CLINICAL REVIEW

David W. Eisele, MD, Section Editor

Analysis of sentinel node biopsy combined with other diagnostic tools in staging cN0
head and neck cancer: A diagnostic meta-analysis
Li-Jen Liao, MD, PhD,1,2 Wan-Lun Hsu, PhD,3 Chi-Te Wang, MD, PhD,1,2 Wu-Chia Lo, MD,2 Mei-Shu Lai, MD, PhD,1,4*
1

Graduate Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan, 2Department of Otolaryngology, Far Eastern
Memorial Hospital, Taipei, Taiwan, 3Genomics Research Center, Academia Sinica, Taipei, Taiwan, 4Center of Comparative Effectiveness Research, National Center of Excellence
for Clinical Trial and Research, National Taiwan University Hospital, Taipei, Taiwan.

Accepted 6 December 2014

Published online 26 June 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/hed.23945

ABSTRACT: Background. The purpose of this was to find a staging

strategy sensitive enough to reduce the risk of occult metastases in cN0
head and neck cancer to below 15% to 20%.
Methods. A total of 73 articles were selected for analysis of the diagnostic performance in staging cN0 head and neck cancer. Hypothetical estimation of negative predictive value (NPV) was calculated based on the
Bayesian theory.
Results. The pooled estimates for sensitivity were 56.4% and 84.9% for
ultrasound-guided fine-needle aspiration (FNA) and sentinel node biopsy
(SNB). The pooled estimates for sensitivity were 47.0%, 56.6%, 48.3%,
and 63.3% for CT, MRI, positron emission tomography (PET), and ultra-

INTRODUCTION
Lymph node status is one of the most important predictors of a poor prognosis in head and neck cancer.1 For
patients with a clinically negative (cN0) neck, there are 2
major management strategies, which include elective neck
dissection (END) and watchful waiting. Cervical lymph
node metastasis staged by palpation has been demonstrated to be inaccurate; the rate of occult cervical nodal
metastases is at least 30% by simple palpation.2 Currently, the National Cancer Comprehensive Networks
practice guidelines3 recommend END for clinical N0 cancer of the oral cavity, oropharynx, hypopharynx, and
supraglottic larynx.
The dilemma in the current clinical management of the
neck is the choice between the possible undertreatment of
30% to 40% of patients with occult metastases and overtreatment of the remaining 60% to 70%. Personalized
management of the cN0 neck, especially in patients with
early head and neck squamous cell carcinoma (SCC),
would benefit greatly from staging techniques that

*Corresponding author: M.-S. Lai, Graduate Institute of Epidemiology and

Preventive Medicine, College of Public Health, National Taiwan University,
Taipei, Taiwan. E-mail addresses: mslai@ntu.edu.tw
Contract grant sponsor: This work was supported by the National Science
Council of the Republic of China (Grant NSC101-2314-B-418-002) and grants
from the Far Eastern Memorial Hospital (FEMH-2013-C-012).

628

HEAD & NECKDOI 10.1002/HED

APRIL 2016

sound, respectively. The pooled estimates for specificity were 88.9%,

82.5%, 86.2%, and 79.1% for CT, MRI, PET, and ultrasound. In estimation, the CT or MRI with SNB strategies had NPV higher than 85% even
when the pretest metastatic rate was 60%.
Conclusion. The SNB procedure has the best performance. A combination of CT/MRI and SNB for cN0 head and neck cancer is preferred.
C 2015 Wiley Periodicals, Inc. Head Neck 38: 628634, 2016
V

