Ross G, Shoaib T, Soutar DS, Camilleri IG, Gray HW, Bessent RG, Robertson AG, MacDonald DG. The Use of Sentinel Node Biopsy to Upstage the Clinically N0 Neck in Head and Neck Cancer. Arch Otolaryngol Head Neck Surg. 2002;128(11):1287–1291. doi:10.1001/archotol.128.11.1287
To investigate the possible role of sentinel node biopsy (SNB) alone to upstage the clinically N0 neck in patients with oral and oropharyngeal squamous cell carcinoma.
Prospective clinical study.
Head and neck referral center.
Patients with primary untreated oral and/or oropharyngeal squamous cell carcinoma accessible to injection and with clinically N0 necks were enrolled in the study.
An SNB was performed after radiocolloid and blue dye injection. Preoperative lymphoscintigraphy and the perioperative use of a gamma probe identified radioactive sentinel nodes and visualization of blue-stained lymphatics identified blue sentinel nodes. If the sentinel node was found negative, there was no further treatment to the neck. If the sentinel node tested positive, a therapeutic neck dissection was performed. All patients underwent regular follow-up at the outpatient clinic to identify possible recurrence.
Main Outcome Measures
Upstaging of the clinically N0 neck by SNB and development of subsequent disease in SNB-negative necks.
An SNB was performed on 57 clinically N0 necks in 48 patients. Sentinel nodes were harvested in 43 (90%) of 48 patients. Fifteen (35%) of 43 patients were upstaged by SNB and 28 (65%) of 43 were staged SNB negative. There was a mean follow-up of 18 months. One patient developed subsequent disease after having been staged negative with SNB. The overall sensitivity of the procedure using the full pathologic protocol was 94% (15/16).
Sentinel node biopsy can be used to upstage the N0 neck in patients with early subclinical nodal disease. However, before it becomes the standard of care in head and neck squamous cell carcinoma, longer follow-up observational trials are needed.
A CONTINUING DEBATE in head and neck cancer is the management of the clinically N0 neck.1,2 Currently accepted management policies are that patients with a greater than 20% chance of subclinical metastases, based on the anatomic site and the size of the primary tumor, should undergo elective neck dissection (END).3- 5 However, such a policy might still overtreat up to 80% of patients, and ENDs carry with them an associated morbidity.6- 8
The theory behind the sentinel node biopsy (SNB) approach is that the lymphatic flow from a tumor travels sequentially to the first-echelon node (sentinel node) and then on to the remaining nodal basin.9 Thus, if the SNB concept holds true, the pathologic evaluation of a harvested sentinel node will give an accurate reflection of disease within the rest of the draining nodal basin. In melanoma, the sentinel node identifies node-positive patients with over a 95% sensitivity,10 thus accurately predicting which patients do not require a lymph node dissection.
The SNB technique is cost-effective in the management of melanoma and has been associated with minimal morbidity compared with END.11 The popularity of the technique in head and neck squamous cell carcinoma is growing.12- 16 Initially we performed SNBs on 40 necks followed by END.12 The pathologic characteristics of the sentinel node reflected that of the END in 16 of 17 cases (sensitivity, 94%).12
Having successfully implemented SNB as a diagnostic tool, we obtained ethical approval to use SNB alone to stage the node-negative neck. The aim of this study was to assess the utility of SNB in patients with N0 necks who might otherwise have undergone END for staging.
Forty-eight patients have entered the study to date. Patients undergoing SNB were admitted the day prior to surgery. The triple diagnostic procedure of preoperative lymphoscintigraphy and intraoperative blue dye and gamma probe has been described.12,13 Briefly, on the day prior to surgery, patients attended the Nuclear Medicine Department where up to 1.1 mCi (40 MBq) of technetium Tc 99m–labeled colloidal human serum albumin—either Nanocoll or Albures (Amersham Health, Little Chalfont, Buckinghamshire, England) in approximately 0.5 to 1 mL of isotonic sodium chloride solution—was injected at as many points as necessary in an attempt to completely surround the tumor on its deep and lateral aspects.17 Static lymphoscintigraphy was performed at 15 minutes, 30 minutes, and 60 minutes after injection, or until the first appearance of sentinel nodes within the neck. The locations of radioactive lymph nodes were marked on the patient's skin with a cobalt 57 solid source pen, and the pen was moved until its position overlaid that of a radioactive node. This position was marked on the skin using indelible ink. The lymphoscintigraphy scans, which can be manipulated to mask the primary injection site, were also available to the surgeon in the operating room.
