This article provides a critical review of the evidence surrounding the management of the clinical node-negative patient with early-stage oral cavity squamous cell carcinoma.
The following points list the level of evidence as based on Oxford Center for Evidence-Based Medicine. Additional critical points are provided and points here are expanded at the conclusion of this article.
- •
The presence of lymph node metastases in oral cavity squamous cell carcinoma (OCSCC) continues to be one of the most important prognostic factors. In clinical node-negative (cN0) early-stage OCSCC, the prevalence of occult nodal disease ranges from 18% to 30% for T1 lesions and 24% to 53% for T2 tumors.
- •
Preoperative factors, including characteristics of the primary tumor, histopathologic features, and preoperative imaging, can help adjust the estimated risk of nodal disease.
- •
The most appropriate management strategy for dealing with the cN0 neck remains controversial, with observation, elective neck dissection, and sentinel lymph node biopsy reported as potential management strategies. There is insufficient evidence to recommend any single management strategy over another (level of evidence 1a-; grade D).
- •
The current literature is hampered by inadequately powered studies regarding the role of elective neck dissection in early-stage clinically node-negative OCSCC.
Overview
Spread to the cervical lymphatics continues to be one of the most important prognostic factors in patients with oral cavity squamous cell carcinoma (OCSCC), with a reduction in survival of at least 30%. As such, the presence of lymphatic spread is an integral consideration when deciding on the use of adjuvant therapy. In patients with a clinically negative neck (ie, those with no evidence of cervical lymphatic spread by physical examination and imaging studies), there is continued controversy as to the most appropriate management strategy.
Elective treatment with neck dissection, sentinel lymph node biopsy, radiation, and observation are proposed strategies for managing the clinically node-negative (cN0) neck. The debate surrounding these management choices has been longstanding, with recent reinvigoration from the increasing interest in transoral surgical approaches and the avoidance of external incisions in the management of early-stage head and neck malignancies.
Despite several decades of intensive debate in the medical literature, the role of elective neck dissection (END) in patients with cN0 OCSCC remains controversial. The conflicting conclusions of a large number of heterogeneous retrospective studies and the lack of large-scale, adequately powered prospective studies contribute to the confusion surrounding this topic. This article provides a critical review of the evidence surrounding the management of the cN0 patient with early-stage OCSCC.
Evidence-based clinical assessment
Choosing an Appropriate Therapeutic Threshold
Before any clinical assessment of the risk of lymphatic metastases, the physician should establish an appropriate treatment threshold beyond which the potential benefits of treatment outweigh the morbidity. This a priori defined threshold forms the backdrop that frames all management decisions regarding the risk of lymphatic metastases. Although little controversy exists at the extremes of risk, the benefit becomes less clear in cases with intermediate risk of occult nodal disease. Having a defined level at which the potential benefits of elective treatment outweigh the risks provides a rational approach to the application of treatment.
Given the fundamental importance of defining a treatment threshold in interpreting estimates of the risk of occult nodal disease and the subsequent application of treatment, it is surprising that little attention has been paid to this question in the literature. In their often-quoted paper, Weiss and colleagues used a decision tree analysis to derive a treatment threshold of 20%. This analysis was based on the risk of nodal disease, the effectiveness of primary and salvage surgery, and a subjective assignment of the usefulness of treatment outcomes. Based on this analysis, they argued that patients with a risk of nodal metastases greater than 20% would benefit from END. Although many continue to use a 20% risk of nodal metastases as a general guideline for performing END, the clinical data used to generate this threshold are based on historical series that may not reflect contemporary treatment outcomes. More recent publications have recommended treatment thresholds ranging from 17% to 40%. Given the lack of high-quality evidence surrounding any of these treatment thresholds, they should be interpreted with caution. Such thresholds should not be interpreted as absolute values that define when treatment is appropriate. Rather, they should serve to emphasize that any interpretation of the risk of nodal metastases needs to be done in the framework of a predefined threshold for action.
Assessing the Risk of Occult Nodal Disease
Assessment of the primary site
Once a threshold for END has been defined, an initial assessment of the risk of nodal disease is based on the site and classification of the primary tumor (T classification). Subsequent evaluation of specific tumor characteristics, including depth of invasion, histologic tumor grade, and presence of perineural and/or lymphovascular invasion, can be used to adjust the probability estimates of occult nodal disease.
