extent of therapeutic lateral neck dissection is not well agreed upon. While the consensus statement from the ATA recommends the removal of levels IIA, III, IV, and VB in a comprehensive therapeutic neck dissection, other authors routinely recommend a full dissection of levels II−V [11]. In DTC, the rate of nodal metastasis at the various lateral neck levels differs widely. The incidence of nodal metastasis at level III (62–67%) and level IV (50–67%) were significantly higher than that at level II (42–56%) and level V (29–40%) [21, 66, 67, 68]. A further distinction of sublevels IIA/B and VA/B were reported. Farrag et al. stated that level IIB rarely had nodal metastasis (8.5%) while all level V metastases were within level VB [68]. Others reported that level IIB nodal metastasis was exclusively accompanied by level IIA nodal metastasis [68, 69]. To date, no randomized controlled trials were published to determine the most appropriate operative extent. But available evidence supports a selective approach where level IIB and VA are only dissected if there is preoperative or intraoperative suspicion of level II or level V involvement. There is an added advantage that avoiding routine level IIB and VA dissection can minimize the risk of spinal accessory nerve injury.

Only a few studies have dealt with the surgical morbidities of lateral neck dissection. Two key complications after lateral neck dissection are lymphatic leakage secondary to thoracic duct damage (0.5–8%), and spinal accessory nerve injury (25−50%) resulting in shoulder dysfunction [27]. Other less common nerve injuries can be related to greater auricular nerve (48%), cervical plexus, sympathetic trunk (5%), and phrenic, hypoglossal, and vagal nerves [28, 63].

2.3 Prophylactic central neck dissection: why and who?

The rationale for prophylactic neck dissection is based on the fact that occult nodal metastasis is common. Despite comprehensive preoperative imaging, there is clinically node-negative (cN0) patients who are found to have unexpected nodal metastasis (pN+) on pathology. The incidence of this occult nodal disease was reported in up to 54% of patients who underwent elective bilateral central neck dissection during total thyroidectomy for DTC [14]. About 50% of which were a bilateral occult nodal disease. Prophylactic neck dissection allows accurate disease staging, improves postoperative risk stratification, and improves serum thyroglobulin levels facilitating postoperative surveillance. Prophylactic neck dissection at the initial thyroid surgery may additionally help to avoid reoperation in the future.

The evidence supporting routine prophylactic central neck dissection is controversial. While some earlier studies reported that prophylactic central neck dissection could reduce the risk of nodal recurrence and cancer-specific survival [29], others did not show such benefit. Aggregating these heterogenous nonrandomized studies, a meta-analysis showed that prophylactic central neck dissection had a lower risk of locoregional recurrence (risk ratio 0.66) than those without neck dissection [30]. However, prophylactic central neck dissection was associated with higher rates of overall morbidity, especially transient hypoparathyroidism. To date, three randomized trials have been published on prophylactic central neck dissection and all failed to show improvement in oncological outcomes or recurrence-free survival [31, 32, 57]. While it was shown that the prevalence of operative morbidities was similar in the group with prophylactic central neck dissection, all these randomized trials were underpowered to demonstrate the difference in survival outcomes and morbidity rates. To this end, ATA examined the feasibility of a multi-institutional prospective randomized controlled trial and concluded that the sample size required would be prohibitively large (>5800) given the low rate of disease recurrence and operative morbidities [33].

Both the ATA guidelines and the American Association of Endocrine Surgeons guidelines recommend a selective approach based on assessment of a patient’s risk factors [59, 70]. For patients with T1/2 tumor, the risks of central nodal metastasis are relatively small, and thus prophylactic central neck is not recommended. Patients with T3/4 tumor, or extrathyroidal extension, or BRAF mutation may be considered at higher risk, and option of prophylactic central neck dissection should be considered. The British Thyroid Association on the other hand recommended personalized decision-making for prophylactic central neck dissection on the basis of one or more high-risk factors (adverse histological subtype, age ≥ 45 years, multifocal, tumors >4 cm, extra-thyroidal extension). This highlights the variability and uncertainty in this aspect of management of DTC.