KEY WORDS: sentinel node biopsy, head and neck carcinoma, N0

neck, meta-analysis, review

increase the accuracy of the assessment of nodal disease,

particularly when these methods are minimally or not at
all dependent on the size of the metastasis. To avoid the
unnecessary treatment of histologically negative necks, a
staging technique must be sensitive enough to reduce the
risk of occult metastases to <15% to 20%, which means
a negative predictive value (NPV) of >80% to 85%.4
With the development of modern imaging modalities,
the American Joint Committee on Cancer has stated that
clinical staging should include physical examination as
well as the results of other imaging modalities.5 Research
is now directed toward finding a staging method sensitive
enough to bring the risk of occult metastases below 15%
to 20%. Previous meta-analyses6 compared the diagnostic
accuracy of different imaging modalities in neck node
evaluation and showed that ultrasound-guided fine-needle
aspiration (FNA) was the most accurate imaging modality
compared to CT, MRI, and ultrasound. However, these
studies examined a mixture of patients diagnosed with
cN1 and cN0, and no study has focused only on patients
with cN0 necks. Although most authors agree that the
combination of several diagnostic tests provides a significant improvement in the accuracy of detecting metastatic
disease,7 the combination of these diagnostic tools has
not been evaluated in detail.
The purpose of this study was to complete a systematic
review and meta-analysis of the performance of different
imaging modalities (ie, CT, MRI, positron emission
tomography [PET], ultrasound, ultrasound-guided FNA,

SENTINEL

NODE BIOPSY COMBINED WITH OTHER DIAGNOSTIC TOOLS IN STAGING CN0 HEAD AND NECK CANCER

and sentinel node biopsy [SNB]), in the evaluation of

neck lymph node metastasis in patients with cN0 head
and neck cancer. Analyses of combined diagnostic tools
in staging cN0 head and neck cancer could also be
simulated.

MATERIALS AND METHODS

Systematic review
All subsites of cN0 head and neck SCC,
including the oral cavity, oropharynx, hypopharynx, and
larynx, were included in this study. Nasopharyngeal cancer, thyroid cancer, and salivary gland cancer were
excluded from this study.
Population.

We included different diagnostic modalities

(ie, CT, MRI, PET, ultrasound, ultrasound-guided FNA,
SNB, and gene-expression profiling) in the evaluation of
neck lymph node metastasis in patients with clinical head
and neck cancer.

Index tests.

Inclusion criteria.
There were 2 major study groups
according to the clinical node staging of the patient population in each study. The first group included studies of
patients who had pathologically positive head and neck
cancer and clinically negative cervical lymph nodes (cN0)
before the imaging examination. The second group
included studies with mixed-patient populations (head and
neck malignancy with both cN0 and cN1). We included
studies with individual patient data available for cN0.
Based on the full text reports, studies were selected if
they fulfilled all of the following inclusion criteria: (1)
histopathology findings for neck dissection (specimens
obtained at surgery) or sufficient follow-up time were
used as the reference standard; (2) the primary tumor and
lymph node metastases were SCCs; and (3) sufficient
data were presented to construct a 2 3 2 contingency
table (sensitivity and/or specificity with absolute numbers
of false positive, false negative, true positive, and true
negative findings) for the imaging modalities compared
using the reference standard.

Studies were excluded if raw data were

not presented. We also excluded studies with both
patients diagnosed with cN0 and cN-positive when individual data for cN0 patients were not available and a 2 3
2 contingency table could not be made because these
studies enrolled both patients with cN0 and cN1. Studies
with <10 cases were also excluded.

Exclusion criteria.

PubMed and CENTRAL database searching

PubMed (up to August 2013) and CENTRAL (via the
Cochrane database up to August 2013) were searched
using the following keywords: (a) (head and neck neoplasms [MH]) NOT (thyroid neoplasms [MH] OR esophageal neoplasms [MH] OR nasopharyngeal neoplasms
[MH] OR salivary gland neoplasms [MH] OR melanoma
[MH] OR parathyroid neoplasms [MH]); (b) diagnostic
imaging [MH]; (c) lymph node [TW] OR neck node
[TW]; and (d) sensitivity [TW] OR accuracy [TW].
Human studies with abstracts in English were included.