Rates of regional metastatic spread differ by oral cavity subsite. Squamous cell carcinoma involving the tongue is the most well-studied subsite of the oral cavity. Early-stage tongue carcinomas have higher rates of metastatic spread than floor of mouth carcinomas. Comparisons with other oral cavity subsites are limited.
The probability of occult nodal disease based on the clinical T classification has been estimated in multiple END studies in early OCSCC ( Table 1 ). Differences in patient populations, subsite distribution, extent of dissection, and methods of histologic node analysis between these studies make comparative estimates of the prevalence of occult nodal disease difficult. In END series in which elective treatment of the neck was applied universally, the prevalence of occult positive nodes ranges from 6% to 25% for T1 OCSCC, whereas the prevalence for T2 OCSCC ranges from 20% to 32%. When the studies comparing observation and neck dissection are included ( Table 2 ), the prevalence of occult node disease in early OCSCC can approach 40% to 50%, although these numbers may be inflated because of selection bias.
| Investigators | n | T1 (%) | T2 (%) |
|---|---|---|---|
| Iype et al | 172 | 25.4 | 29.4 |
| Thiele et al | 122 | 5.9 | 19.7 |
| Civantos et al | 140 | 25.0 | 32.0 |
| Investigators | n | Population | % Occult N+ (END) | % Delayed N+ (Obs) | % Salvaged | % Survival for END, Obs (y) | Recommendation |
|---|---|---|---|---|---|---|---|
| Ebrahimi et al | 153 | T1–T2 OC | 37 | 39 | 21 | 92, 69 (5) | END |
| Dias et al | 49 | T1 OT, FOM | 21 | 28 | 38 | 97, 74 (3) | END |
| Haddadin et al | 137 | T1–T2 OT | 38 | 41 | 35 | 80.5, 53.6 (5) | END |
| Capote et al | 154 | T1–T2 | 0 | 27 | 32 | 92.5, 71.4 (5) | END |
| Cunningham et al | 54 | T1–T2 OT, FOM | 14 | 42 | 56 | 88, 77 (3) | END |
| Lydiatt et al | 156 | T1–T2 OT | 20 | 17 | 50 | 55, 33 | END |
| Duvvuri et al | 359 | T1–T2 OC, OP | 23 | 27 | N/A | 73, 58 | END |
| Keski-Santti et al | 80 | T1–T2 OT | 34 | 44 | 33 | 82, 77 (5) | END |
| Franceschi et al | 149 | T1–T2 OT | 41 | 26 | 41 | 62, 63 (5) | Obs |
| O’Brien et al | 162 | T1–T4 OC, OP | 30 | 9 | 80 | 86, 94 (3) | Obs |
| Khafif et al | 590 | T1–T4 OC, OP, L | 41 | 16 | 49 | 56, 49 | Obs |
| D’Cruz et al | 359 | T1–T2, OT | 20 | 47 | 46 | 74, 68 (5) | Obs |
| Layland et al | 621 | T1–T4 OC | 12 | N/A | 31 | 54, 58 | Obs |
| Liu et al | 131 | T1 OT | 24 | 23 | N/A | 80, 74 (4) | Obs |
Although location and clinical T classification can be obtained from preoperative examination alone, other factors that have been reported to affect the prevalence of occult nodal disease require further diagnostic testing. These factors include histopathologic details such as tumor grade, lymphovascular invasion, perineural spread, and depth of invasion.
The depth of invasion of the primary tumor, particularly for oral tongue squamous cell carcinoma, has been reported to significantly influence the prevalence of occult node positivity. A depth of invasion greater than 4 mm has been associated with an increased risk of occult node metastasis. Asakage and colleagues studied 44 patients with T1 and T2 squamous cell carcinoma of the oral tongue who were treated with partial glossectomy and observation of the neck. At 5 years, 21 of the 44 had developed neck metastases. On multivariate analysis, only the depth of invasion predicted subsequent cervical node metastases with a relative risk ratio of 9.4 (95% confidence interval [CI] 1.5–57.7) for lesions 4 mm or greater in thickness. Additional retrospective studies noted an association between increased tumor thickness and an increased risk of occult node disease.