2.4 Prophylactic lateral neck dissection

Prophylactic lateral neck dissection is generally not recommended. The rationale is that lateral neck dissection can be deferred until the nodal disease become clinically apparent and proven. The concern to reoperate after initial thyroidectomy is lower as the lateral compartment is not entered during the initial surgery. Evidence suggested that removal of microscopic nodal disease in the lateral department had not been proven to improve cancer-specific or disease-free survival [51]. Considering the morbidity associated with lateral neck dissection, lateral neck dissection is in general reserved for proven nodal disease only.

2.5 Sentinel lymph node biopsy in differentiated thyroid cancer

The concept of sentinel lymph node biopsy (SLNB) depends on the stepwise pattern of lymphatic spread and the belief that the sentinel node (SLN), the first node that drains the tumor, will reflect the status of the remaining lymph nodes within the drainage basin.

2.5.1 Various sentinel lymph node techniques

Various SLNB techniques have been studied in PTC. The use of vital dye was the earliest described technique, but it is growing out of popularity due to its disadvantages that (1) lateral compartment SLN cannot be visualized during thyroidectomy; (2) parathyroid glands may have uptake; and (3) the risks of anaphylaxis [71]. The use of radioisotope as an alternative technique allows identification of SLN located outside the central compartment without parathyroid gland uptake. Furthermore, the addition of SPECT/CT imaging can improve preoperative anatomical localization of the SLN. In a meta-analysis including 45 retrospective studies, the SLN detection rates for vital dye technique, radiolabelled lymphoscintigraphy, vital dye plus lymphoscintigraphy, and lymphoscintigraphy plus SPECT/CT were 83, 96, 87, and 93% respectively, while the reported respective false negative rates were 38, 40, 17, and 8% [72].

2.5.2 Using sentinel lymph node biopsy to guide neck dissection

SLNB has been applied in the management of PTC where SLNB-positive patients can undergo dissection of the involved compartments, and negative patients can be spared the risks and added costs of unnecessary procedure. The negative predictive value of each SLNB technique is the most important parameter to determine whether neck dissection should be performed, and it is dependent on the SLN detection rate and the false negative rate of each technique. Relevant literature data was diversified, firstly due to the difficulty to interpret SLN at frozen section; secondly due to variable definition of the SLN resulting in non-sampling of non-dominant nodes; thirdly due to the occasional skip metastasis; and finally due to the learning curve [73, 74]. Therefore, despite promising initial results from SLNB, further high-quality prospective evidence is necessary. Furthermore, the cost-effectiveness of SLNB strategy also remains ambiguous. A retrospective study reported that the cost of implementing SLNB outweighed the potential cost saving from avoided procedures and morbidities [75]. As such, the current use of SLNB remains within clinical trial settings.

3. Management of recurrent nodal disease

DTC recurs in up to 30% of patients. Three quarters of these recurrences are at the cervical and mediastinal lymph nodes, 20% are in the thyroid remnant and 21% are distant with the lungs being the most common [76]. Although recurrences are common, resultant cancer related deaths are not [77].

3.1 Management of resectable nodal recurrence

When patients develop isolated nodal recurrence after previous surgery, further curative surgery, if technically feasible, confers the advantages of avoiding future local complications arising from the recurrent tumor, improving serum thyroglobulin level, and facilitating RAI treatments.

3.1.1 Low volume recurrences

Although surgical removal of recurrence confers advantages, low volume recurrence does not necessarily require surgery. Several observational studies have suggested that low-volume recurrent nodal disease can be indolent and can be managed with active surveillance. In a retrospective cohort of 166 patients with suspicious lateral compartment lymph node of a median size of 1.3 cm, less than 10% of the patients had interval progression by >5 mm, and 14% had complete resolution [78]. Thyroid stimulating hormone suppression therapy should be continued during active surveillance.

