Research Article: Diagnostic Efficacy of Sentinel Lymph Node Biopsy in Early Oral Squamous Cell Carcinoma: A Meta-Analysis of 66 Studies

Date Published: January 20, 2017

Publisher: Public Library of Science

Author(s): Muyuan Liu, Steven J. Wang, Xihong Yang, Hanwei Peng, Gang Han.


The diagnostic efficacy of sentinel lymph node biopsy(SLNB) in early oral squamous cell carcinoma(OSCC) still remains controversial. This meta-analysis was conducted to assess the diagnostic value of SLNB in clinically neck-negative T1-2 OSCC.

A systematic literature search for relevant literature published up to September 11, 2016 was conducted in PubMed, Embase, Web of Science, Cochrane Library and ClinicalTrials, and the reference lists of eligible studies were examined. Data from different studies were pooled to estimate the summary sentinel lymph node(SLN) identification rate, sensitivity, negative predictive value. Summary receiver operator characteristic curve(SROC) was plotted and area under the SROC curve (AUC) was calculated to evaluate the overall diagnostic efficacy. Threshold effect was assessed with use of the spearman correlation coefficient. Between-study heterogeneity was tested using the Q tests and the I2 statistics. Subgroup analyses were conducted in view of the greater effect of different study characteristics on diagnostic efficacy of SLN. Deeks’ funnel plot asymmetry test was performed to evaluate publication bias. Sensitivity analysis was evaluated through omitting studies one by one and comparing the pooled results of random-effects model and fixed-effects model. All analyses were performed using Review Manager (version 5.3.5), Meta-DiSc (version 1.4), Comprehensive Meta Analysis (version 2.0) and STATA (version 12).

66 studies comprising 3566 patients with cT1-2N0 OSCC were included in this meta-analysis. The pooled SLN identification rate was 96.3%(95% CI: 95.3%-97.0%). The pooled sensitivity was 0.87 (95% CI: 0.85–0.89), pooled negative predictive value was 0.94 (95% CI: 0.93–0.95), and AUC was 0.98 (95% CI: 0.97–0.99). Subgroup analyses indicated that SLN assessment with immunohistochemistry(IHC) achieved a significantly higher sensitivity than without IHC.

This meta-analysis suggests that SLNB has a high diagnostic accuracy in cT1-2N0 oral squamous cell carcinoma, and is an ideal alternative to elective neck dissection. Furthermore, the use of IHC can significantly improve SLNB diagnostic sensitivity for early OSCC.

Partial Text

Oral squamous cell carcinoma (OSCC) is one of the most common types of cancer in the world, with a considerable incidence of new cases every year. Approximately 50% of the patients with OSCC present with early stage disease(cT1-2N0) [1]. The main prognostic factor is occult lymph node metastasis in the neck. As it was reported in previous literatures, the overall rate of occult lymph node metastasis is 20%-30% in early stage OSCC patients [2–4]. Therefore, elective neck dissection remains the gold standard treatment in many institutions, resulting in overtreatment in over 70% of early OSCC patients and a considerable morbidity. For this reason, in recent years, sentinel lymph node biopsy(SLNB) has become more important and popular in the cervical treatment of patients with early OSCC. The sentinel lymph node (SLN) procedure is based on the theory that flow from a primary tumor travels sequentially to the sentinel lymph node and subsequently to the remaining lymph node basin [5]. Compared to elective neck dissection, SLNB is less invasive, cost-effective and beneficial to patient quality of life [6–9]. But the diagnostic efficacy of SLNB in early OSCC remains controversial [10–12]. Furthermore, most previous individual studies contained too small of a sample size to yield a valid conclusion. In addition, previous meta-analyses mainly focused on head and neck cancer or oral and oropharyngeal carcinoma [13–16]. However, combining different subset of head and neck cancer with differing clinical characteristics and metastasis patterns, can lead to heterogeneous results for SLNB. Although some previous meta-analyses have conducted subgroup analysis on OSCC, the small included sample size was underpowered to yield credible pooled findings. In recent years, many high quality prospective and some multi-institutional studies on the diagnostic efficacy of SLNB in early OSCC have been published [17–19]. Therefore, we performed a meta-analysis to summarize the diagnostic efficacy of SLNB specially focused exclusively on early OSCC. Additionally, we further stratified results by different clinical and study characteristics in order to explore the potential factors that may affect the diagnostic accuracy and applicability of SLNB.

