Sentinel Node Biopsy or Neck Dissection for Early-Stage Oral Cancer For Long Term Follow-Up: Matched-Pair Analysis for pN0 Necks

Background: Oral cancer spreads preferentially through lymphatics with a high rate of micrometastases. We studied the rate of neck recurrence in patients with T1/T2 squamous cell carcinoma (SCC) of oral cavity with clinically negative necks (cN0) submitted to elective neck dissection (END) or sentinel lymph node biopsy (SLNB) with a histopathologically negative specimen of END (pN0) or sentinel nodes (SN). Methods: Prospective study matched paired with a retrospective control group. We studied forty-three patients: 25 in the SLNB group and 18 in the control group (END), and compared both patient cohorts, with negative lymph node metastases with more than ve years of follow-up. We matched both groups by gender, age, site, stage, pathological status of the primary lesion, adjuvant treatment, length of follow-up by Odds ratio analysis with 95% condence interval and t-test without signicant difference all categories(p>0.05). Results: Both SLNB and END groups had a similar recurrence rate in the neck, without signicant difference. Conclusions: Elective neck dissection may be unnecessary in this study population when SN comes out without metastases after extensive histopathological evaluation.


Introduction
Head and neck squamous cell carcinoma spreads preferentially through the lymphatic system and plays a signi cant role in decreasing survival 1,2 . Some controversy remains as observation and therapeutic neck dissection or elective neck dissection for clinically N0 necks(cN0). CT scan, MR, or PET CT for evaluation of micrometastases had a high rate of false-negative and positive [3][4][5][6][7][8] . Histopathological examination of the surgical specimen following neck dissection is considered the gold standard method for neck staging 9 . We must perform neck dissection in patients whose primary lesion has an expected rate of lymphatic metastases higher than 20% 1,2 . Even patients with early SCC of the oral cavity, more than two millimeters of depth, possess this risk of metastasis, and elective neck dissection (END) must be performed [10][11][12] . Neck dissection, in addition to resection of the primary tumor, increases surgical length and morbidity, but END has a survival advantage over wait and sees policy 13,14 .
In cohorts with subsequent END after negative SLNB, only 3 to 5% were found to have additional metastatic lymph nodes in the surgical specimen. It suggests a low probability of neck recurrence after SLNB without neck dissection when the sentinel node is histopathological negative [23][24][25] . This recurrence rate would be lower than reported in the literature following selective I-III neck dissection (of up to 10%) 26 , even after radical neck dissection in pN0 necks (6.7-18%) 27,28 .
We compared the rate of neck recurrence for two treatment modalities for patients with early-stage SCC of the oral cavity (T1N0 or T2N0). The treatments were excision of the primary lesion with END versus the excision of the primary lesion with SLNB, without subsequent neck dissection when SLNB came out negatively on the permanent section after extensive pathological analysis.

Materials And Methods
We included only patients with SCC (squamous cell carcinoma) of the oral cavity, con rmed by histopathological examination at our institution from 2010 to 2012. We used the AJCC Eighth Edition for the staging of the primary lesions and neck 29 . All participants were radiologically negative for lymphatic metastases by multi-slice CT scan with 128 detectors. Consecutive patients submitted to SLNB, which come out negative after extensive histopathological evaluation, were prospectively followed without subsequent neck dissection. Ethical Committee of XXXXXX approved the study under CAAE:0392.0.146.000-07. All participants signed informed consent.
Only negative sentinel lymph node patients without local recurrences were included in statistical analysis to avoid primary recurrence bias.
Patients with positive margins on a permanent section (despite negative results in frozen section), perineural invasion, or vascular emboli were treated with adjuvant radiation therapy on primary tumor bed and neck.
We used two peritumoral injections of technetium labeled nanocolloid for SLNB.
Lymphoscintigraphy and SPEC-CT were performed in all cases. The neck skin was marked accordingly, and a gamma probe was used to identify the sentinel lymph node intraoperatively. We obtained step serial sections at each 150 um of the sentinel lymph node stained with hematoxylin-eosin, and subsequently, immunohistochemistry for cytokeratin AE-1/AE-3 in negative SLNB on HE.
All patients were followed postoperatively with a CT scan every six months for the rst two years and yearly.
The control group submitted to END was collected with reviewed medical charts of all consecutive patients with T1/T2 oral cancer, cN0 necks. All with radiologically negative necks, from 2000 to 2012 submitted to END of levels I to III. All pathological evaluation was performed with an extensive evaluation of all para n embedded blocks with step serial section of 150 um.
Both groups were match-paired by age, gender, stage, site, histopathological status of the primary lesion and neck dissection specimen, adjuvant treatment, and follow-up length. Only patients with negative surgical specimens for lymphatic metastases were included for pairing, without local recurrences.
Statistics were performed via IBM SPSS Statistics 17 (Windows) software. The odds ratio was used to compare groups by gender, stage, site, adjuvant treatment, and recurrence rate and T-test for continuous categories of age and length of follow-up. P-values < 0.05 were considered statistically signi cant.

