SLN biopsy has greatly evolved since 1955 when Seaman and Powers [32
] used radioactive gold colloid to demonstrate the concept of the first echelon sentinel node. SNL biopsy, however, remained experimental until the landmark melanoma study performed by Morton et al. [33
] in 1992. Using isosulfan blue dye, 82% of sentinel nodes were identified intraoperatively and a false-negative rate of only 1% was achieved. In 1996, Alex and Krag [34
] expanded the success of Morton et al. [33
], using lymphoscintigraphy with Tc99-labelled sulfa colloid and intraoperative γ probing to identify the sentinel node in 90% of patients with cN0 necks [34
]. With the advent of lymphoscintigraphy using Tc99-labelled sulfa colloid and intraoperative y probe, Morton et al. [33
], in a multicenter trial, combined the blue dye method and lymphoscintigraphy with intraoperative γ probing, and found 99% accuracy of identifying the sentinel node compared with 95% using only blue dye. They additionally showed that SLN biopsy with selective completion lymph node dissection had similar survival rates compared with the standard elective neck dissection, thus validating SLN biopsy as the standard of care [6
Despite success by Morton et al. [33
] with SLN mapping, the number of primary head and neck melanoma cases evaluated were low compared with primary tumors of the trunk and extremities. The broadly held belief was that evaluation of the head and neck region by SLN biopsy was risky, given the concentration of critical structures, and may not be any better than cervical lymphadenectomy [35
]. O’Brien et al. studied lymphatic drainage patterns in 97 patients using lymphoscintigraphy [35
]. Based on primary tumor sites and predicted drainage patterns, 34% of preoperative nuclear studies were discordant and 22% of sentinel nodes were found outside of the expected nodal basins [35
] They concluded that complex lymphatic drainage in the head and neck region made accurate identification of sentinel nodes uncertain. Other authors reported similar discordance rates [36
]. Pitfalls of SLN biopsy for melanoma in the head and neck included recurrences in previously dissected and pathologically negative nodal basins and technical difficulties in parotid dissections with compromise to the facial nerve [35
]. Despite these concerns, success rates for sentinel lymph node mapping remained similar to those obtained with SND in cases with melanoma of the head and neck [38
], thus validating the minimally invasive approach. Therefore, SLN biopsy is currently the standard of care for staging the neck in cutaneous head and neck melanoma at many institutions [27••
High success rates of SLN biopsy in melanoma and breast cancers have clearly driven interest for SLN biopsy in staging cN0 necks for patients with mucosal head and neck squamous cell carcinomas. Concern remains regarding the reliability of lymphatic drainage patterns in OCSCC in which, unlike cutaneous melanomas that often spread to superficial nodes first, tumors drain directly into the deep cervical nodes in a more predictable fashion [27••
]. Initial success rates were poor. In 1998, Pitman et al. [39
] performed blue-dye SLN mapping, identifying 0 of 16 sentinel nodes in patients studied [39
]. In 1999, Shoaib et al. [40
] compared 26 patients by randomizing them to undergo blue dye or radiocolloid plus blue dye. The blue dye arm resulted in the detection of only 5 (39%) sentinel nodes with 3 of 5 (60%) being false negatives compared with the rest of the neck specimen. Contrary to the blue dye-only arm, the blue dye plus radioisotope method yielded 12 of 13 (94%) sentinel nodes and showed 100% sensitivity for identifying nodal neck disease based on sentinel node findings.
After the success of Shoaib et al. [40
], many others began to evaluate SLN biopsy using the combined blue dye plus radioisotope method, and found reproducibly strong sensitivities ranging from 66% to 100% [41
]. Small sample sizes limited these studies, thus precluding recognition and implementation by others.
In light of growing interest and improving results from independent investigators, the First International Conference on Sentinel Node Biopsy in Mucosal Head and Neck Cancer was held in 2001 [19
]. Twenty-two centers contributed SLN biopsy data on 316 patients with cN0 neck disease treated with combined SLN biopsy and SND to compare the diagnostic accuracy of the procedure. The SLN identification rate was 95%, 76 (25%) patients were upstaged based on SLN biopsy, and the overall sensitivity was 90%. Eight patients (4%) were found to have false-negative SLN biopsy. Lower sensitivities were observed at institutions performing ≤ 10 cases (sensitivity, 57%) compared with those doing more than 10 cases (94% sensitivity). This confirmed the findings of previous studies in melanoma that showed a learning curve associated with SLN biopsy, and that strong intra-observer reliability is vital to obtaining accurate data [6
]. Although this study included head and neck cancers from multiple subsites and showed poor reliability between institutions based on numbers alone, the results were sufficient to suggest that SLN biopsy should be further studied as a potential modality for staging patients with head and neck mucosal SCC.
