Oral cavity squamous cell carcinoma (SCC) is the most common oral cancer, and many treatments have been evaluated.Reference Ettinger, Ganry and Fernandes1,Reference Liviu Feller2 However, the optimal method for evaluating neck nodes after removal of primary oral cavity lesions in patients with early oral cavity SCC (stage T1 or T2) free from lymph node metastasis (N0) remains unclear. Neck node removal prevents clinical recurrence and significantly increases overall survival.Reference Peters, Senft, Hoekstra, Castelijns, Witte and Leemans3,Reference Ho, Kim, Tighiouart, Gudino, Mita and Scher4 Metastasis to the cervical lymph nodes is very important in the prognosis, reducing survival by 50 per cent.Reference Hart, Nasser, Trites, Taylor, Bullock and Barnes5 The risk of occult lymph node metastasis in oral cavity SCC patients of clinical stage N0 is 20–30 per cent.Reference Braams, Pruim, Freling, Nikkels, Roodenburg and Boering6,Reference Ross, Soutar, MacDonald, Shoaib, Camilleri and Robertson7

Traditionally, elective neck dissection was considered for patients with early-stage oral cavity SCC. Several studies reported that this was better than watchful waiting (until metastasis developed).Reference Oh, Phan, Kim, Low, Gupta and Clark8–Reference Alkureishi, Burak, Alvarez, Ballinger, Bilde and Britten10 Elective neck dissection improved survival and reduced the recurrence rate. However, elective neck dissection may be an unnecessarily invasive approach for patients at low risk of lymph node involvement.Reference Oh, Phan, Kim, Low, Gupta and Clark8,Reference Ding, Xiao, Huang, Yuan, Ye and Xuan9 As elective neck dissection can affect shoulder motility and cause persistent pain and scarring, an alternative is desirable.

Sentinel node biopsy represents a compromise between elective neck dissection and watchful waiting and has often been used to accurately detect occult neck node metastases.Reference de Bree and Nieweg11,Reference Glechner, Wöckel, Gartlehner, Thaler, Strobelberger and Griebler12 Sentinel node biopsy involves the injection of a radiotracer or methylene blue dye to identify the lymph nodes that drain first from the primary cancer,Reference Seenu, Suhani, Srivastava, Parshad, Mathur and Kumar13,Reference Keski-Säntti, Kontio, Törnwall, Leivo, Mätzke and Suominen14 and sensitivity and accuracy are high.Reference Kim, Kim, Kim and Hwang15 However, long-term follow-up data are lacking, and the false-positive rate can reach 36 per cent.Reference Kim, Kim, Kim and Hwang15,Reference Yang, Zhou and Wu16 Few reviews or meta-analyses have compared the utility of sentinel node biopsy and elective neck dissection, which we thus address herein. We also performed detailed subgroup analyses by follow-up period.

This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (‘PRISMA’) guidelinesReference Moher, Liberati, Tetzlaff and Altman17 and recommendations for optimising literature searches for systematic surgical reviews.Reference Goossen, Tenckhoff, Probst, Grummich, Mihaljevic and Büchler18 The protocol was prospectively registered on the Open Science Framework (https://osf.io/agzkf/).

Relevant clinical studies were retrieved from PubMed, Scopus, Embase, the Web of Science and the Cochrane Central Register of Controlled Trials up to March 2022. The search terms were as follows: oral carcinoma, oral neoplasm, oral cavity neoplasms, neoplasm, oral cavity, oral cavity cancer, squamous cancer, sentinel node biopsy, elective neck dissection, prognosis, survival, hazard ratio, overall survival rate and disease-free survival. Reference lists were searched to ensure that no relevant studies were missed. Two independent reviewers removed irrelevant studies (i.e. those that did not discuss prognostic factors and survival rates) by reviewing the title, abstract and text.

The inclusion criteria were studies that had: comparison between sentinel node biopsy and elective neck dissection in terms of the prognosis of early stage (T1 or T2) oral cavity SCC patients; survival data and prognostic predictions including hazard ratios with 95 per cent confidence intervals (CIs) and/or overall survival, disease-free survival or disease-specific survival; human studies published in English; and exclusion of advanced oral cavity SCC (clinically confirmed staging (c)T3–4 or N1) patients and those on drugs that might affect oral cavity SCC development.