3.1.2 Macroscopic resectable recurrence

Larger volume recurrent nodal disease has been associated with poorer cancer-specific survival and is best treated with revision surgery [79]. In the cases where the recurrence occurs in a previously un-operated neck region, a formal neck dissection should be considered. Revision surgery achieves biochemical remission rates of 21−66%. While most modern series suggest that revision surgery can achieve a high clearance rate of structural disease in over 80% of patients [80].

However, in the cases where recurrence occurs in a previously operated area, the risk of re-operative surgery should be balanced against oncological disease clearance. Re-operative surgery for recurrent nodal disease has higher risk of major complications due to the increased technical demand for dissecting scarred tissue and altered anatomy. The morbidity from re-operative surgery is related to the region undergoing dissection, the experience of surgeons, and the degree of scarring from initial surgery. In general, the incidence of permanent and transient hypoparathyroidism was 0–9.5%, and 0–46.3% respectively. While the rate of transient and permanent unexpected RLN palsy was an average of 3.6 and 1.2%, respectively [81].

3.1.3 Decision-making for re-operative neck dissection

The ATA recommends that radiologically localized, cytology confirmed, recurrent central neck nodes ≥8 mm and lateral neck nodes ≥10 mm in the smallest dimension should be considered for surgery. The exact size threshold for re-operative neck dissection is anecdotal by consensus, and available supporting evidence are scarce. The relationship between the size of recurrent structural disease, morbidity from surgery, and response to therapy had been assessed by Lang et al. in their cohort of 130 patients [82]. Lesion >15 mm was an independent risk factor for incomplete biochemical response. The rates of incomplete surgical resection, unexpected vocal cord palsy, and overall morbidity were also significantly higher in patients with lesions >15 mm than those with lesions <15 mm. Hence, the authors propose that the threshold for continued active surveillance can be less stringent and extended to larger lesions.

Besides the size of recurrence, several other factors are also important to be considered during discussion for revision surgery. These include patient’s factors (symptoms arising from recurrent disease, vocal cord status, previous neck irradiation/surgery, motivation for surgery), surgeon’s experience, and disease factors (lesion’s location in relation to vital structures, factors reflecting tumor aggressiveness such as serum thyroglobulin doubling time, speed of radiological growth, iodine avidity, and the presence of adverse molecular markers) [81]. Hence, such a decision is best discussed with a dedicated multidisciplinary team including surgeons, endocrinologists, nuclear medicine physicians, pathologists, and oncologists [83].

3.2 Nonoperative local treatment for nodal recurrence

Radioactive iodine ablation (RAI) may be employed in patients with ¹³¹I-avid low-volume disease detected on WBS or as adjuvant treatment following surgery. To date, no randomized controlled trials had demonstrated superior outcomes with RAI alone or as adjuvant treatment in the setting of locoregional recurrent disease [59]. However, ¹³¹I non-avid lesions are unlikely to respond and empirical RAI is not recommended [84].

With a larger volume of nodal recurrences, percutaneous ethanol injection (PEI) for metastatic cervical lymph node was first reported in the early 1990s−2000s [85]. Studies have reported sonographic successful ablation in up to 84% of treated lymph nodes after repeated sessions of treatment [86]. Most ablated lymph nodes decreased in size and 46% completely disappeared [87]. Radiofrequency ablation (RFA) for metastatic cervical lymph node is another newer modality of nonoperative treatment for recurrent nodal disease. It has been associated with a greater mean volume reduction of 55–95% and complete resolution of the structural lesion in 40–60% [88]. The limitation of these ablative modalities is that only limited small studies are available. They are considered a nonoperative form of lymph node picking and, as such, are best considered in patients with high surgical risks or those who declined surgery.