During surgery, approximately 0.5 to 2 mL of Patent Blue V dye (Laboratoire Guerbet, Aulnay-Sous-Bois, France) was injected into the same site as the radiocolloid. A suitable incision was made in the neck in such a position as to facilitate excision of the scar should a subsequent therapeutic neck dissection (TND) be necessary. Blue-stained lymphatics, if seen, were traced to the first draining lymph node, which was harvested. All radioactive lymph nodes were identified with a Neoprobe 1500 handheld gamma probe (Neoprobe Corporation, Dublin, Ohio). The handheld gamma probe was used to identify radioactive sentinel nodes, including those marked preoperatively during lymphoscintigraphy. To reduce detection of radiation from the injection site, a series of malleable sterilized lead plates were used to mask the injection site, thus aiding in vivo identification of radioactive nodes. Radioactive nodes were excised, and radioactivity within the node was confirmed ex vivo. Sentinel nodes were labeled according to their color, radioactivity, and anatomic neck level.18
The sentinel nodes were fixed in 10% neutral buffered formalin and then bisected through their longest axis. If the thickness of the halves was more than approximately 2.5 mm, the slices were further trimmed to provide additional 2.5-mm-thick blocks. One hematoxylin-eosin (H&E)–stained section was prepared from each histologic block and examined for possible metastasis.
The full pathologic protocol was used to examine nodes that appeared negative following examination with H&E, and these nodes were step serial sectioned at 150-µm levels. One section from each level within the block was H&E stained and examined. If the node still appeared free from tumor, immunocytochemical analysis for cytokeratin (AE1/3) was undertaken. Cytokeratin positivity was compared with the adjacent H&E-stained serial section to confirm that it represented viable tumor cells. If metastatic tumor was found on routine H&E staining, step sectioning, or immunohistochemical analysis, a TND was undertaken in the form of a modified radical neck dissection18 with preservation of the accessory nerve, sternocleidomastoid muscle, and internal jugular vein.
The full pathologic protocol was performed prospectively on the last 27 patients in the study in whom a sentinel node was successfully harvested. Only an H&E examination was performed on the first 16 patients. Two patients who developed nodal disease following SNB had the sentinel nodes reexamined using the full pathologic protocol.
Patients were upstaged according to the current TNM classification.19 Patients upstaged on the demonstration of micrometastasis only within the sentinel node by step serial sectioning or immunocytochemical analysis were classified as pN1mi20 if no subsequent disease was found to be present. Patients with negative SNB findings were followed-up as outpatients with clinical examination of the neck every 3 months, and no further treatment to the neck was carried out. The mean follow-up for these patients has been 18 months (range, 9-32 months).
The male-female patient ratio was 5:3, and the mean age was 61 years (range, 33-88 years). Forty-five N0 patients underwent SNB to stage the clinically negative neck, 36 had SNB performed unilaterally, and 9 had SNB bilaterally. Three N-positive patients underwent SNB to stage the contralateral clinically negative neck and a TND to treat clinically suspected or confirmed nodal metastases on the ipsilateral neck for tumors close to or crossing the midline. In total, SNB was performed in 57 necks from 48 patients. Sentinel nodes were found in 43 of 48 patients, 50 of 57 necks.
The clinical stage of the primary tumor was T1 in 22 patients, T2 in 14, T3 in 1, and T4 in 11. Forty patients had the primary tumor treated by excision and 8 by brachytherapy. The site of the primary squamous cell carcinoma was the anterior tongue in 19 patients, the floor of the mouth in 17, posterior/base of the tongue in 5, retromolar in 3, hard palate in 1, soft palate in 1, buccal mucosa in 1, and the tonsillar fossa in 1. The T stage distribution for oral cavity tumors was T1 in 20, T2 in 13, T3 in 1, and T4 in 8. This compared with the T stage distribution for oropharyngeal tumors of T1 in 2, T2 in 1, and T4 in 3.
