Plaveiselcelcarcinomen van de mondholte metastaseren primair lymfogeen naar lymfeklieren in de hals. In het geval van aantoonbare lymfekliermetastase(n) wordt de hals van de patiënt doorgaans behandeld met een halsklierdissectie. Bij een klinisch (en bij beeldvorming) negatieve hals (cN0) zijn er twee opties: een electieve (profylactische) halsklierdissectie om occulte (dat wil zeggen klinisch en bij beeldvorming niet te detecteren) metastasen te verwijderen of een “watchful waiting” beleid waarbij de hals pas behandeld wordt bij manifeste metastasering tijdens follow-up. Dit dilemma doet zich met name voor bij kleine mondholtecarcinomen waarbij de hals niet geopend hoeft te worden voor resectie van de primaire tumor of reconstructie van het chirurgisch defect in de mondholte. Van patiënten met een klinisch of bij beeldvorming negatieve hals is het bekend dat ongeveer 30% van de patiënten occulte metastasen heeft. Van oudsher bestaat bij deze patiënten dan ook het dilemma of er een electieve halsklierdissectie dient te worden uitgevoerd (overbehandeling bij 70% van de patiënten met hierbij behorende postoperatieve morbiditeit zoals mogelijk een gestoorde schouderfunctie) of dat kan worden volstaan met een afwachtend beleid (onderbehandeling bij 30% van de patiënten met het risico dat een occulte metastase zich zal ontwikkelen tot een grotere metastase met mogelijk uitgebreide en zelfs inoperabele ziekte tot gevolg). Het risico op het ten onrechte afzien van een behandeling van de hals kan verminderd worden door de diagnostiek voorafgaand aan behandeling te verbeteren. Daarnaast is het als onderdeel van de diagnostische work-up mogelijk om een schildwachtklierprocedure (sentinel node procedure) uit te voeren, wat als voordeel heeft dat er een histologische bevestiging van de oncologische klierstatus (N) van de hals komt. Het nadeel van een schildwachtklierprocedure is dat indien een metastase gevonden wordt in tweede instantie een halsklierdissectie moet worden uitgevoerd. De voorliggende vraag is, indien er gekozen wordt voor een stadierende ingreep voor de hals, is een schildwachtklierprocedure dan gelijkwaardig aan een electieve halsklierdissectie bij patiënten met een mondholtecarcinoom en een negatieve hals?

Description of studies

Hasegawa (2021) performed a multi-center randomized trial in Japan to assess the hypothesized noninferiority of a sentinel lymph node biopsy (SLNB) compared to elective neck dissection (ND). The border of clinical relevance for noninferiority was set at 12% for the three-year overall survival of patients with an oral cavity squamous cell carcinoma. Patients were included when they had a T1-2N0M0 oral cavity squamous cell carcinoma (according to the TNM 7^th ed.), when there were no lymph node metastases on contrast enhanced CT of the head and neck, when the patient did not receive prior treatment, and when the patient was 18 years or older. Patients with T1 tumors having ≤ 4mm depth of invasion were excluded, as well as patients with a recurrence after definitive treatment, history of radiation to the neck, had planned or were currently pregnant or lactating, or had other disqualifying reasons as judged by the attending physician. Patients were allocated to a treatment through a stratified randomization (T-stage: T1 versus T2, tumor site: tongue versus other). The clinical neck status of patients was examined by CT and supplemented with ultrasound when necessary. The study center could decide to use MRI and PET/CT additionally at their own discretion. Four patients were excluded after randomization (n=3 did not fulfill the eligibility criteria (n=1 ND, n=2 SLNB), n=1 declined treatment (SLNB)). Tumor resection was performed during surgery. For the group undergoing SLNB, the sentinel lymph node identification was performed with ^99mTc phytate and was injected in the peritumoral mucosa the day before surgery. A gamma probe was used to identify the sentinel node on the day of surgery. During surgery a frozen section analysis was performed on 2mm blocks. When the result was negative, the blocks were paraffin embedded for staining analysis on 4mm sections (hematoxylin and eosin, and cytokeratin). Isolated tumor cells were treated as metastasis-positive. Patients with positive sentinel nodes on frozen section immediately underwent neck dissection (level I-IV or I-V). If the resected specimen was positive in pathological analysis, the neck dissection was performed within 6 weeks of the initial surgery. When no metastasis was detected in the sentinel node and if the resection required a pull-through resection, a supraomohyoid neck dissection was performed. Patients with a negative sentinel node on the contralateral side of the neck received a sentinel node basin dissection. The group undergoing neck dissection received a supraomohyoid neck dissection. Patients with a positive resection margin received a reoperation or radiationtherapy (with or without chemotherapy) at the discretion of the study center. Radiotherapy was initiated within 6 weeks when patients had extracapsular spread of the lymph node metastasis. Concomitant chemotherapy was left at the discretion of the study center. Patients were followed up to a median of 3.1 years. Patients in the SLNB-group (n=134, 66.4% males) had a median age of 63 years (range: 21 to 90) and tumors located at the tongue (n=109), floor of mouth (n=13), lower gingiva (n=7), or buccal mucosa (n=5). Tumor stage was either T1 (n=26) or T2 (n=108) with a pN status of pN- (n=86), pN+ (n=46), or pNx (n=4). Four patients received postoperative radiation and/or chemotherapy. The group undergoing elective ND (n=137, 65.7% males) had a mean age of 63 years (range: 28 to 85). The primary tumor was located at the tongue (n=114), floor of mouth (n=14), lower gingiva (n=6), or buccal mucosa (n=3). Patients had a T-stage of either T1 (n=25) or T2 (n=112). The pN status was either pN- (n=99), pN+ (n=34), or pNx (n=4). Three patients received postoperative radiation or chemotherapy.

