Isosulfan blue dye is used most widely. The dye is injected submucosally around the tumor. The SLNs are stained blue 15–45 min after the injection. The technique needs visualization, and hence, it is essential to expose the entire nodal basin, thereby increasing the invasiveness of the procedure. Moreover, blue dye consists of small particles with a very poor retention in the sentinel lymph node, and the blue color is therefore retained for a short period of time. This is probably because of the fast lymphatic drainage in the head and neck area. Besides, isosulfan blue dye has lower reliability than lymphoscintigraphy. Staining of the mucosa around the tumor may make the surgical excision of the primary tumor difficult. So, this technique is not preferred now for identification of the SLN.

Preoperative dynamic lymphoscintigraphy involves the injection of radiolabeled colloid at the periphery of the tumor. The flow of radiolabeled dye from the primary tumor to the sentinel nodes can be visualized in real time using a gamma camera operating in a continuous mode in both the anteroposterior and lateral views (Fig. 8.2). The position of these nodes where the radioactivity localizes can be marked on the skin. Intraoperative static lymphoscintigraphy involves identifying the nodes with highest radioactivity using a handheld gamma probe. An incision is made at the region of the nodes marked by the preoperative dynamic lymphoscintigraphy (Figs. 8.3 and 8.4). It is important that the incision should be planned in such way that it may be extended to do a neck dissection, should it be required, based on the SLN biopsy result. There is no threshold radioactivity value for the SLN. It varies by the time of injection, quantity of radioisotope used, and the location of the lesion. The nodes with the peak radioactive reading as well as any adjacent nodes that are more than 10 % as hot as the SLN are also removed. After removal, the node has to be checked for radioactivity. The surgical bed should not have radioactivity higher than the background reading. The 10 % rule was developed on the basis of the observation that about 13 % of the metastatic nodes are not those that are the hottest nodes [28]. There is controversy about the number of SLNs to be biopsied for accurately determining the pathological nodal status [29]. This question was addressed by Werner et al. [30]. They performed SNB in 90 patients with clinically N0 head and neck cancer. Up to three SLNs were biopsied in these patients. Overall, 23 of the 90 patients (25.6 %) showed evidence of occult metastasis. It was observed that if only the node with the strongest tracer uptake had been biopsied, the histological evidence of metastasis would have been missed in nine of 23 (39 %) patients. This study clearly suggested that the node with highest radioactivity may not be the pathologically positive SLN, and more than one SLN needs to be harvested for accurate pathological evaluation [31, 32]. Detection of the SLN at level I becomes difficult at times due to the “shine-through” effect of the primary tumor. This is especially true in case of a primary tumor in the floor of the mouth. In this scenario, it is recommended that the primary tumor be removed first before localizing the SLN [33–35].

Lymphoscintigraphy and intraoperative gamma probe identification of the SLN depend upon the ability of the injected radiotracer to be selectively retained within the sentinel node, while minimizing retention at the primary site and transit to downstream lymphatics. Once injected interstitially, radiotracers travel in lymphatic channels to the first echelon nodes, where they are taken up by macrophages. In order to be phagocytosed, these particulates must be within certain size limits. The size of the particle also influences clearance rates from the primary tumor site, transit time through the lymphatics, and retention time within the sentinel node. Larger particles achieve higher retention within the sentinel node and lower transit to second-echelon lymphatics and higher retention within the primary tumor site.

Gold-198 was the first material used for the purpose. This had a particle size of 5 nm. Although this material has greater and faster uptake than any other subsequently developed radioisotopes, the high dose of radiation thwarted its broader clinical use [36]. Iodine-131 and 99mTc were later introduced for lymphoscintigraphy. The 99m Tc attached to sulfur colloid is now the most widely used for lymphoscintigraphy. The advantages of 99mTc sulfur colloids are that they emit only gamma rays and have low radiation exposure, the half-life of 99mTc is only 6 h, and it has a peak energy emission peak of 140 keV. This is within the detection range of most of the gamma camera and handheld gamma probes. The particle size and the attached molecules are the primary factors that determine the rate of uptake into the lymphatics and the filtration within the sentinel node. The optimal particle size of radioisotopes is between 5 and 10 nm [37]. A particle size smaller than 5 nm may be taken up by the vascular system. The radioisotopes may be used as either filtered or unfiltered forms. The filtration allows control of the particle size to a specific size (15–50 nm). The unfiltered nanocolloids have a particle size raging from 5 to 1000 nm [36, 37]. The dose of radioisotope used also varies from 0.5 to 0.8 mCi. Using a 99mTc sulfur colloid in cutaneous lesions, the transit time to the lymph node is less than 1 h. The radioactivity may be retained in the lymph node for an additional 3–6 h. However, for mucosal head and neck tumors, the transit time is less than 30 min. The radioactivity can be detected for 3–6 h after the injection. Ideally, the injection, dynamic scintigraphy, and intraoperative gamma probe localization should be done on the same day.