To our knowledge, this is the largest meta-analysis focused on the diagnostic efficacy of sentinel lymph node biopsy in early oral squamous cell carcinoma. In this meta-analysis of 66 studies comprising more than 3500 patients, SLNB yielded a pooled identification rate of 96.3%(95% CI: 95.3%-97.0%), a pooled sensitivity of 0.87(95%CI: 0.85–0.89), a pooled negative predictive value of 0.94 (95% CI: 0.93–0.95) and an AUC of 0.98 (95% CI: 0.97–0.99). The high pooled negative predictive value implied that only 6% of SLN-negative early oral cavity cancer patients would result in a false-negative regional recurrence during follow-up. This is similar to the regional recurrence rate after elective neck dissection in clinically neck-negative early OSCC reported by previous literature [80], and is far lower than the acceptable threshold of 20% cervical lymph node metastasis rate for prophylactic neck dissection. Therefore, elective neck dissection could be omitted in SLN-negative early OSCC patients. Moreover, the pooled sensitivity implies that 87% of occult cervical lymph node metastases could be diagnosed by SLNB and the false-negative rate is 13%. The occult lymph node metastasis rate has been reported to be 20%-30% for cT1-2N0 OSCC [2–4]. Therefore, we can estimate that SLNB applied to all early OSCC patients would result in a 2.6%-3.9% regional recurrence rate. This regional recurrence rate is acceptable when considering the serious complications and 70% overtreatment rate in traditional prophylactic neck dissection procedure. Overall, these pooled findings indicated that SLNB had an ideal diagnostic accuracy for predicting occult cervical lymph node metastases in early oral cancer patients and was an ideal alternative to neck dissection. In the previous meta-analyses focusing on the diagnostic efficacy of SLNB in head and neck cancer or oral/oropharyngeal cancer, Tim reported a pooled sensitivity of 0.92 (95%CI: 0.86–0.95) in oral cancer subgroup(n = 508), while Thompson reported a pooled sensitivity and negative predictive value of 0.94 (95%CI: 0.89–0.98) and 0.96 (95%CI: 0.93–0.99) respectively in the subset of oral cavity tumors(n = 631) [14, 15]. Compared to these previous meta-analyses, our research found a lower sensitivity of 0.87(95%CI: 0.85–0.89)(n = 3506). Since those two meta-analyses were published many years ago, we further stratified our results by publication year and found that the pooled sensitivity of early publications(2000–2008) in current meta-analysis was 0.92(95%CI: 0.87–0.95), more similar to the results reported by previous meta-analyses, and better than late publications(2009–2016). A possible reason for this difference may be that SLNB researches in early publications were still during the validation stage, and elective neck dissection of levels I-III was the gold standard for SLN-negative cases in most of these publications(69.2%, 18/26). But in more recent publications, most SLNB research studies use clinical follow-up as their gold standard for SLN-negative cases and only 35%(14/40) of studies were still using elective neck dissection(levels I-III) as their gold standard. Thus, we speculate that: (1) there may have occult lymph node metastases in level IV, level V or even contralateral neck that would be missed by the elective neck dissections in most of the earlier publications, resulting in an overestimated sensitivity; (2) SLNB with neck dissection is definitely easier than SLNB without neck dissection and this may also lead to a higher pooled sensitivity in the validation stage.

Our results confirmed that SLNB had a high diagnostic accuracy in cT1-2N0 oral squamous cell carcinoma, and was an ideal alternative to elective neck dissection. We also found that H&E with IHC yielded much better diagnostic sensitivity than H&E alone. However, further clinical trials are required to verify the clinical utility and application of SLNB by frozen section but not by postoperative pathological assessment. In particular, further studies on the diagnostic accuracy of automated quantitative real-time PCR (qRT-PCR) assay for intra-operative SLN frozen section are required [81, 82].