Articles

Sentinel-lymph-node biopsy in patients with breast cancer

before and after neoadjuvant chemotherapy (SENTINA):
a prospective, multicentre cohort study
Thorsten Kuehn, Ingo Bauerfeind, Tanja Fehm, Barbara Fleige, Maik Hausschild, Gisela Helms, Annette Lebeau, Cornelia Liedtke,
Gunter von Minckwitz, Valentina Nekljudova, Sabine Schmatloch, Peter Schrenk, Annette Staebler, Michael Untch

Summary
Background The optimum timing of sentinel-lymph-node biopsy for breast cancer patients treated with neoadjuvant Published Online
chemotherapy is uncertain. The SENTINA (SENTinel NeoAdjuvant) study was designed to evaluate a specific May 15, 2013
http://dx.doi.org/10.1016/
algorithm for timing of a standardised sentinel-lymph-node biopsy procedure in patients who undergo neoadjuvant S1470-2045(13)70166-9
chemotherapy. See Online/Comment
http://dx.doi.org/10.1016/
Methods SENTINA is a four-arm, prospective, multicentre cohort study undertaken at 103 institutions in Germany S1470-2045(13)70180-3
and Austria. Women with breast cancer who were scheduled for neoadjuvant chemotherapy were enrolled into the Interdisciplinary Breast Centre,
study. Patients with clinically node-negative disease (cN0) underwent sentinel-lymph-node biopsy before neoadjuvant Department of Gynaecology
and Obstetrics, Klinikum
chemotherapy (arm A). If the sentinel node was positive (pN1), a second sentinel-lymph-node biopsy procedure was Esslingen, Esslingen, Germany
done after neoadjuvant chemotherapy (arm B). Women with clinically node-positive disease (cN+) received (Prof T Kuehn MD); Department
neoadjuvant chemotherapy. Those who converted to clinically node-negative disease after chemotherapy (ycN0; of Gynaecology and Obstetrics,
arm C) were treated with sentinel-lymph-node biopsy and axillary dissection. Only patients whose clinical nodal Klinikum Landshut, Landshut,
Germany (I Bauerfeind MD);
status remained positive (ycN1) underwent axillary dissection without sentinel-lymph-node biopsy (arm D). The Department of Gynaecology
primary endpoint was accuracy (false-negative rate) of sentinel-lymph-node biopsy after neoadjuvant chemotherapy and Obstetrics, University of
for patients who converted from cN1 to ycN0 disease during neoadjuvant chemotherapy (arm C). Secondary endpoints Düsseldorf, Düsseldorf,
included comparison of the detection rate of sentinel-lymph-node biopsy before and after neoadjuvant chemotherapy, Germany (Prof T Fehm MD);
Department of Pathology
and also the false-negative rate and detection rate of sentinel-lymph-node biopsy after removal of the sentinel lymph (B Fleige MD) and Department
node. Analyses were done according to treatment received (per protocol). of Gynaecology and Obstetrics
(Prof M Untch MD),
Findings Of 1737 patients who received treatment, 1022 women underwent sentinel-lymph-node biopsy before Multidisciplinary Breast
Centre, Helios Klinikum
neoadjuvant chemotherapy (arms A and B), with a detection rate of 99·1% (95% CI 98·3–99·6; 1013 of 1022). Berlin-Buch, Berlin, Germany;
In patients who converted after neoadjuvant chemotherapy from cN+ to ycN0 (arm C), the detection rate was 80·1% Spital Rheinfelden,
(95% CI 76·6–83·2; 474 of 592) and false-negative rate was 14·2% (95% CI 9·9–19·4; 32 of 226). The false-negative Rheinfelden, Switzerland
rate was 24·3% (17 of 70) for women who had one node removed and 18·5% (10 of 54) for those who had two sentinel (M Hausschild MD);
Department of Gynaecology
nodes removed (arm C). In patients who had a second sentinel-lymph-node biopsy procedure after neoadjuvant and Obstetrics (G Helms MD)
chemotherapy (arm B), the detection rate was 60·8% (95% CI 55·6–65·9; 219 of 360) and the false-negative rate was and Department of Pathology
51·6% (95% CI 38·7–64·2; 33 of 64). (A Staebler MD), University
Medical Centre Tübingen,
Tübingen, Germany;
Interpretation Sentinel-lymph-node biopsy is a reliable diagnostic method before neoadjuvant chemotherapy. After Department of Pathology,
systemic treatment or early sentinel-lymph-node biopsy, the procedure has a lower detection rate and a higher false- University Medical Centre,
negative rate compared with sentinel-lymph-node biopsy done before neoadjuvant chemotherapy. These limitations Hamburg-Eppendorf,
Hamburg, Germany
should be considered if biopsy is planned after neoadjuvant chemotherapy.
(A Lebeau MD); Department of
Gynaecology and Obstetrics,
Funding Brustkrebs Deutschland, German Society for Senology, German Breast Group. University Hospital
Schleswig-Holstein Campus
Lübeck, Lübeck, Germany
Introduction locally advanced disease and is being used increasingly
(C Liedtke MD); German Breast
Axillary-lymph-node status is one of the strongest for early-stage breast cancer.4 This therapeutic approach Group, Neu Isenburg, Germany
prognostic factors for patients with breast cancer and it provides in-vivo chemosensitivity testing and prognostic (Prof G von Minckwitz MD,
guides adjuvant local and systemic treatment decisions. information. Patients with an unfavourable tumour-to- V Nekljudova PhD); University
Women’s Hospital Frankfurt,
In recent years, sentinel-lymph-node biopsy has replaced breast ratio can be downstaged to allow less radical
Frankfurt, Germany
full axillary-lymph-node dissection as a staging procedure surgery and to increase the rate of breast-conserving (Prof G von Minckwitz);
for patients who undergo primary surgery and have treatment.5,6 Elisabeth-Krankenhaus Kassel
clinically negative lymph nodes. Sentinel-lymph-node Timing of sentinel-lymph-node biopsy in the neo- gGmbH, Kassel, Germany
(S Schmatloch MD); and Breast
biopsy provides an accurate assessment of histological adjuvant setting is controversial. Reliable data for the Cancer Center, General
nodal status and is associated with less acute and chronic detection rate, accuracy (the false-negative rate), and the Hospital Linz, Linz, Austria
morbidity than axillary-lymph-node dissection.1–3 Neo- number of regional relapses are available for when (P Schrenk MD)
adjuvant chemotherapy is established for treatment of biopsy is done before systemic adjuvant treatment in