Tilmanocept is a novel agent. It is a 99mTc-labeled non-particulate radiotracer that contains multiple mannose moieties with high affinity for the CD206 receptor found on macrophages and dendritic cells, enhancing targeting to these cells within the SLN. Studies in breast cancer and melanoma showed that tilmanocept may have improved clearance from the site of the primary tumor and enhanced retention within the sentinel node when compared to sulfur colloid [38, 39]. Because of the rapid clearance and prolonged retention within the sentinel nodes, patients could be injected preoperatively from immediately prior to surgery up to 30 h preceding surgery. A single institution reported their experience as part of this larger multicenter trial in their initial report of 20 clinically node-negative patients [40]. The NPV was 100 % for five patients with floor-of-mouth tumors. Complete results of the multicenter phase III trial are yet to be published.

Intraoperative detection of metastatic deposits in sentinel node biopsy is important to make the sentinel node biopsy procedure patient friendly and to avoid a staged second procedure. Recent studies have shown reasonable negative predictive value with frozen section analysis (83 %) [41–43]. Intraoperative evaluation with frozen section was accurate in detecting macrometastasis but was not effective in detecting micrometastasis and isolated tumor cell deposits [44]. Moreover, if smaller deposits (mainly isolated tumor cells) [41] are located within the tissue used for frozen section analysis, it would be missed in eventual analysis for micrometastasis. It has been argued that frozen section analysis prolongs the duration of the procedure. But, by coordinating with the histopathology service, the frozen section result can be made available during the resection of primary tumor.

Imprint cytology as an alternative to frozen section analysis of lymph nodes is reported [45]. In the study by Trivedi et al. [44], the detection rates of imprint cytology and frozen section were identical. As in frozen section, imprint cytology failed to detect micrometastasis and isolated tumor cells. For better utilization of SLN biopsy, it is essential to develop a novel technology that can detect smaller deposits intraoperatively in SLNs. Intraoperative ultrarapid IHC and intraoperative real-time reverse transcriptase–polymerase chain reaction (RT–PCR) evaluation [46] may aid in the future to improve the sensitivity of intraoperative detection of occult metastasis.

Routine pathologic evaluation of a neck node consists of identifying each individual node, bisecting the node at its center and then staining one or two sections to find light microscopic evidence of metastatic deposits [47]. This, in reality, is an incomplete examination, where central sections serve as a proxy for the whole node. If deposits were small and present in other regions of the node, they would be missed. Studies have shown that routine evaluation misses up to 21 % of disease nodes in breast cancer [48]. It has been shown that SSS with hematoxylin–eosin stain and IHC and molecular methods identify smaller metastasis more accurately. Nelson [49] has reported that hematoxylin–eosin staining with step sections identifies one cancer cell among 10,000 normal cells. IHC identifies one tumor cell among 100,000 normal cells. RT–PCR is the most sensitive of all. It identifies one cell among 1 million normal cells. In clinical practice, SSS with hematoxylin-eosin staining upstages the tumor in 10 %, whereas IHC further upstages it up to 10 % more [26, 27]. In a study by Trivedi et al. [44], routine pathologic evaluation detected occult metastasis in 13 cases (16.2 %) but missed metastasis in seven cases. SSS with hematoxylin–eosin stain and IHC identified the metastasis in 20 cases (25 %) and hence further upstaged the neck by about 9 %. Other studies [26, 27] also have reported similar results. SSS was necessary to detect micrometastatic deposits. The routine pathologic evaluation was not sufficient to detect this metastasis. IHC was needed only to identify isolated tumor cells. The clinical significance of these smaller foci of metastasis is not established.