Garrel (2020) performed a multi-center open-label randomized controlled trial in France to investigate the equivalence of sentinel lymph node biopsy to elective neck lymph node dissection. The border of equivalence was defined at a recurrence rate difference of 0.10 (H_null: SLNB-END ≥ 0.10; H_alternative: SLNB-END < 0.10). Eligible patients had to be older than 18 years with health insurance, signed an informed consent, were not participating in another trial, and had an operable T1-2N0M0 primary oral or oropharyngeal squamous cell carcinomas (diagnosed by biopsy with histopathologic analysis in the month prior to inclusion). Patients were excluded when they had treatment for other cancers, non-invasive tumors (high grade dysplasia, in situ carcinoma), inadequate tumor resection, contraindications (for sentinel lymph node biopsy, lymph node dissection, radiotherapy, medical imaging), allergy or intolerance to contrast product, pregnancy, refusal to accept full treatment, when follow-up was not possible or when follow-up was refused, when already treated for the tumor, when chemotherapy or immunotherapy was received in the prior 6 months, or when there was a history of neck surgery or radiotherapy. Preoperatively a contrast enhanced CT or MRI (when there were contraindications for CT) was performed to check for the clinically negative neck (cN0). All participants (n=307) received excision of the primary tumor. After randomization 28 participants were excluded (intervention: n=15, control: n=13) due to high grade dysplasia (n=2), carcinoma in situ (n=10), history of surgery for oral cavity carcinoma (n=1), history of radiotherapy for oropharyngeal carcinoma (n=1), R1 margin without completion (n=1), refusal of surgery (n=1), decision for radiotherapy (n=1), refusal of random allocation (n=4), withdrawal of consent (n=1), synchronous lung cancer (n=1), urgent carotid surgery (n=1), erroneous inclusion of T4 tumor (n=2), investigator decision (n=1), or neck dissection despite ITC1 only (n=1). Participants in the intervention group received treatment with sentinel lymph node biopsy and neck dissection in case of positive sentinel lymph nodes (SLNB-group, n= 140 after exclusions). Preoperative sentinel lymph node identification was performed with radiotracer injections and lymphoscintigraphy. A portable gamma probe was used to identify sentinel lymph nodus during surgery and intraoperative histopathology was performed by imprint cytology or frozen sections. Positive tumor invasion was defined as the presence of at least one micrometastasis (tumor tissue 200mm-2mm) or one macrometastasis (tumor tissue > 2mm). Isolated tumor cells (< 200mm) were not considered to be nodal invasion. When sentinel lymph node positivity was detected post-surgery, a neck dissection was performed. Eight persons in the SLNB-group received a neck dissection because sentinel lymph node biopsies failed and twelve patients received a neck dissection after initial negative intraoperative histopathological analysis (total neck dissections in SLNB-group: 21/140, 15%) The control group (END-group, n=139 after exclusion) received neck dissection. Patients received adjuvant radiotherapy if two or more lymph nodes (including the sentinel node, when applicable) were positive. Concomitant adjuvant chemotherapy was proposed when poor prognostic factors were present (for example vascular, perineural, or muscular invasion). Follow-up was performed by the surgeon every 2 months (first year), every 4 months (following year), and once yearly (up to the fifth year). Every visit had a clinical examination. A neck-thorax CT was performed at year 1 and year 2 post-surgery. Functional follow-up was performed at month 2, 4, 6, 12, and 24 post-surgery. Mean follow-up duration was 4.95 years (SD: 2.45).