Reviews, letters to the author, comments, and case reports

were excluded.
From the selected abstracts, 2 reviewers (W.C.L. and
C.T.W.) separately screened the full text of these potentially eligible articles in which CT, MRI, PET, ultrasound,
and SNB were utilized. Reference lists were manually
screened for additional relevant articles.

Quality assessment
The Quality Assessment of Diagnostic Accuracy
Studies8 quality assessment tool was used to evaluate the
relevant study design characteristics of each study. This
tool and the definitions of the characteristics have previously been fully described.8 We assigned a design characteristic with a score of 1 if the evaluation criteria were
met or 0 if the design characteristic was not preset or
was unclear. Each study that met the inclusion criteria
was analyzed by 2 independent reviewers. When there
was a discrepancy between the reviewers, a consensus
reviewer (L.J.L.) resolved these differences.

Meta-analysis and simulation

Because there was no false positive in ultrasoundguided FNA and SNB, the specificity is 100%. Univariate
meta-analyses with random-effects approach were performed to calculate the combined sensitivities of
ultrasound-guided FNA and SNB. The random-effects
approach allows for interstudy heterogeneity. For other
diagnostic tools (CT, MRI, ultrasound, and PET), the
threshold effect has an impact on the sensitivity and specificity. The interpretation of diagnostic accuracy should
take both sensitivity and specificity into consideration.
Therefore, univariate meta- analyses are not appropriate
for these tools. Pooled sensitivities and specificities of
CT, MRI, ultrasound, and PET in neck lymph node
metastasis of a cN0 neck from individual studies were
calculated using a bivariate random-effects model.9 The
overall performance was estimated with the area under
summary receiver operating characteristic curve (AUC),
which was fitted using Moses model,10,11 adding 1/2 to
all cells of the studies with a zero value. To avoid publication bias and a learning curve effect,12 we excluded
studies with a case number <10.
NPV was used to evaluate the risk of occult metastases.
The formula is shown as follows:
Negative predictive value 5 specificity*(1-prevalence)/
[specificity*(1- prevalence 1 (1-sensitivity)* prevalence].
To increase the NPV, serial tests combined with multiple tests are proposed. Sequential tests combined with
multiple tests are proposed and simulated for cN0 neck
evaluation. A scenario of cN0 diagnosed by palpation
received serial tests. Test 1 should be routinely performed
with CT/MRI; if test 1 reveals a negative result, the
patient will receive a second examination (ultrasound,
ultrasound-guided FNA, PET, or SNB). Patients with any
positive test result should receive neck dissection. If a
patient has 1 positive test, no more tests are needed.
Patients with 2 negative results should receive the watchful waiting management policy.
With the assumption of the independence of tests based
on the Bayesian theory and the collected data with
HEAD & NECKDOI 10.1002/HED

APRIL 2016

629

LIAO ET AL.

FIGURE 1. The search process.

different pre-examination probabilities, the postexamination positive and negative neck nodal probability can be
calculated; simulation of overtreatment and undertreatment can also be simulated.

RESULTS
The search process is shown in Figure 1. The abstracts
and titles of 206 primary studies were identified for initial
review based on the described search strategy. Full-text
reviews were required for 195 publications to determine
study eligibility. Five studies with cases number <10
were excluded. Subsequently, a total of 73 articles were
selected based on agreement between the 2 reviewers.
Twenty-two studies evaluated multiple tests at the same
time. Ten studies fulfilled all the inclusion criteria for
CT, 7 studies for MRI, 12 studies for PET, 9 studies for
630