There is now a standard recommendation for pathological evaluation of the SLN [50]. The node is dissected free of any fat and bisected through its long axis. If each half of the node is more than 2.5-mm wide, they are then further sectioned longitudinally so that each section is not more than 2.5 mm. SSS is then carried out at 150-mm intervals. At each step, four sections are obtained. One section will be stained by H&E and the other by IHC for cytokeratin. The remaining sections are retained for any additional or repeat study (Fig. 8.5). If any cytokeratin-positive cells are identified, they are compared with the adjacent H&E section to confirm that the positivity was due to tumor cells. It has been recognized that individual cells indigenous to lymph node milieu may be keratin positive and hence present a risk of false-positive detection in some instances [51, 52]. Once recognized, this problem can be solved by judicious application of interpretation skills. A standard and uniform reporting of the SLN needs to be adopted to compare results [50].

The term micrometastasis is erroneously used for any metastasis detected by histologic analysis of clinically negative (N0) neck. However, histologically detected metastases are correctly termed as occult metastases, which can be further stratified based on histopathologic criteria. Hermanek et al. [53] proposed histopathologic classification of occult metastasis for breast cancer into macrometastasis, micrometastasis, and isolated tumor cells. Widely used staging scheme for breast cancer formally defines macrometastasis (Fig. 8.6) as those metastatic deposits more than 2 mm in diameter. The micrometastasis is defined as metastatic deposit between 2 and 0.2 mm in diameter (Fig. 8.7). Deposits less than 0.2 mm are defined as isolated tumor cells [54] (Fig. 8.8). This can be either single cell deposits or a cluster of tumor cells. Unlike in breast cancer, there is no uniform histopathologic staging available for occult metastases in head and neck cancer. Some studies have included 3 mm as the upper limit of size of micrometastasis [55], but these studies do not always mention lower limit for micrometastasis. Most studies though use 2 mm as the upper limit and 0.2 mm as the lower limit for micrometastasis [56]. Metastatic deposits less than 0.2 mm are generally defined as isolated tumor cells [53, 54, 56, 57].

The technical feasibility of SNB is ascertained by the probability of identifying sentinel lymph nodes with preoperative lymphoscintigraphy and intraoperative detection using a handheld gamma probe. High rates of sentinel lymph node identification in oral cancer have been reported suggesting that the procedure is feasible. Pilot studies from 20 centers contributing to the Second International Conference on Sentinel Node Biopsy in Mucosal Head and Neck Cancer were collectively analyzed [50]. Among 379 patients with clinically N0 disease, sentinel lymph nodes were identified in 366 patients, for an identification rate of 97 %. A systematic review and diagnostic meta-analysis of all published literature on SNB in oral cancer through 2003 were performed by Paleri and colleagues [61]. The analysis included 301 patients with oral cavity primary tumors from 19 studies. The study reported an overall sentinel lymph node identification rate of 97.7 %. A European multi-institutional study was initiated in 2002. The 5-year results of this study [35] showed that, of the 227 SLNBs reported, 134 were performed on early-stage (T1-2) oral cavity cancer. In these patients, sentinel nodes were correctly identified in 93 % (124 of 135) of patients. The ability to identify the sentinel node was lower in patients with floor-of-mouth tumors compared to all other subsites (88 % vs. 96 %, p = 0.138). The meta-analysis by Govers et al. [62] had 847 cases of early-stage oral cavity or oropharynx in total. At least one sentinel node was detected in almost all patients included from the studies, and a sentinel node biopsy could thus be performed in 835 patients. The sentinel node detection rates ranged from 91 to 100 %.