The importance of a tumor depth 4 mm or greater has been validated as an important predictor of occult node metastasis in prospective randomized controlled trials evaluating END versus observation for early-stage OSCC. In a study comparing hemiglossectomy alone or with elective radical neck dissection for early-stage oral tongue squamous cell carcinoma, Fakih and colleagues showed an increased rate of occult nodal disease (67% vs 8%; P <.01) for tumors with a depth of 4 mm or greater in the END arm of the study. In the observation arm, they showed a corresponding increased rate of delayed regional node metastases in tumors with a depth of 4 mm or greater (76% vs 22%; P <.01). Similar trends were noted in a prospective study by Kligerman and colleagues that compared resection alone versus resection with selective neck dissection (levels I–III) for early-stage OCSCC. The investigators observed a trend toward an increase in occult node positivity with tumors greater than 4 mm in depth (30% vs 7%; P = .11).
Because of these findings, many have recommended END when the depth of invasion exceeds 4 mm. Although the evidence strongly suggests that an increase in the depth of the primary tumor is significantly associated with an increased risk of node positivity, interpretations of the degree to which it influences this risk and the best cutoff value in tumor thickness are clouded by significant heterogeneity in the study populations, extent of node dissection, and techniques for measuring tumor thickness.
Perhaps the greatest limitation to the use of the 4-mm tumor thickness cutoff is the difficulty in obtaining this information before it is needed. Although staged END can be performed after treatment of the primary tumor and pathologic analysis of the specimen, many surgeons prefer to address the primary tumor and regional lymphatics at the same surgical setting. A preoperative biopsy can provide an estimate of tumor thickness but may be subject to sampling error. To overcome these limitations, Taylor and colleagues reported using ultrasound (US) to assess the depth of the primary lesion. In a consecutive series of 21 patients with oral tongue and floor of mouth squamous cell carcinoma, the investigators noted a high concordance between pathologic tumor thickness and preoperative US estimation of thickness. However, the results of this study have yet to be replicated in a large series of patients with OCSCC and therefore should be interpreted with caution.
Other histopathologic parameters, such as lymphovascular invasion, tumor grade, and perineural spread, have less consistent relationships with the risk of cervical node metastases than either clinical tumor classification or thickness. As such, the usefulness of these parameters to adjust predictions regarding the risk of occult node disease is not clear.
Assessment of the neck
The sensitivity of physical examination for the detection of cervical node metastases is reported to range from 60% to 80%, which is less than the useful level for decision making.
Imaging
A variety of imaging techniques can provide increased sensitivity and specificity for the detection of regional node metastases. US, magnetic resonance imaging (MRI), computed tomography (CT), photon emission tomography (PET), and PET-CT/MRI fusion have all been examined in the context of the N0 neck, with varying accuracy.
The usefulness of these imaging studies lies, in part, in their ability to shift the probability or risk of nodal metastasis past (either above or below) the predetermined threshold for treatment. The decision to obtain imaging of the neck should consequently be made in light of the estimated probability of occult disease based on the initial examination of the primary lesion. In some patients, the initial risk of nodal metastases may exceed a threshold at which testing for the sake of determining whether END is necessary is unwarranted. For instance, a large, deeply invasive T2 tongue carcinoma may have an estimated risk of nodal metastases that is high enough to warrant END regardless of the findings of any imaging study. In this case, imaging may provide valuable additional information about the primary tumor or potentially the extent of dissection, but would not alter the decision of whether or not treatment is needed.
Several prospective studies have compared the performance of various imaging modalities in the examination of the neck in patients with OCSCC: Stuckensen and colleagues compared the ability of PET, MRI, CT, and US to detect cervical node metastases in 106 patients with OCSCC who underwent END. Using pathologic assessment as the gold standard: PET had the greatest accuracy: sensitivity 70%, specificity 82%; US: 84%, 68%; CT: 66%, 74%; MRI: 64%, 69%. In a study of 463 patients with OCSCC, Liao and colleagues reported that preoperative 18F-fluorodeoxyglucose (FDG) PET had a sensitivity and specificity of 77.7% and 58.0%, respectively. When examination is restricted to patients with clinically N0 disease, the reported sensitivities and specificities decline markedly. Ng and colleagues prospectively compared the performance of CT, MRI, and PET in 134 patients with OCSCC without palpable adenopathy. Overall, 35 (26.1%) had neck metastases. The best sensitivity and specificity for the detection of occult nodal disease was with PET imaging visually correlated to CT/MRI (57.1% and 96%, respectively) compared with PET alone (51.4% and 91.9%) or CT/MRI alone (31.4% and 91.9%). Based on these data, the posttest probability of occult nodal metastasis with a negative PET correlated with CT/MRI was 3.3% in T1 tumors and 9.2% in T2 tumors.