The exclusion criteria were: reviews; case reports; studies on other head and neck cancers such as nasopharyngeal, oropharyngeal, hypopharyngeal or salivary cancer; and a lack of adequate prognostic data. The search strategy is shown in Figure 1.

Fig. 1. Diagram of the selection of studies for meta-analysis.

All data were extracted by two independent reviewers, who also assessed study quality. Differences were resolved by panel discussion. We recorded the first author, year of publication, country, type of cancer, and number, age, sex and T-stage of the patients. For overall survival, disease-free survival and disease-specific survival, hazard ratios (with 95 per cent CIs) were either describedReference Flach, Bloemena, Klop, van Es, Schepman and Hoekstra19–Reference Seferin, Pinto, Lin, Leite, Gimenes and Dedivitis31 or calculated as described by Tierney et al.Reference Tierney, Stewart, Ghersi, Burdett and Sydes32 and Parmar et al.Reference Parmar, Torri and Stewart33 If both multivariate and univariate analyses were used to evaluate overall survival, the hazard ratios and 95 per cent CIs generated by multivariate analysis were extracted.Reference Zhou, Guo, Zhang, Liu and Dou34 The Risk of Bias in Non-Randomized Studies of Interventions (‘ROBINS-I’) and Cochrane risk of bias tool for randomised trials (‘RoB 2’) were used to assess study quality in line with the Grading of Recommendations, Assessment, Development, and Evaluations guidelines.Reference Sterne, Hernán, Reeves, Savović, Berkman and Viswanathan35,Reference Sterne, Savović, Page, Elbers, Blencowe and Boutron36

R statistical software (R Foundation for Statistical Computing, Vienna, Austria) was utilised for meta-analysis. Homogeneity was assessed using the Q statistic. The degree of heterogeneity was indicated by the I2 value (76–100 per cent = high; 50–75 per cent = moderate; 25–49 per cent = low). Parameters with I2 values less than 50 per cent were analysed using a fixed-effects model, whereas those with I2 values more than 50 per cent were analysed with a random-effects model. Subgroup analyses were performed by follow-up period (3, 5 and 10 years). We used Begg's funnel plots and the Egger linear regression test to evaluate publication bias. For sensitivity analysis, we removed each item individually to assess its contribution to the observed effect.

A total of 12 studies with 10 583 patients were included. Their characteristics and bias assessment results are shown in Tables 1, 2 and 3 in the supplementary material, available on The Journal of Laryngology & Otology website.

Table 1. The characteristics of the included studies

Table 2. Methodological quality of the included studies: Risk of Bias in Non-Randomized Studies of Interventions

Table 3. Methodological quality of the included studies: Risk of Bias 2

We found no significant differences in overall survival (hazard ratio = 1.12; 95 per cent CI, 0.93 to 1.36), disease-free survival (hazard ratio = 1.08; 95 per cent CI, 0.88 to 1.33) or disease-specific survival (hazard ratio = 0.87; 95 per cent CI, 0.65 to 1.15) between sentinel node biopsy and elective neck dissection. Heterogeneity was not detected in analyses of pooled overall survival, disease-free survival and disease-specific survival data (I2 less than 50) (Figure 2). Neither the Egger nor Begg's test showed any publication bias in terms of overall survival (p = 0.84) or disease-specific survival (p = 0.76) (Figure 3a and b). Although mild bias was apparently present in disease-free survival for sentinel node biopsy (p = 0.00; Figure 3c), the Duval and Tweedie trim and fill method showed no significant difference between the observed and adjusted values (hazard ratio, 1.08, p = 0.47 vs 1.04, p = 0.71). Thus, we concluded that the disease-free survival data were not biased. In the sensitivity analyses, the overall survival, disease-free survival and disease-specific survival data did not change on omission of any individual study (Figure 4).