The next step in evaluating SNB lies in determining the extent to which occult metastases are identified. The finding of occult metastasis within the sentinel node on pathologic evaluation results in upstaging of the node-negative N0 neck. The rate of upstaging by SNB must be compared with that by elective neck dissection, which is currently the best available method of detecting occult metastasis. Pathologic evaluation of specimens stained with H&E and elective neck dissection upstages approximately 30 % of patients. In the Canniesburn trial [26], where the sentinel lymph nodes were examined by a routine H&E staining, serial sectioning, and immunohistochemical analysis for cytokeratin, upstaging of disease occurred in 34 % of cases. Nodal metastasis was identified by H&E staining alone in 26 % of cases and by additional pathologic means in 11 % of cases. Therefore, although the extent to which SNB upstages the neck by traditional pathologic methods seems similar to that with elective neck dissection, the use of additional pathologic methods results in perhaps an even greater level of detection of disease. Additional pathologic methods, such as serial sectioning and immunohistochemical analysis, may increase the identification of micrometastasis. The identification of micrometastases creates a problem for the current staging system for oral cancer. Patients with micrometastatic disease found by additional pathologic methods and with no additional disease in the neck specimen are upstaged from clinically N0 to a new classification of pNmi, or pathologic nodal micrometastasis, for which the prognosis is unknown.

Many single-institution series have published the results of sentinel lymph to accurately stage the neck compared to elective neck dissection [63–76]. In the European multi-institutional study [35], with a minimum 5-year follow-up, the overall sensitivity of the procedure was 91 %. The patients with floor-of-mouth tumors demonstrated lower negative predictive values (NPV) compared to other oral cavity sites (88 % vs. 98 %). These findings led the authors to recommend the use of SLNB as a reliable staging procedure in all oral cavity sites with the exception of the floor of the mouth. The ACOSOG trial [77], a multi-institutional trial from the United States, (Z0360) included 140 patients with early-stage (T1-2) oral cavity cancer. SNB was performed, followed by immediate elective lymph node dissection. The negative predictive value for SNB in this study, which consisted primarily of oral tongue and floor-of-mouth carcinomas, was 94 % (95 % CI: 0.88–0.98) on routine histologic analysis, with improvement to 96 % with immunohistochemical staining. SNB appeared to perform slightly better for smaller lesions, with NPV of 100 % for T1 lesions compared to 94 % for T2 lesions. The similar conclusions of these multiple published experiences support SNB as an accurate procedure for staging the neck in early-stage oral tongue cancer. In the meta-analysis by Paleri et al. including 301 patients with oral cavity squamous cell carcinoma as well as a smaller subset of 49 patients with oropharynx cancer [61], the pooled sensitivity of SNB for the study group was 0.926 (95 % CI: 0.852–0.964) using a random effects model, with individual ranges of 0.75–1. A more recent meta-analysis examining the diagnostic accuracy of SNB in head and neck cancer [78] included 766 patients in 26 studies between 1970 and 2011, in which the pooled NPV was 96 % (95 % CI: 94– 99 %). In the subset of 593 patients with early-stage (T1/2) consistency of the findings across the multiple studies included in these meta-analyses demonstrates that the accuracy of SLNB is reproducible. As with most procedures, however, surgeon experience does seem to affect the results. In the ACOSOG trial, increased surgeon experience with SLNB correlated with a higher negative predictive value when compared with surgeons with less experience (100 % vs. 95 %) [77]. In a yet another meta-analysis by Govers et al. [62], 21 studies (847 patients) could be included. Most of these patients had oral cavity squamous cell carcinoma (OCSCC). The pooled data showed an overall sensitivity of 0.93 [95 % CI 0.90–0.95]. Negative predictive values were ranging from 0.88 to 1. Subgroup analysis showed no significant differences in subgroups. As with most procedures, however, surgeon experience does seem to affect the results. In the ACOSOG trial [77], increased surgeon experience with SLNB correlated with a higher negative predictive value when compared with surgeons with less experience (100 % vs. 95 %). A single-center study of 79 patients with clinically (after ultrasound-guided FNAC) N0 early oral cancer found a sentinel lymph node detection rate of 99 %, a sensitivity of 91 %, and a negative predictive value of 90 % [79]. This study showed evidence that the previously reported promising short-term results can be sustained through long-term follow-up.

In summary, based on the sensitivity and specificity of SLNB in early-stage oral cancer, it seems that the accuracy of SNB is similar to that of elective neck dissection. Based upon the data, the performance of SNB is best for small oral tongue lesions and for floor-of-mouth lesions; the procedure appears to be less accurate.