Kyzas and colleagues performed a meta-analysis examining the diagnostic performance of 18F-FDG PET in patients with head and neck squamous cell carcinoma. For the cN0 subpopulation, the investigators found 10 studies with 311 patients for analysis. For this subpopulation, the sensitivity of PET was only 50% (95% CI 37%–63%), whereas specificity was 87% (95% CI 76%–93%). The positive likelihood ratio was 3.83 (95% CI 1.90–7.75) and the negative likelihood ratio was 0.57 (95% CI 0.43–0.77).
This means that, with an estimated pretest probability of occult nodal metastasis of 20%, the posttest probability with a positive PET scan shifts to 49%. The posttest probability with a negative PET scan is 12%, which is less than the commonly cited threshold of 20% for consideration of END. Therefore, a PET scan may be beneficial in the initial evaluation of the cN0 neck, depending on an individual’s treatment threshold.
Evidence-based clinical assessment
Choosing an Appropriate Therapeutic Threshold
Before any clinical assessment of the risk of lymphatic metastases, the physician should establish an appropriate treatment threshold beyond which the potential benefits of treatment outweigh the morbidity. This a priori defined threshold forms the backdrop that frames all management decisions regarding the risk of lymphatic metastases. Although little controversy exists at the extremes of risk, the benefit becomes less clear in cases with intermediate risk of occult nodal disease. Having a defined level at which the potential benefits of elective treatment outweigh the risks provides a rational approach to the application of treatment.
Given the fundamental importance of defining a treatment threshold in interpreting estimates of the risk of occult nodal disease and the subsequent application of treatment, it is surprising that little attention has been paid to this question in the literature. In their often-quoted paper, Weiss and colleagues used a decision tree analysis to derive a treatment threshold of 20%. This analysis was based on the risk of nodal disease, the effectiveness of primary and salvage surgery, and a subjective assignment of the usefulness of treatment outcomes. Based on this analysis, they argued that patients with a risk of nodal metastases greater than 20% would benefit from END. Although many continue to use a 20% risk of nodal metastases as a general guideline for performing END, the clinical data used to generate this threshold are based on historical series that may not reflect contemporary treatment outcomes. More recent publications have recommended treatment thresholds ranging from 17% to 40%. Given the lack of high-quality evidence surrounding any of these treatment thresholds, they should be interpreted with caution. Such thresholds should not be interpreted as absolute values that define when treatment is appropriate. Rather, they should serve to emphasize that any interpretation of the risk of nodal metastases needs to be done in the framework of a predefined threshold for action.
Assessing the Risk of Occult Nodal Disease
Assessment of the primary site
Once a threshold for END has been defined, an initial assessment of the risk of nodal disease is based on the site and classification of the primary tumor (T classification). Subsequent evaluation of specific tumor characteristics, including depth of invasion, histologic tumor grade, and presence of perineural and/or lymphovascular invasion, can be used to adjust the probability estimates of occult nodal disease.
Rates of regional metastatic spread differ by oral cavity subsite. Squamous cell carcinoma involving the tongue is the most well-studied subsite of the oral cavity. Early-stage tongue carcinomas have higher rates of metastatic spread than floor of mouth carcinomas. Comparisons with other oral cavity subsites are limited.
The probability of occult nodal disease based on the clinical T classification has been estimated in multiple END studies in early OCSCC ( Table 1 ). Differences in patient populations, subsite distribution, extent of dissection, and methods of histologic node analysis between these studies make comparative estimates of the prevalence of occult nodal disease difficult. In END series in which elective treatment of the neck was applied universally, the prevalence of occult positive nodes ranges from 6% to 25% for T1 OCSCC, whereas the prevalence for T2 OCSCC ranges from 20% to 32%. When the studies comparing observation and neck dissection are included ( Table 2 ), the prevalence of occult node disease in early OCSCC can approach 40% to 50%, although these numbers may be inflated because of selection bias.
