144 / 220
level. Demographic data, pathological features, location
of SN, and survival data were collected for each patient.
Statistical analysis was performed using R survival
package
[19]
. Univariate survival analysis models were
built using Kaplan
e
Meier product-limit estimator for
overall survival (OS), disease-specific survival (DSS) and
DFS, and multivariate and models with univariate
continuous covariates were built using Cox proportional
hazards model. Table analysis on 3-year outcomes (such
as recurrence within 3 years of SNB) was performed
using either chi-square or Fisher’ exact to test signifi-
cance, depending upon the distribution of the variable in
question.
3. Results
The patient and carcinoma characteristics are shown in
Table 2
.
3.1. Lymphatic drainage characteristics
A total of 483 neck sides were examined from 415 patients
with 1342 SNs harvested. There were a mean of 2.75 SN
per neck or 3.2 SN per patient (range 1
e
10), with an
average size of 11.8 mm (range 3
e
30 mm). The primary
tumour was positioned in the midline in 11.4% (N
Z
46)
and laterally in 88.6% (N
Z
369) of cases. Lateral tu-
mours drained ipsilaterally in 87% of cases (320 in 369)
but in 10% (40 cases) they drained bilaterally and in 2.4%
(9 cases) exclusively to the contralateral neck. Sixty
percent (28 in 46) of midline lesions drained bilaterally.
3.2. Occult cervical disease
SNB detected metastasis in 94 patients (23%), 16 of
whom had extra-capsular spread (17%). Of the 75 cases
classified by the UICC guidelines, 12 (16%) contained
ITC, 36 (48%) contained micrometastasis and 27 (36%)
macrometastasis.
Fifteen patients with a negative SNB subsequently
developed isolated cervical metastasis with a negative
primary tumour site (one with concomitant distant
metastasis) and these were recorded as a false-negative
biopsy. Therefore, of 415 patients, 109 had occult
metastasis. SNB had a sensitivity, negative predictive
value and FNR of 86%, 95%, and 14%, respectively. The
FNR, sensitivity and negative predictive value for the
three most common tumour sites are shown in
Table 3
.
In the 49 patients with unexpected bilateral or
contralateral drainage from a lateral carcinoma, a pos-
itive SN was identified in seven (two bilateral and five
solely contralateral).
All 94 patients with a positive SNB underwent neck
dissection. In seven cases, dissection was bilateral, giving
a total of 101 neck dissections, of which 47% (47 in 101)
were selective, and the remainder modified radical.
In 85% of cases, no further positive nodes were found in
the completion specimen. Of the patients with additional
Table 2
SENT patient demographics and tumour characteristics.
Characteristic
Overall n
Z
Negative SNB
Positive SNB
Effect of characteristic
on sentinel node status
Total patients
415
321 (77%)
94 (23%)
Male
247 (60%)
194 (60%)
53 (56%)
Gender p
Z
0.48
Female
168 (40%)
127(40%)
41 (44%)
Median age (years, range)
61 (28-92)
61(28
e
92)
62 (29
e
91)
Age p
Z
0.74
Primary tumour site
Oral tongue anterior 2/3
213 (51%)
157/213 (74%)
56/213(26%)
Topographic site p
Z
0.1
Oral tongue posterior 1/3
43 (10.5%)
32/43(74%)
11/43 (26%)
Buccal mucosa
17 (4%)
13/17 (76%)
4 /17(24%)
Floor of mouth
101 (25%)
87/101 (86%)
14/101 (14%)
Hard palate
3 (1%)
3/3 (100%)
0
Lower alveolus/gingival
8 (2%)
4/8 (50%)
4/8(50%)
Lower lip
6 (1.5%)
6 /6 (100%)
0
Upper alveolus/gingival
5 (1%)
5/5 (100%)
0
Retromolar
9 (2%)
6/9 (66%)
3/9 (33%)
Soft palate
10 (2%)
8/10 (80%)
2/10 (20%)
T stage
T1
296 (71%)
239 (74%)
57 (61%)
T stage p
Z
0.032
T2
119 (29%)
82 (26%)
37(39%)
Neck dissection (for SNB
þ
or recurrent disease)
Yes
121(29%)
27(8%)
94 (100%)
N/A
No
294(71%)
294 (92%)
0
Radiotherapy (n
Z
36) or chemoradiotherapy (n
Z
12) (
>
1 positive node, ECS or recurrence)
Yes
48 (12%)
23 (7%)
25 (27%)
N/A
No
367 (88%)
298(93%)
69 (73%)
Statistical testing by chi-square or two-sample t-test depending upon characteristic.
SENT, Sentinel European Node Trial; SN, sentinel node; SNB, sentinel node biopsy; N/A, not applicable; ECS, extracapsular spread.
C. Schilling et al. / European Journal of Cancer 51 (2015) 2777
e
2784
124