HEAD & NECKDOI 10.1002/HED

APRIL 2016

ultrasound, 5 studies for ultrasound-guided FNA, and 55

studies for SNB. The results are summarized in Table 1.
The meta-analysis results are shown in Table 2.
Because of no false-positive results in ultrasound-guided
FNA and SNB, the specificity is 100%. Univariate analyses with the random-effects approach were performed to
calculate the combined sensitivities of ultrasound-guided
FNA and SNB. The pooled estimates for sensitivity are
56.4% (95% confidence interval [CI] 5 44.7% to 67.4%)
and 84.9% (CI 5 82.0% to 87.5%) for ultrasound-guided
FNA and SNB, respectively. For bivariate analysis, the
pooled estimates for sensitivity were 47.0% (CI 5 38.2%
to 56.0%), 56.6% (CI 5 39.8% to 71.9%), 48.3%
(CI 5 30.9% to 66.1%), and 63.3% (CI 5 54.1% to
71.1%) for CT, MRI, PET, and ultrasound, respectively.
The pooled estimates for specificity were 88.9%
(CI 5 82.0% to 93.3%), 82.5% (CI 5 68.6% to 91.1%),

SENTINEL

NODE BIOPSY COMBINED WITH OTHER DIAGNOSTIC TOOLS IN STAGING CN0 HEAD AND NECK CANCER

TABLE 1. Summary of reviewed literature.

Method

No. of studies
Publication year
Cancer site
Oral cavity
Larynx
Mixed
Other
Design
Prospective
Retrospective
Center
One
Multiple
T classification
T1T2 only
T1T4

CT

MRI

Ultrasound

Ultrasound-guided FNA

PET/CT

SNB

10
19852005

7
19922005

9
19922005

5
19922004

12
19952010

55
20012013

3
0
7
0

3
0
4
0

5
0
4
0

1
0
4
0

4
0
7
1

27
2
7
19

9
1

6
1

8
1

5
0

12
0

52
3

9
1

7
0

8
1

4
1

11
1

52
3

0
10

1
6

2
7

0
5

0
10

23
32

Abbreviations: FNA, fine-needle aspiration; PET, positron emission tomography; SNB, sentinel node biopsy.

86.2% (CI 5 76.9% to 92.1%), and 79.1% (CI 5 73.4% to

83.8%) for CT, MRI, PET, and ultrasound, respectively.
The AUC values were 81.1% (CI 5 56.2% to 100%),
79.1% (CI 5 65.7% to 92.5%), 82.6% (CI 5 69.4% to
95.9%), 80.7% (CI 5 73.9% to 87.4%), 97.0%
(CI 5 84.9% to 100%), and 98.0% (CI 5 96.1% to 99.9%)
for CT, MRI, PET, ultrasound, ultrasound-guided FNA,
and SNB, respectively.
Figure 2 and Table 3 show the results of the simulation
with the sensitivity analysis. With pre-examination nodal
metastasis was set at 20%, 30%, 40%, and 60%, and the
postexamination NPV is demonstrated in Table 3. With a
range from 0 to 1, the posterior probability of nodal
metastasis is simulated in Figure 2. For CT then SNB or
MRI then SNB strategies, even with a very high occult
rate by palpation (up to 60%), NPV will be still higher as
88%. The negative result will reduce the final occult metastatic rate below 12%.
For the base case analysis, if the pretest neck occult
nodal prevalence is set at 40%, the proportion of patients
receiving END and watchful waiting is shown in Table 4.
For CT plus SNB staging strategies, 43% patients will
receive END, and 57% patients will receive watchful
waiting management. In the END group, 85% is metastatic positive; in the watchful waiting group, 6% is metastatic positive, and 94% is metastatic negative. Compared
to other staging methods, CT/MRI plus SNB strategies

have a smaller overtreatment rate in the END group and

undertreatment rate in the watchful waiting group. The
overtreatment and undertreatment rates are 15% versus
6% for the CT plus SNB strategy and 22% versus 5% for
MRI plus SNB, respectively.

