Thyroid cancer is the most common endocrine malignancy worldwide and the 9^th most common malignancy in Europe and Northern America (1). Incidence varies across the world, with the highest reported in South Korea (111.3/100 000), where thyroid cancer is the most common cancer in women, due to excessive screening. In most countries, the incidence of thyroid cancer has been clearly on the rise in the last decades. In the United States, the incidence tripled between 1975 and 2009, and the steepest, about 10-fold increase caused by overdiagnosis was observed in South Korea (2). This increase is attributed almost exclusively to papillary thyroid cancer (PTC), which accounts for from 85% to over 95% of thyroid cancer diagnoses (3, 4). In the vast majority of cases, however, the prognosis remains excellent and the mortality is low.

Surgery constitutes the most commonly used curative treatment for PTC. Depending on the preoperative staging, lobectomy or total thyroidectomy is recommended (5). Lymph node metastases in PTC are very common and present in the majority of patients at the moment of diagnosis (6). In the case of preoperatively or intraoperatively discovered lymph node metastases, the recommendation concerning the lymph nodes is clear: therapeutic compartment-oriented dissection (5). Prophylactic central lymph node dissection (CLND), however, is still debated, mainly due to the fact that it increases the frequency of postoperative hypoparathyroidism.

Fluorescence imaging, a relatively new tool in endocrine surgery, facilitates the identification and preservation of parathyroid glands and can contribute to lower frequency of postoperative hypoparathyroidism. It can therefore be of great use to minimise the risks associated with CLND, while allowing to take advantage of all its potential benefits.

The aim of this review is to synthesise the current scientific literature on the use of fluorescence imaging in thyroidectomy with lymph node dissection, and to examine the arguments for and against prophylactic CLND in the light of reported outcomes of this imaging technique.

The frequency of lymph node metastases in PTC ranges from 44.5% to more than 60% (6–8) and tends to decrease with age (7). Also microscopic lymph node metastases in papillary thyroid microcarcinoma are common: they were present in 40% to 64% of patients who underwent central neck dissection (8). Metastatic lateral lymph nodes are found during prophylactic neck dissection in up to 44% of patients with papillary thyroid microcarcinoma (9). Despite the extremely frequent nodal involvement, occurring even at early disease stages, the prognosis is excellent, with 10-year disease-specific survival reaching 98.6% (10).

The impact of lymph node metastasis on patient outcome has been extensively studied, including in large epidemiological studies. Most studies report that the presence of metastatic lymph nodes is associated with lower survival and/or higher disease recurrence (11–16), while others report similar outcomes in patients with and without lymph node metastases (17). However, data from large epidemiological studies seem to outweigh the doubts, suggesting that metastases are associated with worse patient outcomes (11–13, 15). Besides, most studies do not differentiate between preoperatively diagnosed, pathologically-diagnosed macroscopic and microscopic lymph node metastases.

Therapeutic lymph node dissection is systematically performed for clinically apparent metastases (cN1) identified either preoperatively by imaging or palpation, or intraoperatively by the surgeon. Dissection of clinically uninvolved lymph nodes (cN0) is called prophylactic neck dissection. Compartment-oriented therapeutic neck dissection for the central and lateral compartments of the neck is currently not subject to debate, and is recommended by most guidelines and societies, including the 2015 ATA guidelines (5). Prophylactic lateral neck dissection is not recommended, because the lateral neck constitutes a separate compartment, which remains untouched during the initial thyroidectomy, so it can be dissected at a later stage without any significant additional risk. On the other hand, prophylactic CLND has been the subject of extensive scientific discussion which has so far led to no clear consensus. According to the ATA 2015 guidelines, prophylactic CLND should be considered for patients with advanced primary tumours (T3 and T4), or if the information obtained will be used to plan further therapy (5).