Fig. 2. Forest plots of (a) overall survival, (b) disease-free survival and (c) disease-specific survival. TE = estimated treatment effect; seTE = standard error of treatment estimate; HR = hazard ratio; CI = confidence interval

Fig. 3. Funnel plot of (a) overall survival, (b) disease-free survival, (c) trim filled disease-free survival and (d) disease-specific survival.

Fig. 4. Sensitivity analysis of (a) overall survival, (b) disease-free survival and (c) disease-specific survival. HR = hazard ratio; CI = confidence interval

Most enrolled studies reported survival rates per different follow-up periods from 1 to 10 years and usually at 3, 5 and 10 years; we therefore performed subgroup analyses at these times. In all three subgroups, the survival outcomes (overall survival, disease-free survival and disease-specific survival) were consistently similar between sentinel node biopsy and elective neck dissection.

To the best of our knowledge, this is the largest meta-analysis (including subgroup analyses by follow-up period) to compare sentinel node biopsy with elective neck dissection for early-stage oral cavity SCC patients with clinically N0 necks (no metastases). We found no significant difference in overall survival (hazard ratio = 1.12), disease-free survival (hazard ratio = 0.96), or disease-specific survival (hazard ratio = 1.08). Ding et al. reported similar five-year disease-free survival and overall survival in sentinel node biopsy and elective neck dissection groups in their review article.Reference Ding, Xiao, Huang, Yuan, Ye and Xuan9 However, they included only six prospective studies and did not discuss disease-specific survival. Saleem et al. found no significant difference in disease-free survival or overall survival between patients who underwent sentinel node biopsy and elective neck dissection.Reference Saleem, Peng, Zhu, Wong, Pereira and Tham37 However, unlike our study, their disease-free survival data were heterogeneous. Saleem et al.Reference Saleem, Peng, Zhu, Wong, Pereira and Tham37 included 10 studies in a meta-analysis, but one of those studies (Hiraki et al.Reference Hiraki, Fukuma, Nagata, Shiraishi, Kawahara and Matsuoka38) compared sentinel node biopsy and no neck dissection groups. Recently, Gupta et al. found no significant difference in overall survival between sentinel node biopsy and elective neck dissection groups.Reference Gupta, Maheshwari, Kannan, Nair, Chaturvedi and Agarwal39 Isolated neck nodal and locoregional recurrences were compared, and there was no significant difference; however, only three studies were included.

Five of the studies included in our meta-analysis had high weights because of their large sample sizes.Reference Fan, Zeng, Peng, Guo, Wang and Zhang20,Reference Cramer, Sridharan, Ferris, Duvvuri and Samant24,Reference den Toom, Boeve, Lobeek, Bloemena, Donswijk and de Keizer27–Reference Hasegawa, Tsukahara, Yoshimoto, Miura, Yokoyama and Hirano29 However, subgroup analyses showed minimal heterogeneity. Thus, although the studies varied somewhat in terms of design and quality, the variations did not affect the results, and the survival outcomes of sentinel node biopsy and elective neck dissection were similar. In contrast to Saleem et al., we found no significant difference in overall survival, disease-free survival or disease-specific survival between sentinel node biopsy and elective neck dissection in any study (Figure 2), possibly because our subgroups were defined by follow-up period (3, 5 and 10 years).Reference Ding, Xiao, Huang, Yuan, Ye and Xuan9,Reference Saleem, Peng, Zhu, Wong, Pereira and Tham37,Reference Gupta, Maheshwari, Kannan, Nair, Chaturvedi and Agarwal39,Reference Crocetta, Botti, Pernice, Murri, Castellucci and Menichetti40 Earlier meta-analyses did not perform subgroup analyses by follow-up period. In our analyses, the outcomes of sentinel node biopsy and elective neck dissection were similar. Also, we included three more papers than the largest previous meta-analysis.Reference Saleem, Peng, Zhu, Wong, Pereira and Tham37