3.3 Management of extensive/unresectable nodal recurrence

Advanced nodal recurrence may involve extensive soft-tissues with potential laryngeal, tracheal, esophageal, or carotid sheath invasion. While there is no strict description to define unresectable diseases, extensive organ resections such as hemi- or total laryngectomies, circumferential tracheal resection, and esophagectomies have been described [89]. The related morbidity and the anticipated functional impairment must be accepted by the patient. Even in patients with distant metastases and concomitant loco-regional recurrent disease compromising the aerodigestive tract, palliative re-operative surgery may be considered on a case-by-case basis, followed by adjuvant RAI treatment [90].

In extensive unresectable nodal recurrences, RAI alone is unlikely able to eradicate the disease as the absorbed radiation dose is generally inadequate. External beam radiotherapy (EBRT) can be considered for locoregional control in such patients. However, acute treatment-related morbidities, including dermatitis, mucositis, and dysphagia, are not uncommon and efficacy has only been reported in retrospective cohorts [91]. To date, there are no randomized trials that address specific indications for EBRT in patients with recurrent DTC. Hence, individual practice is variable.

3.3.1 Other systemic treatment for extensive/unresectable recurrent diseases

Patients with gross symptomatic recurrent disease that cannot be alleviated with surgery or EBRT or had become RAI refractory, are candidates for systemic therapy. Inhibition of protein kinases that function in key signaling pathways can regulate tumor proliferation, angiogenesis, metastasis, and apoptosis. Furthermore, inhibition of kinases involved in the mitogen-activated protein kinase pathway has been shown to re-express genes of iodine metabolism and thus allow restoration of RAI uptake in RAI refractory DTC [92]. Somatic mutation testing can be performed to identify oncogenic targets such as gene rearrangements in NTRK, RET, or BRAF. This may guide the use of mutation-specific kinase inhibitor such as TRK inhibitors (e.g. Larotrectinib, Entrectinib) and RET inhibitors (e.g. Selpercatinib, Pralsetinib). However, BRAF inhibitors (e.g. Vemurafenib, Dabrafenib) are currently nonapproved in thyroid cancers and have been used off-labeled.

Given the high cost of performing mutation analysis and the limited participation in clinical trials, anti-angiogenic multi-kinase inhibitors (aaMKI) such as Lenvatinib or Sorafenib are alternative agents. No head-to-head comparisons of aaMKI are available. Most of these aaMKI target vascular endothelial growth factor receptors (VEGFR) in angiogenic pathways. Of note, patients with risks of bleeding e.g. recent major surgery, haemorrhagic brain metastasis, or recent thromboembolic events are relatively contraindicated for MKI. Furthermore, tracheoesophageal fistulation have been reported in patients with prior EBRT and MKI treatment.

If a patient is intolerant to, or the disease is refractory to kinase inhibitors, conventional cytotoxic agents (e.g. Doxorubicin) may be alternative options but experience is sparse.

4. Conclusions

Nodal metastasis is common in differentiated thyroid cancers and its pattern of spread is well recognized. Despite improvements in technology, imaging has limited sensitivity to detect nodal metastasis before initial surgery or when disease recurs. Occult nodal metastasis is very common and may explain persistent disease or early recurrence. However, distinction between macroscopic and microscopic nodal metastasis and their prognostic implications must be made clear.

Many unanswered questions regarding management of nodal metastasis remain. Prophylactic central neck dissection may remove occult nodal metastasis but its impact on survival lacks high-quality evidence support and is best reserved for selected patients by experts. Sentinel nodal biopsy is an attractive concept but further evidence from research is required. Even when nodal metastasis is detected before the initial surgery, the optimal extent of therapeutic neck dissection remains debated. When disease ultimately recurs, the decision to operate is complex, taking into consideration the increased rate of morbidities in re-operative surgeries. Active surveillance may be best for low-volume recurrent disease while other nonoperative treatments can be considered in patients not suitable for surgery.

Conflict of interest

The authors declare no conflict of interest.