The rationale for prophylactic CLND is the high rate of occult nodal metastases in PTC (up to 40%-60%) (6–8) and the lack of a reliable method to detect them preoperatively or intraoperatively. Preoperative ultrasonography detects only up to 20% of involved lymph nodes, and its sensitivity is limited, particularly in the central compartment, due to the overlying thyroid gland. Ito et al. calculated that preoperative ultrasonography detects central lymph node metastases with 99% specificity, but only just over 10% sensitivity (18). Prophylactic CLND provides precise staging information, which can guide the decision to administer radioactive iodine postoperatively. The size, number and ratio of metastatic lymph nodes are important factors in assessing the risk of recurrence (19). This information about affected lymph nodes in the central compartment can be accurately obtained only through CLND. Bonnet et al. found that lymph node status affected the decision on postoperative radioactive iodine treatment in 30.5% of patients with PTC <2 cm (20). The value of more precise risk estimation lies not only in identifying patients who are at higher risk of recurrence and would benefit from radioactive iodine treatment, but also in reliably selecting those for whom radioactive iodine therapy can be omitted, avoiding overtreatment. In the study by Bonnet et al., radioactive iodine treatment could have been avoided in 42% of patients when the pN0 status and other low-risk features were confirmed. At the time of the study, however, radioactive iodine was given to all patients with lymph node metastases who had intermediate- or high-risk cancers as defined in the 2009 ATA and 2006 European recommendations (21, 22). Conversely, in low risk thyroid cancer patients, an ipsilateral prophylactic CLND upstages 9% of the patients to intermediate risk (23). However, in a recent randomized controlled clinical trial by Leboulleux et al., follow-up without radioactive iodine therapy was non-inferior to follow-up with radioactive iodine ablation in terms of recurrence outcomes in patients with low-risk differentiated thyroid cancer and Nx/N0 status (24).

More precise diagnosis of the lymph node status provides accurate indications not only for radioactive iodine treatment, but also for adequate initial staging and follow-up. The initial pN0 status in otherwise low-risk tumours might obviate the need for repetitive exams and associated high healthcare costs, while, the pN1 status might lead to a more careful follow-up.

Many retrospective studies have investigated the impact of CLND on patient outcomes (i.e. recurrence and survival), but the results are conflicting. Contrary to what was previously claimed (25), prospective trials on the effects of CNLD are feasible. An ongoing randomized clinical trial (ESTIMABL 3) studies prophylactic CLND for thyroid cancer with the frequency of excellent response after total thyroidectomy with or without CNLD and surgical morbidity as major endpoints (NCT03570021). After analysing thirty retrospective studies and five meta-analyses, the European Society of Endocrine Surgeons issued a consensus report stating a lower risk of recurrence after prophylactic CLND and recommending that CLND should be performed by surgeons with sufficient experience (26). A recent meta-analysis of 17 studies and 4437 patients who underwent thyroidectomy with and without prophylactic CLND for PTC found a significantly higher risk of recurrence in the latter group (27). Another meta-analysis of 13 studies and 3331 patients found that those treated with thyroidectomy and prophylactic CLND for PTC had the risk of recurrence reduced by 35% (28). A number of groups showed significantly lower recurrence in the central compartment of the neck after CLND (26, 29–31), while others observed no impact of CLND on local recurrence (32–36). Concerning survival, most studies consistently report that the impact of prophylactic CLND on disease-specific survival is limited or negligible (29, 37, 38).

A commonly mentioned argument against CLND is the higher morbidity burden, particularly of hypoparathyroidism. In a meta-analysis of 13 studies, Lang et al. showed that hypocalcaemia was the main postsurgical morbidity in patients undergoing thyroidectomy with prophylactic CLND for PTC, and that when it was excluded, the overall morbidity was similar to patients treated with thyroidectomy alone (28). A number of studies, including the consensus report of the European Society of Endocrine Surgeons, reported no difference in frequency of permanent hypoparathyroidism in patients who underwent thyroidectomy with and without CLND (26, 29, 31, 39, 40).

Other studies found a significantly higher rate of permanent postoperative hypoparathyroidism in patients undergoing thyroidectomy with CLND (35, 38, 41). A recent study using Scandinavian registry data analysed over 700 patients who underwent thyroidectomy with or without CLND for pT1-T3 PTC and found that permanent hypoparathyroidism was significantly higher in the CLND group (6.6 vs 2.3%) (41).

Permanent postoperative hypoparathyroidism has lifelong consequences: higher risk of cardiac and renal morbidities, renal stones, and mental health problems, as well as frequent paraesthesia, basal ganglia calcifications and impaired cognitive function (42, 43). Recent studies showed that permanent postoperative hypoparathyroidism can even be associated with an increased risk of renal insufficiency, any type of malignancy, cardiovascular events and higher mortality rates (44, 45). Moreover, the management of postoperative hypoparathyroidism increases considerably the financial burden on healthcare systems (46).

Fluorescence occurs when a substance excited with light of a particular wavelength emits light of a longer wavelength and lower energy. Fluorescence imaging takes advantage of this property by illuminating tissues with a filtered light of a specific wavelength. The target tissue absorbs the light and then emits light of longer wavelength, which is captured and processed. The substance responsible for fluorescence is called “fluorophore” and can be extrinsic, like contrast agents (indocyanine green (ICG) is the most used contrast agent in surgery), or intrinsic, like melanin or porphyrin, which are occurring in tissues naturally.