As a consequence of this conference, several single institution studies were carried out comparing SLN biopsy to the gold standard SND to determine the validity of SLN biopsy. Most studies were small and were limited in obtaining statistical power. In 2005, Paleri et al. [42
] performed a meta-analysis on 19 of these studies in order to increase the power of the studies combined. They reviewed 367 patients, and 301 patients with oral cavity lesions. They found a 97.7% sentinel node identification rate. The combined sensitivity of SLN biopsy was 92.6% with a false-negative rate of 3%. The pooled data allowed decision tree analysis to determine the overall payoff differences between SND and SLN biopsy in terms of recurrence, disease-free survival, and mortality. This showed no statistical difference between the two modalities, but did not consider morbidity data. Although no significant differences between SND and SLN biopsy were observed, a formal standardized prospective study was needed before definitive treatment guideline changes could be supported.
In 2004, Ross et al. [43
] performed a prospective randomized study of 134 patients with T1/T2 oral cavity and oropharyngeal SCC. They randomized 79 patients to receive SLN biopsy alone and 55 patients to receive SLN biopsy plus elective neck dissection. Sentinel node detection rates were 93%, and 42 (34%) patients were upstaged as a result of SLN biopsy. Overall sensitivity was 93% with an average follow-up of 2 years. Importantly, sensitivity for floor-of-mouth lesions (FOM) proved to be only 80%, significantly decreased compared with other subsites in which sensitivity was 100%. Of those treated with SLN biopsy only and negative pathologic results, 3.8% had subsequent disease within 24 months. Those treated with SLN biopsy plus elective neck dissection had a 4% false-negative rate. Follow-up results of this study at 5 years revealed a small drop in sensitivity to 91% but a strong negative predictive value of 95% [44
]. This study clearly showed that SLN biopsy was as good as elective neck dissection with regard to identifying metastatic disease. Importantly, it also identified the FOM as a site prone to failure with SLN biopsy.
Stoeckli et al. [45••
] performed an observational study of 51 patients looking at the effects of SNB with observation, and SND reserved only for patients with positive pathologic sentinel lymph nodes. Of the 51 patients, 40% of patients were upstaged and the negative predictive was 94% with two false-negative results. Recurrence rates were 6% in patients with SLN biopsy and observation. This parallels recurrence rates of 3% to 7% seen in patients treated with SND [46
]. Importantly, 13% of sentinel nodes were identified in regions not normally dissected with SND. With a reported regional recurrence of 3% to 7%, SLN biopsy performed as effectively as SND, and may have outperformed SND by identifying disease outside the expected nodal basin.
In 2006, Civantos et al. [25
] reported their experience with SLN biopsy. Specifically, they examined 106 patients at risk for regional metastases who were cN0. Pathologic staging was performed with SLN biopsy followed by SND. Of the 106 patients, 43 had oral cavity lesions, whereas others had cutaneous melanomas. Upstaging of the neck occurred in 46.5%, and 65% of patients with a positive SLN biopsy had additional positive nonsentinel nodes. SLN biopsy was 100% accurate for T1 lesions, had a false-negative rate of 10% and a negative predictive value of 92% [25
]. Fourteen percent of cases had sentinel nodes outside expected nodal basins, confirming the results of Ambrosch et al. [46
These results formed the basis for the American College of Surgeons Oncology Group Z0360 Validation Trial [27••
]. This was a well-developed multi-institutional study with strict patient selection and protocol guidelines for assessment of SLN biopsy in patients with OCSCC. Patients deemed to be cN0 by CT or MRI were eligible. PET imaging was not accepted. A total of 137 patients were enrolled. Preliminary reports revealed a sentinel node identification rate of 99.3%, an average of 3.14 sentinel nodes harvested per patient, and an upstaging or occult metastases rate of 26%. Despite surgical specimens only being evaluated with H&E staining, the negative predictive value was 94%. Again, FOM lesions were associated with a lower negative predictive value compared with other oral cavity sites, 88.5% versus 95.4%, respectively. A false-negative metastatic rate of 5% again fell within the 3% to 7% recurrence rate seen with SND. The addition of step sectioning and immunohistochemistry, currently underway, will likely improve false-negative rates at the conclusion of the study.