www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9 1

 Articles

Correspondence to: patients with clinically node-negative (cN0) disease,7,8 but disease before neoadjuvant chemotherapy. A few, small,
Prof T Kuehn, Interdisciplinary limited data are available in the context of neoadjuvant retrospective series have been published of patients who
Breast Center, Department for
Gynecology and Obstetrics,
chemotherapy. One advantage of doing sentinel-lymph- presented initially with cN+ disease and converted to
D-73730 Esslingen, Germany node biopsy before neoadjuvant chemotherapy is that negative axillary status after neoadjuvant chemotherapy;
t.kuehn@klinikum-esslingen. knowledge of the initial histological nodal status can be detection rates were 77·6–98·0% of patients and false-
de used to guide postoperative locoregional treatment negative rates were registered for 5·6–35·5%.14–19 In a
decisions. However, the best surgical approach in the prospective multicentre study of sentinel-lymph-node
axilla for the 20–40% of patients with initially positive biopsy after neoadjuvant chemotherapy,20 the detection
lymph nodes (cN+) who are downstaged after rate was 94·6% for patients with cN0 disease before
neoadjuvant chemotherapy to a clinically negative lymph neoadjuvant chemotherapy and 81·5% for those who
node status (ycN0) is unclear.5,6 In this population, biopsy presented initially with cN+ status; false negatives were
done after neoadjuvant chemotherapy would increase noted in 9·4% and 15·0%, respectively. However, the low
the overall rate of axilla-conserving treatment. statistical power of this study precludes reliable evidence
Furthermore, growing evidence suggests that the nodal for subgroups of patients who initially have cN0 or cN+
stage after neoadjuvant chemotherapy reflects prognosis disease.
more accurately than does initial axillary status9 and Khan and colleagues21 reported 33 patients undergoing
could, in the near future, lead to tailoring of regional a second sentinel-lymph-node biopsy after neoadjuvant
treatment.10 chemotherapy. These patients had initially presented
The feasibility and accuracy of doing sentinel-lymph- with a clinically negative axilla before neoadjuvant
node biopsy after neoadjuvant chemotherapy is of some chemotherapy but were found to have an involved
concern. For example, lymphatic drainage from the breast sentinel lymph node at a first sentinel-lymph-node
could be impaired, thus hampering detection of the biopsy done before neoadjuvant chemotherapy. The
sentinel lymph node. Furthermore, tumour regression in detection rate for this second sentinel-lymph-node biopsy
the axilla could follow a non-uniform pattern, leading to was 97·0% and the false-negative rate was 4·5%.21
an unacceptable false-negative rate. Many cohort studies The SENTINA (SENTinel NeoAdjuvant) study was
have been done of sentinel-lymph-node biopsy after designed to provide reliable data for the feasibility and
neoadjuvant chemotherapy, and findings of three meta- accuracy of a standardised sentinel-lymph-node biopsy
analyses showed detection rates of 63–100% and false- procedure in different settings before and after neo-
negative rates of 0–39%.11–13 However, most of these trials adjuvant chemotherapy. We included several clinical
were either retrospective or single-centre in design and scenarios, with the aim to ascertain the best timing
included only a few patients who had predominantly cN0 strategy for sentinel-lymph-node biopsy in breast cancer
patients treated with neoadjuvant chemotherapy.