Results

Neck recurrence

Hasegawa (2021) observed 15 regional recurrences (11.2%) in the sentinel lymph node biopsy-group, while 13 (9.5%) regional recurrences were observed in the neck dissection group. A relative risk was calculated from this data (RR=1.18, 95%CI: 0.58 to 2.38). In natural frequencies (using the baseline risk of 9.5%), a neck dissection results in 95 patients with regional recurrences from a group of 1000 patients compared to 113 patients (95%CI: 56 to 217) with regional recurrences out of 1000 patients for a sentinel lymph node biopsy. According to these numbers, sentinel lymph node biopsy can result in a range of 39 less patients to 122 more patients with regional recurrences per 1000 patients compared to neck dissection.

Garrell (2020) reported 13 recurrences without relapse of the primary tumor in the sentinel lymph node biopsy-group (9.3%), compared to 14 recurrences in the elective neck dissection-group (10.1%). There were no statistically significant differences when tested with a Chi²-test (p=0.82). From this data a relative risk was calculated: RR = 0.92 (95%CI: 0.45 to 1.89) with the point estimate favoring the sentinel lymph node biopsy procedure. In natural frequencies (using the baseline risk of 10.1%), a neck dissection then results in 101 patients with regional recurrences from a group of 1000 patients compared to 93 patients (95%CI: 46 to 191) with regional recurrences out of 1000 patients for a sentinel lymph node biopsy. According to these numbers, sentinel lymph node biopsy can result in a range of 55 less patients to 90 more patients with regional recurrences per 1000 patients compared to neck dissection.

Disease-specific survival

Garrel (2020) reported the 2-year and 5-year disease specific survival. The 2-year specific survival in the sentinel lymph node biopsy-group was 93.0% (95%CI: 87 to 96), compared to 95.5% (95%CI: 90 to 98) in the neck dissection-group. The 5-year disease-specific survival was 87.1% (95%CI: 79 to 92) in the sentinel lymph node biopsy group versus 88.6 (95%CI: 82 to 93) in the neck dissection group. No statistically significant difference between groups were found (p=0.68).

Overall survival

Hasegawa (2021) reported the 3-year overall survival. In the sentinel lymph node biopsy-group the survival was 87.9% (one-sided lower limit of the 95%CI: 82.4%) compared to 86.6% (one-sided lower limit of the 95%CI: 80.9%) survival in the neck dissection group at three years.

Garrel (2020) investigated the 2-year and 5-year overall survival. The 2-year overall survival in the sentinel lymph node biopsy-group was 88.7% (95%CI: 82 to 93) versus 92.6% (95%CI: 87 to 96) in the neck dissection-group. The 5-year overall survival in the sentinel lymph node biopsy-group was 82.2% (95%CI: 74 to 88) compared to 81.8% (95%CI: 74 to 88) in the neck dissection-group. No statistically significant differences between groups were found (p=0.42).

Quality of life

Hasegawa (2021) reported the quality of life using the Neck Dissection Quality of Life Questionnaire. Subscales in this questionnaire that described outcomes of interest in this guideline module are described under the respective outcome (i.e. the subscales: pain, shoulder drop, reach above). The other quality of life subscales are described in Table 1 for 1, 3, 6, and 12 months postoperatively. Medians were calculated from the reported frequency table.