Sentinel node biopsy shows if there is a need for neck dissection; patients without neck node metastases can thus avoid unnecessary dissection. Elective neck dissection is associated with minimal complications.Reference Hernando, Villarreal, Alvarez-Marcos, Gallego, García-Consuegra and Junquera41 The sensitivity of sentinel node biopsy for head and neck cancer was reported as 92 per centReference Pitman, Johnson, Brown and Myers42; in another study, it was 82.7 per cent, and the specificity was 98.1 per cent.Reference Kim, Kim, Kim and Hwang15,Reference Yang, Zhou and Wu16,Reference Schilling, Stoeckli, Haerle, Broglie, Huber and Sorensen43 The Sentinel European Node Trial recommended sentinel node biopsy for patients with clinical N0 oral cavity SCC.Reference Schilling, Stoeckli, Haerle, Broglie, Huber and Sorensen43 Sentinel node biopsy is also recommended by the National Comprehensive Cancer Network.Reference Network44 Sentinel node biopsy is also effective in patients with other cancers and is widely used to evaluate breast cancer patients without metastases.Reference Glechner, Wöckel, Gartlehner, Thaler, Strobelberger and Griebler12 However, sentinel node biopsy has certain disadvantages. The node closest to the injection site may be too bright, which can lead to error.Reference Mahieu, de Maar, Nieuwenhuis, Deckers, Moonen and Alic45 Also, radiotracers may reduce sentinel node biopsy effectiveness. In patients with other cancers, the use of methylene blue dye was associated with high false-negative rates. When methylene blue was combined with indocyanine green or a technetium-based radiotracer, sentinel nodes were effectively detected.Reference Kim, Ku, Yang and Byeon46 Sentinel node biopsy accuracy was enhanced by dynamic lymphoscintigraphy and the use of a same-day protocol.Reference Tartaglione, Vigili, Rahimi, Celebrini, Pagan and Lauro47,Reference Nieuwenhuis, Pijpers, Castelijns and Snow48 Sentinel node biopsy tracers require further study.

The main strength of our meta-analysis is that it included 12 studies with 10 583 patients and showed publication bias in terms of overall survival or disease-specific survival (Figure 3). Some bias in terms of the disease-free survival for sentinel node biopsy was initially suggested, but the Duval and Tweedie trim and fill method indicated otherwise. Therefore, we concluded that the studies were not biased and the data are thus clinically relevant.

Our meta-analysis had some limitations. First, most of the included studies were non-randomised, so patient characteristics and tumour subsites may have differed among the groups, which would have slightly affected the results. However, few survival studies are randomised.Reference Djurisic, Rath, Gaber, Garattini, Bertele and Ngwabyt49 Garrel et al. included mainly early stage oral cavity SCC patients with no clinical lymph node metastases, but also a small number of oropharyngeal cancer patients (less than 13 per cent).Reference Garrel, Poissonnet, Moyà Plana, Fakhry, Dolivet and Lallemant28 Patients with human papillomavirus-positive oropharyngeal cancer have good prognoses and should be analysed separately.Reference Mallen-St Clair, Alani, Wang and Srivatsan50 Sundaram and Subramanyam included mainly T1 and T2 patients but also a small number of T3 patients (less than 13 per cent).Reference Sundaram and Subramanyam26 Second, the tracers differed among the included studies. Several studies used tracers detectable by single-photon emission computed tomography–computed tomography,Reference Carvalho GMd, Kohler, Guimarães and Crespo22,Reference Moya-Plana, Aupérin, Guerlain, Gorphe, Casiraghi and Mamelle25,Reference den Toom, Boeve, Lobeek, Bloemena, Donswijk and de Keizer27,Reference Seferin, Pinto, Lin, Leite, Gimenes and Dedivitis31 but Hasegawa et al. used a tracer evident on single-photon emission–computed tomography.Reference Hasegawa, Tsukahara, Yoshimoto, Miura, Yokoyama and Hirano29 Third, we lacked data on smoking history and co-morbidities, which may affect survival.

Our meta-analysis addressed the problems of previous meta-analyses and also included more studies. The overall survival, disease-free survival and disease-specific survival did not differ significantly between early-stage oral cavity SCC patients without clinical neck metastases undergoing sentinel node biopsy and elective neck dissection. Sentinel node biopsy, which has no life-threatening side effects, is preferable to elective neck dissection for patients with early stage clinically node-negative oral cavity SCC.