Clinically node-negative Clinically node-positive Methods

(cN0) (cN1 or cN2)
Study design
The SENTINA study is a four-arm, prospective,
Sentinel-lymph-node multicentre cohort study undertaken at 103 centres in
biopsy Germany and Austria. We enrolled patients with breast
cancer who were scheduled for neoadjuvant chemo-
Pathologically node-negative Pathologically node-positive therapy, which had to include at least six cycles of an
(pN0sn) (pN1sn) anthracycline-based regimen recommended by German
guidelines for use in the neoadjuvant setting. All patients
Neoadjuvant chemotherapy
provided written informed consent. The protocol was
reviewed by a central ethics committee (University of
Tübingen) and approved by local ethics committees and
Conversion to clinically Disease remains clinically
node-negative disease node-positive competent authorities.
(ycN0) (ycN1)
Procedures
We allocated participants to four study arms according to
Sentinel-lymph-node Sentinel-lymph-node clinical (palpation and ultrasound) axillary nodal status
No axillary-lymph-node Axillary-lymph-node
biopsy and axillary- biopsy and axillary-
dissection
lymph-node dissection lymph-node dissection
dissection before and after neoadjuvant chemotherapy (figure 1).
Arm A included patients with clinically node-negative
disease (cN0) who underwent sentinel-lymph-node
biopsy before neoadjuvant chemotherapy and received
Arm A Arm B Arm C Arm D
no further axillary surgery if they had a histologically
negative sentinel lymph node (pN0sn). Arm B included
Figure 1: SENTINA trial design patients with a positive sentinel node before neoadjuvant

2 www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9

 Articles

chemotherapy (pN1sn) who underwent a second biopsy subcutanous, peritumoral) and injected dose and
procedure after chemotherapy, followed by axillary- volume, within the frame of the national consensus, and
lymph-node dissection. Arm C included patients with we recorded these details. We did histological assessment
initially node-positive disease (cN1 or cN2) who of the sentinel lymph node, which included slicing of
underwent neoadjuvant chemotherapy and then had nodes at 2·0–3·0 mm intervals, embedding of all slices
both sentinel-lymph-node biopsy and axillary-lymph- in paraffin, and complete step-sectioning of slices at
node dissection if they converted to a clinically negative intervals of 500 μm or less.22 We did not require
axillary status (ycN0). Finally, arm D included patients immunohistochemistry to be used. We classified tumour
who had suspicious nodes before and after neoadjuvant deposits with the Union for International Cancer Control
chemotherapy (ycN1) and who subsequently received TNM classification system.24
axillary-lymph-node dissection. All patients underwent preoperative clinical assessment
We standardised the sentinel-lymph-node biopsy of lymph-node status by palpation and axillary ultrasound.
procedure according to recommendations made by We asked institutions to describe lymph-node status as
multidisciplinary consensus of the German Society for either positive (including cN1 and cN2) or negative. We
Senology and guidelines from the American Society of judged patients to be node-negative when palpation and
Clinical Oncology.22,23 Use of radiocolloid and preoperative ultrasound showed no suspicious nodes (cN0). In case of
lymphoscintigraphy was necessary for all patients, but palpable nodes, we regarded patients as node negative if
additional application of blue dye was optional. We ultrasound showed normal-sized or enlarged lymph
allowed individual choice of injection site (periareolar, nodes with regular morphological structure of hilum and

2234 patients scheduled for neoadjuvant

chemotherapy and entered into trial

1334 patients with clinically node-negative 900 patients with clinically node-positive
disease (cN0) disease (cN+)

1 screening failure
102 no second report

1334 had sentinel-lymph-node biopsy 797 received neoadjuvant chemotherapy

3 progressive disease 16 progressive disease

3 incomplete data 2 implausible data
7 no sentinel lymph node detected 1 no planned axillary surgery
before neoadjuvant chemotherapy 10 incomplete data
182 no second report 1 no breast surgery (patient’s wish)
2 withdrawal of informed consent

684 were pathologically node-negative (pN0) 455 were pathologically node-positive (pN1) 642 converted to clinically node-negative 123 remained clinically node-positive (ycN+)
disease (ycN0)

2 screening failure 1 screening failure 50 not treated per protocol

(no neoadjuvant chemotherapy) (no neoadjuvant chemotherapy)
1 withdrawal of informed consent