DISCUSSION
There are several diagnostic modalities that can be used
for staging the nodal status of cN0 head and neck cancer.
This is the first study to compare different modern diagnostic technologies, including SNB, in staging cN0 head
and neck cancer at the same time. Simulation with CT/
MRI in combination with different tools was also
performed.
CT and MRI are the standard evaluation tools in head
and neck cancer evaluation. CT and MRI improve clinical
staging and are routinely used in many centers. The feasibility of operation can be evaluated preoperatively by CT
or MRI, which represents a great advancement in modern
surgery. CT and MRI can evaluate the primary tumor and
neck at the same time. The malignancy criteria for the
nodal metastasis of the neck in CT and MRI have been
defined by a number of authors and continue to evolve as
new-generation scanners provide improved clarity and
resolution. Although they have limitations, the most current malignancy criteria include the following13: (1) a
minimal diameter of 15 mm for nodes located in level II

TABLE 2. Results of the meta-analysis and parameters used in sequential testing analysis.

CT
MRI
PET
Ultrasound
Ultrasound-guided FNA
SNB

Studies

Neck sides

Sensitivity

Specificity

AUC

QUADAS

10
7
12
9
5
55

315
261
602
391
181
2469

47.0% (38.2% to 56.0%)

56.6% (39.8% to 71.9%)
48.3% (30.9% to 66.1%)
63.3% (54.1% to 71.1%)
56.4% (44.7% to 67.4%)
84.9% (82.0% to 87.5%)

88.9% (82.0% to 93.3%)

82.5% (68.6% to 91.1%)
86.2% (76.9% to 92.1%)
79.1% (73.4% to 83.8%)
100%
100%

81.1% (56.2% to 100%)

79.1% (65.7% to 92.5%)
82.6% (69.4% to 95.9%)
80.7% (73.9% to 87.4%)
97.0% (84.9% to 100%)
98.0% (96.1% to 99.9%)

8.7 (6.311.1)
8.6 (6.310.9)
10.3 (8.212.3)
9.0 (5.910.1)
9.8 (8.511.1)
11.7 (9.314.0)

Abbreviations: AUC, area under receiver operating characteristic curve; QUADAS, Quality Assessment of Diagnostic Accuracy Studies; PET, positron emission tomography; FNA, fine-needle aspiration; SNB, sentinel node biopsy.

HEAD & NECKDOI 10.1002/HED

APRIL 2016

631

LIAO ET AL.

FIGURE 2. Simulation results of the conditional probability analysis. The x-axis shows the pretest neck lymph node occult metastatic rate ranging
from 0 to 1. The curve in gray color is the positive neck metastatic rate with a negative test result posted for each strategy. The upper black curves
are the positive neck metastatic rates with positive results. For sentinel node biopsy (SNB), even with very high occult rate by palpation (up to
60%), negative predictive value (NPV) will be higher than 85%. The negative result will reduce the final occult metastatic rate below 15%.

TABLE 3. Negative predictive value of sequential testing under conditional probability simulation.
Prior probability, %
Examination

CT 1 PET
CT 1 ultrasound
CT 1 ultrasound-guided FNA
CT 1 SNB
MRI 1 PET
MRI 1 ultrasound
MRI 1 ultrasound-guided FNA
MRI 1 SNB

60

40

30

20

65
71
72
88
68
73
74
89

81
84
85
94
83
86
87
95

87
89
90
96
88
91
91
97

92
93
94
98
93
94
95
98

Abbreviations: PET, positron emission tomography; FNA, fine-needle aspiration; SNB, sentinel
node biopsy.
Negative predictive values (NPVs) depend not only on the sensitivity and specificity of diagnostic tests but also on prior test disease prevalence. For CT/MRI plus SNB strategies, the
NPV values are 88% to 89% with a prior probability of 60%.