Paras et al. first described the clinical application of near-infrared autofluorescence (NIRAF) in the detection of parathyroid glands (47). When excited with infrared light of 785 nm wavelength, the parathyroid glands emit light of 820 nm wavelength. These findings were subsequently validated in larger patient cohorts, in benign and malignant thyroid disease, as well as in diverse parathyroid pathologies, all consistently showing stronger autofluorescence of healthy and hyperfunctioning parathyroid glands compared to all the other neck tissues, including the thyroid (48–50). Of note, the fluorophore responsible for NIRAF in the parathyroid glands is currently unknown.

A number of studies have shown that mapping of the parathyroid glands position with NIRAF imaging at various stages of thyroidectomy leads to earlier and more frequent identification of the glands, lowers the rate of inadvertent resection, and allows to salvage more glands from the resected specimen (51–53).

Whether intraoperative detection of NIRAF lowers postoperative hypoparathyroidism and hypocalcemia has also been a subject of numerous studies. Most studies, including several randomized controlled trials, consistently show lower frequencies of postoperative transient hypoparathyroidism (54–57). However, some recent randomized controlled trials found no significant difference in postoperative calcium and parathormone concentrations in NIRAF and control groups (58, 59). No reduction in the frequency of permanent hypoparathyroidism has been observed so far (54–56).

The autofluorescence signal of the parathyroid glands remains unaffected by blood perfusion: in order to assess gland vascularity and viability, other methods, in particular ICG angiography, are necessary.

Unlike autofluorescence imaging, contrast-enhanced fluorescence imaging does not exploit the intrinsic properties of the tissue. The technique requires administration of an extrinsic contrast agent – a fluorescent dye. ICG is a fluorescent dye widely used in intraoperative imaging since the 1950s. It has an excellent safety profile, with very few reported adverse reactions (< 1 per 200 000). ICG quickly binds to plasma proteins and remains almost entirely confined to the vascular system. It is taken up by the hepatocytes and excreted into the bile. ICG has a very fast washout and metabolism, so repeated administrations during a single intervention are possible.

ICG angiography can be used at the beginning of the dissection, to identify and preserve well-vascularized parathyroid glands, or after the dissection, to predict postoperative outcomes. Used at the beginning of the surgery, ICG angiography can aid identification of the parathyroid glands’ vessels, and therefore help preserve the parathyroid glands and the vessels during thyroid dissection. Other studies have shown that intraoperative ICG angiography can be used to accurately evaluate parathyroid vascularization and thus gland function after thyroid resection (60–62). Vidal Fortuny et al. conducted a randomized clinical trial which demonstrated that in the case of at least one well-perfused parathyroid glad on ICG angiography, no patient had hypoparathyroidism after total thyroidectomy and that there was no need for blood testing and calcium/vitamin D supplementation (61). This technique can also be helpful in deciding whether a parathyroid gland should be left in situ or autotransplanted, depending on its viability.

Reports on the use of fluorescence imaging in CLND are scarce. To date, only three studies, all performed in Asia, evaluated the use of NIRAF and/or ICG angiography in thyroidectomy with CLND for thyroid cancer (63–65). All three groups included patients undergoing thyroidectomy with unilateral or bilateral, prophylactic and therapeutic CLND. Table 1 summarizes basic information about the three studies.

Ouyang et al. evaluated the impact of ICG angiography on the identification and preservation of the parathyroid glands, as well as on the postoperative parathyroid hormone and calcium concentrations in patients undergoing bilateral axillo-breast approach (BABA) robotic thyroidectomy with unilateral and bilateral CLND (both therapeutic and prophylactic) for PTC (63). ICG angiography was used twice, to identify the parathyroid glands and their blood vessels to guide dissection, and to assess parathyroid perfusion after the removal of the specimen ( Table 2 ).

Kim et al. investigated the impact of NIRAF imaging on postoperative calcium and parathyroid hormone concentrations in a large cohort of patients undergoing thyroidectomy and CLND (unilateral and bilateral) for thyroid cancer (65). The researchers proposed the concept of “parathyroid gland mapping”, that is localisation of the glands through NIRAF imaging at early stages of the dissection, when no glands are yet visible to the naked eye. If the parathyroid glands were not visible at that stage, NIRAF imaging was used a second time, after limited dissection was performed, and the glands previously covered by tissue were exposed. NIRAF imaging was also performed on the specimen, to search for inadvertently resected parathyroid glands ( Table 2 ).