There were 7 necks in 6 patients in whom a sentinel node could not be identified. Four of these patients had floor-of-mouth tumors, and 2 had tongue tumors. The sentinel node identification rate was 43 (90%) of 48 patients. The total number of sentinel nodes harvested was 104 from 43 patients (mean, 2.4 nodes per patient). Table 1 summarizes the findings. The largest number of nodes was in level II. Positive nodes were found in levels I through III but not in level IV. There were no retropharyngeal, level V, or level VI nodes identified with lymphoscintigraphy preoperatively or the gamma probe intraoperatively, and no search was made for blue nodes in these regions via a separate incision. A total of 62 nodes were hot and blue, 30 were hot only, and 12 were blue only. There was 1 patient in whom a blue-only node was positive when a hot node was not positive.
In 15 (35%) of 43 patients, the SNB findings upstaged the clinically N0 neck. Thirteen patients underwent subsequent neck dissection. The upstaging of the 15 patients following SNB and subsequent neck dissection of 13 found 7 patients upstaged from pN0 to pN2b, 3 patients upstaged from pN0 to pN1, 3 patients from pN0 to pN1mi, and 2 patients from pN2b to pN2c. The results of each upstaged patient and the subsequent TND are provided in Table 2. In 2 of these patients the sentinel node was initially staged negative using H&E. Both patients developed subsequent disease, but examination of the initial sentinel node using the full pathologic protocol in retrospect found that they were sentinel node positive. The upstaging of the clinically N0 neck with SNB including these 2 patients was 25% (5/20) of T1 patients, 42% (5/12) of T2 patients, and 45% (5/11) of T3/T4 patients. Upstaging according to site of tumor is shown in Table 3. The distribution of negative and positive sentinel nodes for oral cavity tumors compared with oropharyngeal tumors is provided in Table 4. The mean follow-up has been 18 months (range, 9-32 months). During this follow-up, 3 patients staged SNB negative have died from unrelated causes, 2 have developed a second primary tumor, 1 patient has developed local recurrence, and 1 developed a primary bronchial carcinoma. One patient staged SNB negative using the full pathologic protocol developed subsequent disease 26 months after SNB (Table 2). The true sensitivity of the procedure is therefore 94% (15/16) when using the full pathologic protocol. The remaining 20 SNB-negative patients remain disease free.
When we used a combination of blue dye and radiocolloid injection, our success rate in identifying sentinel nodes within the neck was 90% (43/48). This compares favorably with the rate of sentinel node identification in patients undergoing SNB for cutaneous lesions of the head and neck.10,21 It is also similar to the identification rate found by members of our group on the clinically N0 neck in conjunction with a neck dissection.12 It would seem, therefore, that surgical access does not affect the successful harvesting of sentinel nodes.
A total of 22 nodes contained metastases in 15 patients; 95% (21/22) of positive sentinel nodes were detected by radiocolloid, and 82% (18/22) of positive nodes were found by blue dye. There were 4 positive hot nodes in which no blue dye was seen. There was only 1 case, a posterior/base-of-tongue tumor, in which a blue-only sentinel node was positive where no other positive sentinel nodes were found. It is more likely in tumors with more difficult access (such as posterior/base-of-tongue tumors) that the injections will not be in the same position because the injection of blue dye is carried out under general anesthesia, whereas the injection of radiocolloid occurs with the patient awake. We have found that the blue dye and gamma probe supplement each other in the identification process.
The procedure in head and neck cancer may be more technically demanding than in other areas of the body. The close proximity of sentinel nodes to the primary site, especially when the primary site is the floor of the mouth, renders gamma probe identification of radioactive nodes in the submandibular and submental triangles difficult. Four of the 6 patients in whom a sentinel node was not identified had floor-of-mouth tumors. One patient had bilateral nodes visible on lymphoscintigraphy, but no sentinel nodes were subsequently harvested. Another patient with bilateral nodes visible on lymphoscintigraphy had 2 sentinel nodes harvested from 1 neck only. In an attempt to improve identification for floor-of-mouth tumors, we use malleable lead plates. These shield the injection site, reduce shine-through and scatter from the primary site, and aid radiolocalization.