632

HEAD & NECKDOI 10.1002/HED

APRIL 2016

and 10 mm for nodes located in other levels; (2) groups

of 3 or more borderline nodes (1 to 2 mm smaller; cluster); (3) nodes of any size with evidence of central necrosis; (4) loss of tissue planes (fat planes), indicating
extracapsular extension.
CT and MRI are valuable adjuncts to palpation alone
and are especially valuable in those head and neck sites
that are not accessible to palpation, such as the retropharyngeal, paratracheal, and upper mediastinal nodes, and in
assessing extranodal disease.
CT and MRI comprise the basic examination for staging head and neck cancers because the primary tumor can
be checked in detail for oncologic soundness via en bloc
resection during surgery. For neck nodal evaluation, the
pooled estimates for sensitivity were 47.0% (38.2% to
56.0%) and 56.6% (39.8% to 71.9%) for CT and MRI,
respectively. The pooled estimates for specificity were
88.9% (82.0% to 93.3%) and 82.5% (68.6% to 91.9%)
for CT and MRI, respectively. However, the performance

SENTINEL

NODE BIOPSY COMBINED WITH OTHER DIAGNOSTIC TOOLS IN STAGING CN0 HEAD AND NECK CANCER

TABLE 4. Hypothetical overtreatment and undertreatment with different management strategies and simulation of different diagnostic and management
strategies with a pretest prevalence occult metastatic rate set at 40%.
Watchful waiting
%

END %

PN1
CT 1 PET
CT 1 ultrasound
CT 1 ultrasound-guided
FNA
CT 1 SNB

43
67
50
64
37
82
43
85

PN-*
33
36
18
15

PN1
57
19
50
16
63
15
57
6

PN-

PN1
MRI 1 PET

81
MRI 1 ultrasound
84
MRI 1 ultrasound-guided FNA
85
MRI 1 SNB
94

Watchful
waiting%

END %

48
64
54
62
43
76
48
78

PN-*
36
38
24
22

PN1
52
17
46
14
57
13
52
5

PN83
86
87
95

Abbreviations: END, elective neck dissection; PET, positron emission tomography; FNA, fine-needle aspiration; SNB, sentinel node biopsy.
* PN- in END group means overtreatment.

PN1 in watchful waiting group means undertreatment.

The table indicates the percentage of patients who will receive END and watchful and waiting management in each strategy and the percentage of patients who are really positive metastatic (PN1)
and really negative metastatic (PN-) in each management strategy.

of CT or MRI may not be effective for predicting which

patients with early cN0 head and neck SCC can safely
avoid prophylactic neck dissection.
PET using 18-F fluoro-2-deoxyglucose as a metabolic
tracer has shown promise in detecting lymph node metastases. PET scan was never reported to be cost-effective in
the classification of cN0 patients and the guidance of further treatment strategy.14 However, the pooled estimates
for sensitivity and specificity were 48.3% (30.9% to
66.1%) and 86.2% (76.9% to 92.1%) for PET, respectively, in our meta-analysis. The sensitivity of PET may
still not be adequate to guide the management of a cN0
neck.
The use of ultrasound and ultrasound-guided FNA
cytology for the N0 neck has been popularized in Europe
since the late 1980s and is now gaining acceptance in
other countries as well. The ultrasound criteria for metastatic lymph nodes includes absent hilar blood flow in
Doppler sonography, round shape (short to long axis ratio
>0.5), central necrosis and extracapsular invasion, and
short-diameter axis larger than 7 mm.15 The major
advantage of combining ultrasound and ultrasound-guided
FNA cytology is that an ultrasound is capable of detecting very small lymph nodes, whereas cytology will assure
that the specificity remains almost 100%. In our systematic review and meta-analysis, the pooled estimates for
sensitivity and specificity were 63.3% (54.0% to 71.7%)
and 79.1% (73.4% to 83.8%) for ultrasound, respectively.
The pooled estimate for sensitivity was 56.4% (44.7% to
67.4%).
The latest report16 for ultrasound-guided FNA in cN0
neck evaluation included 285 consecutive patients with
early T1 or T2 oral cancer; of these, 234 were followed
by the wait and scan policy, and 51 underwent END.
Survival rates were compared between the groups, and
correction for confounding factors was performed. With
regard to survival in patients with early stage oral cancer
and a cN0 neck, a wait and scan policy using strict
ultrasound-guided FNA surveillance is justified because
survival is not negatively influenced. Using a wait and