Yin et al. conducted a randomized controlled trial of NIRAF imaging combined with ICG angiography in CLND (64). NIRAF imaging was used in the initial stages of the dissection to identify the parathyroid glands and again, after the removal of the specimen. ICG angiography was used at the end of the surgery, to evaluate blood perfusion of the glands and select those qualifying for autotransplantation ( Table 2 ).

Ouang et al. showed that the mean number of identified and in situ preserved parathyroid glands was higher in the ICG group (3.74 vs 3.15, p < 0.001; and 3.12 vs 2.74, p = 0.007, respectively). Patient who underwent ICG angiography were also more likely to have all four parathyroid glands identified and preserved. Yin et al. identified and preserved in situ more parathyroid glands in the NIRAF than in the control group (3.6 vs 3.2, p < 0.001; and 1.3 vs 1.0, p < 0.001, respectively). The number of inadvertently resected and salvaged glands were not reported.

Kim et al. found that the rate of inadvertently resected parathyroid glands in the specimen was 6.9% in the NIRAF group and 12.8% in the control group (p = 0.021), but observed no differences in the number of identified parathyroid glands.

The reported impact of the introduction of fluorescence imaging protocols on the frequency of postoperative hypoparathyroidism is summarised in Table 3 .

The results suggest that fluorescence imaging may reduce temporary hypoparathyroidism in patients undergoing thyroidectomy with CLND. This could happen through several mechanisms: better identification of the parathyroid glands and their vascular pedicles, lower rates of inadvertent parathyroid resection, and facilitated decision-making on parathyroid glands autotransplantation.

Two of the reviewed studies showed that fluorescence imaging helped lower the frequency of hypoparathyroidism. Ouyang et al. observed no statistical difference, possibly due to the small sample size (63). A relatively high initial frequency of temporary hyperparathyroidism (46.6% in the control group) reported by Kim et al. (65) decreased significantly to 33.7% in the NIRAF group, which is similar to the median incidence of transient hypoparathyroidism after conventional thyroidectomy reported in a recent meta-analysis (up to 38%) (66). Ouyang et al. and Yin et al. observed no cases of permanent hypoparathyroidism (63, 64). In the study by Kim et al., the rate of permanent hypoparathyroidism was 4.2% in the NIRAF group, and was not significantly different from the control group’s 4.6% (65). However, certain significant differences between the two groups were identified: larger tumour size, and higher frequency of lymph node metastases in the central and lateral compartments in the NIRAF group. The reported frequency of permanent postoperative hypoparathyroidism was slightly higher than the median incidence reported for conventional thyroidectomy (0% to 3%) (66). This could be related to the fact that the researchers in this study only used NIRAF imaging, which does not provide any information about the perfusion and viability of the glands.

All three cited studies were performed at high-volume expert centres, where permanent postoperative hypoparathyroidism is already relatively rare, so a larger sample size is necessary to capture statistically significant differences between the fluorescence imaging and control groups. Certain unaccounted for patient and tumour characteristics, such as age, BMI, size of the tumour and size and extent of nodal metastases, could have contributed to the fact that no difference in the frequency of permanent hypocalcemia was observed in the cited studies. More studies are needed to assess, if application of NIRAF imaging could reduce the rate of permanent hypocalcemia rate in T3-T4 tumours as compared to T1-T2. This would facilitate the decision on whether to proceed to a prophylactic CLND based on the tumour stage.

The use of fluorescence imaging in thyroidectomy with CLND for thyroid cancer facilitates the early identification of parathyroid glands, helps to identify more parathyroid glands per intervention, lowers the number of inadvertently resected glands and lowers the frequency of transient hypoparathyroidism. Further studies are needed to evaluate the technique’s impact on permanent hypoparathyroidism. A combination of different fluorescence imaging techniques (NIRAF to map parathyroid glands; ICG angiography to locate the vascular pedicle intraoperatively, and to evaluate gland perfusion postoperatively) might yield even better results. Fluorescence imaging can be useful to minimise the risk of hypoparathyroidism associated with CLND, while allowing to take advantage of all its potential benefits. With further development, fluorescence imaging techniques might increase the safety of thyroid cancer surgery with neck dissection and thus shift the paradigm to more frequently recommended prophylactic CLND.

PK: Conceptualization (lead); writing – original draft (lead); formal analysis (lead); writing – review and editing (equal). MD, SL: writing – review and editing (equal). FT: Methodology (lead); writing – review and editing (equal). All authors contributed to the article and approved the submitted version.