Table 1 Median scores on Quality of Life subscales as measured by Hasegawa (2021) with the Neck Dissection Quality of Life Questionnaire

Shoulder morbidity

Hasegawa (2021) measured ‘shoulder drop’ and ‘reach above’ using the Neck Dissection Quality of Life Questionnaire. Since these subscales specifically describe an aspect of shoulder morbidity, these subscales were described here. Hasegawa (2021) also used the shoulder abduction test. Results are presented in Table 2 for 1, 3, 6, and 12 months postoperatively. Median scores were derived from the reported data in a frequency table.

Table 2 Median scores for shoulder morbidity, from Hasegawa (2021)

Garrel (2020) assessed shoulder mobility with the neck-shoulder impairment scale (self-reported). Results were displayed in a figure and exact scores can only be approximated. Significance tests at different time points on the questionnaire’s items are summarized in Table 3. Garrell (2020) also reported the proportion of patients capable of reaching 180° shoulder abduction without pain or effort. Results are summarized in Table 4.

Table 3 Approximated scores and statistical differences between groups on the neck-shoulder impairment scale as reported by Garrel (2020)

Table 4 Proportion capable of reaching 180° shoulder abduction without pain or effort, from Garrell (2020)

Hematoma

None of the included studies reported the occurrence of hematomas as an outcome.

Pain

Hasegawa (2021) measured ‘pain’ using the Neck Dissection Quality of Life Questionnaire. Since this subscale specifically described the outcome of interest, the subscale was reported here. The measurement scale of the instrument was 1 to 5, where higher scores indicated a higher quality of life (i.e. less pain). Table 5 summarized the findings of Hasegawa (20201) at 1, 3, 6, and 12 months postoperatively. Median scores were derived from the reported data in a frequency table.

Table 5 Median scores of the pain subscale of th Neck Dissection Quality of Life Questionnaire, as measured by Hasegawa (2021)

Postoperative oedema

None of the included studies reported the occurrence of postoperative oedema as an outcome.

Costs

None of the included studies reported costs as an outcome.

Level of evidence of the literature

The level of evidence regarding the outcome measure neck recurrence was downgraded by 3 levels because of study limitations (1 level for risk of bias: unclear block size for randomization in one of the RCTs, unclear procedures for random sequence generation, unclear procedures for allocation concealment); number of included patients (2 levels for imprecision: the confidence interval of the pooled estimate crosses both borders of clinical decision-making (0.8 and 1.25)); publication bias was not assessed.

The level of evidence regarding the outcome measure disease-specific survival was downgraded by 2 levels because of study limitations (1 levels for risk of bias: unclear block size for randomization, unclear procedures for random sequence generation, unclear procedures for allocation concealment, no blinding in the RCT); number of included patients (1 level for imprecision: the trial protocol calculated 164 patients per arm in a sample size calculation, this number was not met in the actual inclusion); publication bias was not assessed.

The level of evidence regarding the outcome measure overall survival was downgraded by 2 levels because of study limitations (1 level for risk of bias: unclear block size for randomization in one of the RCTs, unclear procedures for random sequence generation, unclear procedures for allocation concealment); number of included patients (1 level for imprecision: Hasegawa (2021) calculated a sample size for a 12% difference (border of non-inferiority); using the sample size calculation provided in the protocol of Hasegawa (2021) a sample size for a 5% difference was calculated resulting in n=592 participants per arm (D_{between-group}=0.05, Z_{one-sided alpha 0.05}=1.65, Z_{two-sided beta 0.2}=1.28 P_a=0.85, P_b=0.85); SLNB arms consisted of n=134/n=137, ND arms consisted of n=137/n=139); publication bias was not assessed.

The level of evidence regarding the outcome measure shoulder morbidity was downgraded by 3 levels because of study limitations (2 levels for risk of bias: unclear block size for randomization in one of the RCTs, unclear procedures for random sequence generation, unclear procedures for allocation concealment, no blinding in the RCTs, outcomes on the shoulder impairment scale were not reported at 24 months in one RCT); number of included patients (1 level for imprecision: SLNB arms consisted of n=134/n=137, ND arms consisted of n=137/n=139); publication bias was not assessed.