scan follow-up strategy instead of END, unnecessary

neck dissection and its accompanying morbidity can be
avoided in 72.2% of patients. The combination of multiple diagnostic tests for watchful waiting management in
cN0 head and neck cancer is necessary.
In an attempt to more specifically select lymph nodes
that potentially contain metastases, SNB has been introduced extensively in melanoma and breast cancer.17 The
sentinel lymph node is likely to be the first lymph node
to harbor metastasis and can be used to provide information on the rest of the nodal basin. It is usually identified
by the peritumoral injection of a radioactive colloid or
blue dye. Preoperative lymphoscintigraphy, intraoperative
visualization of blue coloration, and intraoperative radionuclide detection with a gamma probe allow for the identification of the sentinel lymph node. After surgical
removal, this node can be studied meticulously by histopathological examination using stepped serial sectioning
and immunohistochemistry. If the sentinel lymph node
contains metastatic tumor cells, treatment of the neck is
recommended, sometimes in a second procedure. The sentinel lymph node procedure might be more precise than
other imaging procedures and less invasive than END.
Moreover, it is associated with significantly less postoperative morbidity and better shoulder function compared to
END.18 In our meta-analysis, the pooled estimates for
sensitivity and specificity were 84.9% (82.0% to 87.5%)
and 100%, respectively, for SNB. The diagnostic performance of SNB is better than that of other tests
(Table 2).
An increasing number of studies include modern diagnostic tools, such as ultrasound-guided FNA and SNB,
and even molecular markers in staging. Our meta-analysis
included all of the diagnostic tools in the evaluation of
cN0. We also identified 1 study using gene expression
profile in staging the cN0 neck; however, as only 1 study
was found, no meta-analysis was performed. The American Joint Committee on Cancer has stated that clinical
staging should include a physical examination as well as
the results of other imaging modalities. Research is now
HEAD & NECKDOI 10.1002/HED

APRIL 2016

633

LIAO ET AL.

directed toward finding a staging strategy sensitive

enough to bring the risk of occult metastases below 15%
to 20%, which means an NPV higher than 80% to 85%.
The methodology of our study is a comprehensive systematic review and meta-analysis of diagnostic tests in
staging cN0 head and neck SCC. Moreover, we performed a simulation of combined strategies. In this metaanalysis, we found that the diagnostic performance of
SNB (sensitivity, specificity, and AUC; Table 2) is superior to other diagnostic tests. For NPV simulation (Table
3 and Figure 2), CT/MRI plus SNB strategies have a high
NPV and even prior occult metastasis is as high as 60%;
NPV is 88%, indicating a posterior probability of occult
metastasis as low as 12%. According to Pillsbury and
Clark7, for these cN0 patients, occult neck lymph node
metastasis may be found in 8% to 55% of cases depending on the subsites and clinical stage of the cancers.
For base-case simulation (Table 4), with a pretest
occult rate set at 40%, CT then ultrasound-guided FNA
showed the most patients (63%) receiving the watchful
waiting policy, but 15% will be false negatives and will
suffer from delayed metastasis. SNB is the most accurate
diagnostic test among these tools. By combining CT and
MRI with SNB, fewer patients will suffer from overtreatment and undertreatment.
There were some limitations to our study because this
was a literature review and simulation from meta-analysis
results. Further trials to compare these management strategies are necessary, although it is rather difficult to conduct such a trial.

CONCLUSIONS
Based on our systematic review and meta-analysis of
the scenario simulation of 6 diagnostic tools, the SNB
procedure has the best diagnostic performance. The combination of CT/MRI and SNB for nonpalpable clinical N0
head and neck cancer is preferred.

Acknowledgment
The authors thank Miss Yu-Ping Chengs help in preparing the manuscript.