The upstaging of the clinically N0 neck occurred in 5 of 20 T1 patients, 5 of 12 T2 patients and 5 of 11 T3/T4 patients, which reflects the incidence of metastatic spread for these T classifications. The most common site of upstaging was for retromolar tumors, which were upstaged in all cases (3/3) compared with posterior/base-of-tongue (2/4), palatal (1/2), floor-of-mouth (4/14), and anterior tongue (5/18) tumors. Retromolar tumors are difficult to diagnose at an early stage, and even small tumors may be present for some time before presentation. It may be that retromolar tumors metastasize earlier than other oral cavity tumors. Six (46%) of 13 sentinel nodes harvested from 2 (33%) of 6 patients with oropharyngeal tumors were positive compared with 16 (28%) of 91 nodes of the 13 (35%) of 37 patients with oral cavity tumors.
In 1 patient, the sentinel node did not reflect the draining lymph node basin, and this patient developed subsequent disease 26 months after the SNB procedure. During the initial SNB procedure a hot blue sentinel node was harvested from level III. The procedure was hampered by the patient's high (>30) body mass index (BMI; calculated by dividing the patient's weight in kilograms by the squared height in meters). The subsequent neck dissection 26 months later found disease in levels I and II. It is likely that in this case the sentinel node harvested was a second-echelon node and the true sentinel node was not harvested. This patient had a floor-of-mouth tumor and, as previously described, the closeness of the primary site may have been the reason why the incorrect node was initially harvested. Undoubtedly the procedure was made more difficult by the patient's high BMI. Most neck recurrences will be identified within the first year and almost all within 18 months.22 In this case the detection of neck recurrence was also impaired by the high BMI. We do not expect such a delay in presentation within the 20 SNB-negative patients who are currently being followed up.
Fifteen patients were upstaged with SNB. Two patients have not had a TND (both having micrometastatic disease in only 1 sentinel node). The first was an 80-year-old woman who developed angina during the SNB procedure, and it was felt that performing a neck dissection was an anesthetic risk. The second patient had a positive sentinel node on the contralateral N0 neck, the clinically N-positive ipsilateral neck having been already treated by a neck dissection. The ipsilateral neck received radiotherapy, the field of which included the contralateral neck with micrometastatic disease. It was decided not to perform a TND in this case to avoid the morbidity of a bilateral neck dissection. Both patients remain disease free.
In total, 13 patients have had a TND. Two of these patients had a dissection following subsequent disease after being staged SNB negative on routine H&E staining. On additional pathologic evaluation of the sentinel nodes, they were found to be positive in retrospect. On additional pathologic evaluation of the sentinel nodes, they were found to be positive in retrospect. In the remaining 11 TND specimens, the sentinel nodes were the only positive nodes in 7 (64%) of 11 cases. In 4 (36%) of 11 patients, there was further disease found during the subsequent TND. The upstaging of disease with additional pathologic evaluation highlights the importance of additional pathologic evaluation of sentinel nodes.
The upstaging of patients with the finding of micrometastasis only has been classified as pN1 (mi).18 It is undetermined whether the presence of micrometastasis will alter prognosis; this determination will require the longer-term follow-up possible only in a larger group of patients. It is also undetermined what the optimal treatment is for patients in whom micrometastasis only is seen within a sentinel node. We would recommend from the above evidence that all patients with micrometastasis within a sentinel node undergo a TND. Currently, we do not recommend any further adjuvant therapy for these patients unless further disease is demonstrated following the TND. All patients in the present study upstaged to N1, N2b, or N2c by SNB underwent postoperative radiotherapy to the upstaged neck.
The true sensitivity of the procedure using the full pathologic protocol was 94% (15/16), which is similar to previous results in the context of a neck dissection.12 If SNB has a similar sensitivity to that of an END while being more cost-effective and carrying less morbidity, then it should be an improvement in quality of care.11 Multicenter trials are required before SNB can be used as standard of care in the treatment of the clinically N0 neck. Details of our current multicenter trial are available at http://www.canniesburn.org/sentinel/sentinelframe.html (accessed August 5, 2002).
Accepted for publication May 8, 2002.
This study was funded by a grant from the Chief Scientist Office, Scottish Executive Health Department, Edinburgh, Scotland. Joyce Hope prepared the pathologic preparations of the sentinel nodes.
Corresponding author and reprints: Gary Ross, MRCSEd, Plastic Surgery Unit, Canniesburn Hospital, Switchback Road, Bearsden, Glasgow G61 1QL, Scotland (e-mail: email@example.com).