The level of evidence regarding the outcome measure quality of life was downgraded by 3 levels because of study limitations (2 levels for risk of bias: unclear procedures for random sequence generation, unclear procedures for allocation concealment, no blinding in the RCT, missing data on the subscales at 12 months was 12.4% (ND) and 19.4% (SLNB)); number of included patients (1 level for imprecision: SLNB arms consisted of n=134, ND arms consisted of n=137); publication bias was not assessed.

The level of evidence regarding the outcome measure pain was downgraded by 3 levels because of study limitations (2 levels for risk of bias: unclear procedures for random sequence generation, unclear procedures for allocation concealment, no blinding in the RCT, 13.1% (ND) and 19.4%(SLNB) missing data at 12 months); number of included patients (1 level for imprecision: SLNB arms consisted of n=134, ND arms consisted of n=137); publication bias was not assessed.

The level of evidence regarding the outcome measures hematoma, postoperative oedema, and costs could not be GRADEd, since none of the included studies reported this outcome.

A systematic review of the literature was performed to answer the following questions:

What are the (un)beneficial effects of a sentinel lymph node biopsy on neck recurrence, disease specific survival, overall survival, quality of life, shoulder morbidity, hematomas, pain, and postoperative oedema compared to an elective neck dissection in patients with cT1-2N0 oral cavity carcinomas?

P: patients with a cT1-2N0 oral cavity carcinoma;

I: sentinel lymph node biopsy;

C: elective neck dissection;

O: neck recurrence, disease specific survival, overall survival, quality of life, shoulder morbidity, hematomas, pain, and postoperative oedema, costs.

Relevant outcome measures

The guideline development group considered neck recurrence, disease-specific survival, and overall survival as critical outcome measures for decision making; and quality of life, shoulder morbidity, hematomas, pain, and postoperative oedema as important outcome measures for decision making.

A priori, the working group did not define the outcome measures listed above but used the definitions used in the studies.

The working group defined a minimal clinically relevant difference as:

• 0.8 or 1.25 as borders for clinical decision-making for risk or odds ratios of neck recurrence.
• 5% difference or more (absolute) and HR< 0.7 in disease-specific survival.
• 5% difference or more (absolute) and HR< 0.7 in overall survival.
• A difference of 10 points on the quality of life instrument EORTC QLQ-C30 or a difference of a similar magnitude on other quality of life instruments.
• Statistically significant less complications/adverse events.
• Statistically significant better functional outcomes.

Search and select (Methods)

The databases Medline (via OVID) and Embase (via Embase.com) were searched with relevant search terms from 2000 until July 15^th, 2020. The detailed search strategy is depicted under the tab Methods. The systematic literature search resulted in 201 hits. Six systematic reviews were selected and screened for eligibility. Systematic reviews were selected based on the following criteria: concerns patients with a cT1-2N0 oral cavity carcinoma, sentinel lymph node biopsy was compared to (elective) neck dissection, at least one outcome of interest was reported, included randomized trials. The only relevant systematic review (Crocetta, 2019) did not find any randomized studies up to their search date on the 30^th of April in 2019. To update this search, we selected all studies (n=25) published between the 1^st of January 2019 and the 15^th of July 2020 from our initial search. Thirteen of these 25 studies were initially selected based on title and abstract screening, using the following criteria: concerns patients with a cT1-2N0 oral cavity carcinoma, sentinel lymph node biopsy was compared to elective neck dissection, at least one outcome of interest was reported, and the study was a randomized trial. None of the 13 studies met the selection criteria.

On the 1^st of July 2021 the search was updated and the databases Medline (via OVID) and Embase (via Embase.com) were searched with relevant search terms, resulting in 25 new hits. Five studies were selected based on title and abstract. Studies were selected using the aforementioned criteria.

After reading the full text, three studies were excluded and two randomized trials were selected.

Results

Two randomized controlled trials were included. Important study characteristics and results are summarized in the evidence tables. The assessment of the risk of bias is summarized in the risk of bias tables.