634

HEAD & NECKDOI 10.1002/HED

APRIL 2016

REFERENCES
1. Walk EL, Weed SA. Recently identified biomarkers that promote lymph
node metastasis in head and neck squamous cell carcinoma. Cancers
(Basel) 2011;3:747772.
2. van den Brekel MW, Castelijns JA, Stel HV, Golding RP, Meyer CJ, Snow
GB. Modern imaging techniques and ultrasound-guided aspiration cytology
for the assessment of neck node metastases: a prospective comparative
study. Eur Arch Otorhinolaryngol 1993;250:1117.
3. NCCN practice guidelines for cancer of the head and neck. Available at:
http://wwwnccnorg/professionals/physician_gls/f_guidelinesasp#site. Accessed
January 30, 2014.
4. Weiss MH, Harrison LB, Isaacs RS. Use of decision analysis in planning a
management strategy for the stage N0 neck. Arch Otolaryngol Head Neck
Surg 1994;120:699702.
5. AJCC Cancer Staging Manual seventh edition. Available at: https://cancerstagingorg/Pages/defaultaspx. Accessed January 30, 2014.
6. de Bondt RB, Nelemans PJ, Hofman PA, et al. Detection of lymph node
metastases in head and neck cancer: a meta-analysis comparing US, USgFNAC, CT and MR imaging. Eur J Radiol 2007;64:266272.
7. Pillsbury HC III, Clark M. A rationale for therapy of the N0 neck. Laryngoscope 1997;107:12941315.
8. Whiting PF, Weswood ME, Rutjes AW, Reitsma JB, Bossuyt PN, Kleijnen
J. Evaluation of QUADAS, a tool for the quality assessment of diagnostic
accuracy studies. BMC Med Res Methodol 2006;6:9.
9. Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman
AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005;58:982990.
10. Moses LE, Shapiro D, Littenberg B. Combining independent studies of a
diagnostic test into a summary ROC curve: data-analytic approaches and
some additional considerations. Stat Med 1993;12:12931316.
11. Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A. Meta-DiSc: a
software for meta-analysis of test accuracy data. BMC Med Res Methodol
2006;6:31.
12. Ross GL, Shoaib T, Soutar DS, et al. The First International Conference on
Sentinel Node Biopsy in Mucosal Head and Neck Cancer and adoption of a
multicenter trial protocol. Ann Surg Oncol 2002;9:406410.
13. Lee SH, Huh SH, Jin SM, Rho YS, Yoon DY, Park CH. Diagnostic value
of only 18F-fluorodeocyglucose positron emission tomography/computed
tomography-positive lymph nodes in head and neck squamous cell carcinoma. Otolaryngol Head Neck Surg 2012;147:692698.
14. Govers TM, Hannink G, Merkx MA, Takes RP, Rovers MM. Sentinel node
biopsy for squamous cell carcinoma of the oral cavity and oropharynx: a
diagnostic meta-analysis. Oral Oncol 2013;49:726732.
15. Hodder SC, Evans RM, Patton DW, Silvester KC. Ultrasound and fine needle aspiration cytology in the staging of neck lymph nodes in oral squamous cell carcinoma. Br J Oral Maxillofac Surg 2000;38:430436.
16. Flach GB, Tenhagen M, de Bree R, et al. Outcome of patients with early
stage oral cancer managed by an observation strategy towards the N0 neck
using ultrasound guided fine needle aspiration cytology: no survival difference as compared to elective neck dissection. Oral Oncol 2013;49:157164.
17. Leusink FK, van Es RJ, de Bree R, et al. Novel diagnostic modalities for
assessment of the clinically node-negative neck in oral squamous-cell carcinoma. Lancet Oncol 2012;13:e554e561.
18. Schiefke F, Akdemir M, Weber A, Akdemir D, Singer S, Frerich B. Function, postoperative morbidity, and quality of life after cervical sentinel
node biopsy and after selective neck dissection. Head Neck 2009;31:503
512.