2017 Section 7 Green Book

C. Schilling et al. / European Journal of Cancer 51 (2015) 2777 e 2784

Table 3 NPV, sensitivity and FNR by tumour location where a false-negative result is recorded as isolated neck recurrence following a negative sentinel node biopsy. Tumour False-negative rate Sensitivity NPV

Table 5 Univariate Kaplan e Meier (for categorical) or Cox PH (for contin- uous) analysis of factors influencing overall survival following SNB (significance levels *0.05, **0.01, ***0.001). Factor Overall survival (OS) p value

Anterior tongue Posterior tongue Floor of mouth

14% (9/65) 21% (3/14) 13% (2/16)

85% 79%

94% 91% 98% 95%

Age of patient

0.003* (Cox PH)

87.5%

Site of tumour (grouped per location anterior versus posterior oral cavity)

0.755

Total SENT group 14% (15/109)

86%

p Z 1

Fisher’ exact test

T size (T1 versus T2)

0.465 0.142

Depth of invasion ( > 4 versus 4 mm) Degree of differentiation (well versus moderate, poor) Sentinel node status (SNB þ versus SNB ) Metastasis type (ITC versus Mi, Ma) Total positive nodes (0, 1, 2, > 2) Extra-capsular spread (no versus yes) Margin ( 1 versus > 1 mm)

NPV, negative predictive value; SENT, Sentinel European Node Trial.

0.004*

positive non-sentinel nodes, 13 in 15 (87%) were located in the same neck level as the SN or an adjacent neck level.

0.741

0.000083***

0.032*

0.000016***

3.3. Outcome

0.029*

RT (no, yes)

0.93

In this cohort of patients, 3-year figures for OS, DFS and DSS were 88%, 92% and 94%, respectively. Disease recurred in 56 patients ( Table 4 ). Univariate analysis of the factors that affected outcome (overall survival) was investigated with Kaplan e Meier survival analysis for categorical vari- ables and Cox proportional hazards for continuous variables (such as age of patient) ( Table 5 , Fig. 1 e 3 ). A multivariate Cox proportional hazards model was then run with all variables that showed a univariate p value < 0.25. The resulting multivariate Cox proportional hazards model found that the grouped number of pos- itive nodes (p Z 0.0008) and SN status (p Z 0.003) were the only significant factors. Morbidity of SNB was minimal. Minor complications were seroma [1] , haematoma [8] , local infection [3] , and lymphoedema [1] . There were two notable complica- tions: one phrenic nerve palsy and one patient had a cerebellar stroke secondary to surgery. Mean hospital stay following SNB and primary tumour resection was 5.7 d (range 0 e 30) with 161 pa- tients discharged within 3 d of their surgery. Lengths of stay varied considerably by country (average of 9 d in Belgium compared to 3 d in Denmark). 3.4. Complications

PH, proportional hazards; SNB, sentinel node biopsy; ITC, isolated tumour cells; Mi, micrometastasis; Ma, macrometastasis; RT, radiotherapy.

3.5. Adjuvant therapy

Adjuvant therapy (RT or chemoradiotherapy) was given to 12% (48 in 415) of patients. In the SNB-positive group, 27% (25 in 94) received adjuvant therapy (more than one positive node or extracapsular spread e ECS) but was used more freely (80%) in the false-negative SNB group (12 in 15) to help salvage patients. There was no significant survival difference between those with and without adjuvant radiotherapy (p Z 0.67). The results of the study demonstrate clearly the value and safety of SNB for staging the N0 neck in routine clinical practice. The principal aim of the study was to establish whether SNB is a safe oncological procedure. This has been confirmed with DFS of 92% at 3 years following treatment. The second objective was to determine, in the context of oral and oropharyngeal cancer, whether SNB was an effective diagnostic test for microscopic deposits of metastatic cancer. The study showed conclusively that the SNB technique works effectively in the oral cavity. The injection of radiotracer (lymphoscintigraphy) will define an SN in the vast majority of patients ( > 99%). In this cohort of patients with a 3:1 distribution of T1:T2 oral squamous carcinoma and radiologically N0 neck, it transpired that 26% (109 in 415) had occult cervical disease. The SNB technique failed to detect occult metastasis in 14% (15 in 109) of patients, only half of whom (53.3%: 8 in 15) were amenable to salvage. This is somewhat counterbalanced through identifi- cation of unexpected contralateral lymphatic drainage by SNB. This occurred in 12% (49 in 369) of cases and in seven instances, the contralateral SN was positive. Thus, 6% (7 in 109) of occult cervical metastasis would have 4. Discussion

Table 4 Recurrences at 3 years.

Total (n Z 415)

SNB negative (n Z 321)

SNB positive (n Z 94)

Local ( distant)

18 (4.3%)

13 (4.0%) 8 (2.5%) 15 (4.7%)

5 (5.3%) 1 (1.0%) 14 (15%)

Local and neck 9 (2.2%)

Neck ( distant)

29 (7.5%)

Distant

0

0

0

Outcome following recurrence Dead with disease

14 (4.3%) 19 (5.9%) 2 (0.6%) 2 (0.6%)

16 (17%) 2 (2.1%) 2 (2.1%)

Alive no disease

Dead with no disease Alive with disease

0

SNB, sentinel node biopsy.

125