Results
Our study comprised 43 patients, 25 of them in the SLNB group and 18 in the control group (END), as detailed in Table 1. The average age of patients in the SLNB group was 58.9(SD 5.4y) years and, in the END, was 58.1(SD 6.4y) years; there was no signi cant difference between groups(p > 0.05). The length of follow-up in both groups was from 5 to 8 years with an average of 6.7y and 5.8 y for SLNB and END, respectively (SD: SLNB 1.3y and 2.2y END) years at the closure of study without signi cant difference in both groups(p > 0.05).
In 8% of SLNB cases and 15% of END group included in this study received postoperative radiation therapy due to the presence of positive margins, perineural invasion, or vascular emboli without statistical difference.
In this study's cohorts, none recurred in the SLNB group nor the END group, respectively, without a statistical difference between these recurrences rates. All patients were alive without disease.
Almost 80% of patients with early-stage SCC of oral cavity submitted to END were pathologically negative. They could be identi ed on SLNB technique avoiding unnecessary neck dissection with a low rate of false-negative cases [23][24][25] . The prevailing view regarding END is that it should be performed in a patient with a cN0 only if the risk of occult metastasis is higher than 20% [30][31][32] . However, negative SNLB with a false-negative rate of 3 to 5% poses END unnecessary in these cases, as this risk is lower than the 20% considered a threat.
Prospective studies comparing postoperative complications and quality of life in SCC of the oral cavity and oropharynx submitted to either SLNB or END observed fewer complications and improved quality of life in the SLNB group 32,33,34 .
Although the high rate of sensitivity and negative predictive value of SLNB in oral cancer, no studies compare the rate of neck recurrence in SLNB and END with pathologically negative nodes in large prospective randomized trials like melanoma studies. Only one study compared a large retrospective cohort of patients of a database from the National Cancer Data Base without pairing both groups of patients 35 . Other two multicenter prospective trials observed a non-inferiority of SLNB over END regarding recurrence free, disease speci c and overall survival, but not all patients had a longer follow-up 36-37 . Our study was the rst to evaluate the recurrence rate between SNB and END groups with a longer follow-up, with a proposal of pairing patients statistically. We observed that both the SLNB and END groups had no recurrences in the neck with the same recurrence rates, i.e., 0%, suggesting the unnecessary elective neck dissection when the sentinel lymph node is negative. This low rate of recurrence could be due to small number of cases or adjuvant treatment could be contributed to this rate. But when entire neck dissection specimen is evaluated with extensive pathological evaluation, when it comes pathologically negative it is truly negative leading to a low rate of regional recurrences. With a shallow rate of false-negative cases of 2% even after an extensive evaluation of these non-sentinel lymph nodes with step serial section and immunohistochemistry 38 , if these micrometastases recur, this rate of recurrence is still comparable to neck recurrence after selective neck dissection with histopathological negative necks. Considering most patients would have histopathological negative necks, it would be possible to select which patient will need neck dissection and who will not need it to save every negative sentinel lymph node 39 . The only disadvantage of the SLNB technique would be performing two-step surgery in a positive sentinel node. Still, this situation would occur in the minority of patients, and a two-week interval will resolve the problem after a complete pathological evaluation of the specimen. This two-step surgery will decrease with real-time rapid quantitative reverse transcription PCR of lymph node with results in 30 minutes with more than 97% accuracy reaching even 100% 40,41 . For more evidence of the reliability of SLNB regarding recurrence and survival, we will need more multicenter prospective randomized trials, with longer followup. Negative predictive value and sensitivity are very high for early-stage oral cancer in all studies of SLNB. However, the sample size of this study is small, representing a major weakness of our study. But given the nature of the research and prospective studies' toughness with large samples, particularly regarding early-stage oral cancer patients our study could, also, ll the gap of comparison results of SNLB and END with pathological negative nodes. It seems an excellent option to avoid END when sentinel lymph node comes out as negative after comprehensive pathological evaluation, with less morbidity and high cost-effectiveness.

Conclusion
END may be unnecessary in the early stage, T1/T2 oral cavity SCC when sentinel lymph node comes out as negative after extensive pathological evaluation with immunohistochemistry and step serial section.

Declarations
Con ict of Interest: Nothing to declare.
Financial disclosure: Nothing to declare.

Source of funding: none
Informed consent was obtained from all patients.
All data and material are available. The datasets during and/or analysed during the current study available from the corresponding author on reasonable request.