682 had neoadjuvant chemotherapy 453 had neoadjuvant chemotherapy

11 progressive disease 9 progressive disease

3 discontinued chemotherapy 1 discontinued chemotherapy
6 sentinel nodes, axilla, or both 5 incomplete data
found after chemotherapy 78 not treated per-protocol

662 had no axillary treatment (arm A) 360 received sentinel-lymph-node biopsy and 592 received sentinel-lymph-node biopsy and 123 received axillary-lymph-node dissection
axillary-lymph-node dissection (arm B) axillary-lymph-node dissection (arm C) (arm D)

Figure 2: Study profile

www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9 3

 Articles

defined the false-negative rate as the ratio of the number

Arm A, cN0 Arm B, cN0 Arm C, cN+ Arm D, cN+ p
pN0sn (n=662) pN1sn (n=360) ycN0 (n=592) ycN+ (n=123) of patients with a negative sentinel lymph node and one
or more positive non-sentinel lymph nodes to the number
Age (years) ·· ·· ·· ·· 0·27
of patients with at least one involved lymph node among
Mean 49 49 50 51 ··
people in whom at least one sentinel node was detected.
Median (range) 48 (20–75) 48 (26–78) 49 (22–98) 50 (29–87) ··
A secondary outcome of our study was the detection rate
Clinical tumour size before ·· ·· ·· ·· 0·014
NACT of sentinel-lymph-node biopsy before and after neo-
≤20 mm 19 (3%) 12 (3%) 21 (4%) 8 (7%) ·· adjuvant chemotherapy (in patients in arms B and C). We
>20 to ≤50 mm 499 (75%) 256 (71%) 472 (80%) 93 (76%) ·· defined successful sentinel-lymph-node detection as
>50 mm 27 (4%) 22 (6%) 46 (8%) 16 (13%) ··
surgical removal of one or more lymph nodes visualised
Unknown 117 (18%) 70 (19%) 53 (9%) 6 (5%) ··
by lymphatic mapping, using either radiocolloid alone or
Grading ·· ·· ·· ·· <0·0001
in combination with a blue dye. A further outcome of the
study was the detection rate and false-negative rate of a
G1 33 (5%) 17 (5%) 14 (2%) 5 (4%) ··
second sentinel-lymph-node biopsy procedure after identi-
G2 213 (32%) 171 (48%) 216 (36%) 46 (37%) ··
fication and removal of a positive sentinel lymph node
G3 315 (48%) 123 (34%) 258 (44%) 48 (39%) ··
before neoadjuvant chemotherapy (in patients in arm B).
Unknown 101 (15%) 49 (14%) 104 (18%) 24 (20%) ··
ER/PR status ·· ·· ·· ·· <0·0001
Statistical analysis
Both negative 260 (39%) 73 (20%) 213 (36%) 45 (37%) ··
Using data from the German multicentre validation trial
One or both positive 336 (51%) 256 (71%) 319 (54%) 61 (50%) ··
for sentinel-lymph-node biopsy in primary surgery,25 we
Unknown 66 (10%) 31 (9%) 60 (10%) 17 (14%) ··
assumed a false-negative rate of 7% in arms B and C and
HER2 status ·· ·· ·· ·· 0·0021
calculated the sample size to exclude 10% in each of
Negative 461 (70%) 236 (66%) 359 (61%) 80 (65%) ··
these arms, with a one-sided 95% CI. In every arm, we
Positive 134 (20%) 92 (26%) 173 (29%) 26 (21%) ··
needed at least 196 patients with positive nodal status
Unknown 67 (10%) 32 (9%) 60 (10%) 17 (14%) ·· (calculated with nQuery Advisor, version 6.02). Based on
Lymphovascular invasion ·· ·· ·· ·· <0·0001 findings of a pilot study, we expected that 13% of the
No 512 (77%) 241 (67%) 372 (63%) 62 (50%) ·· entire study population would have a positive axillary
Yes 38 (6%) 68 (19%) 130 (22%) 45 (37%) ·· status after neoadjuvant chemotherapy in arm B and
Unknown 112 (17%) 51 (14%) 90 (15%) 16 (13%) ·· 14% in arm C, resulting in a total number of 1508 patients
Histological tumour type ·· ·· ·· ·· 0·022 for the study. Analyses were per protocol. We used
Ductal invasive 521 (79%) 275 (76%) 476 (80%) 91 (74%) ·· Pearson χ² tests to compare rates across groups, with
Lobular invasive 41 (6%) 42 (12%) 35 (6%) 11 (9%) ·· exact Pearson 95% CIs for the false-negative rate and
Other 45 (7%) 18 (5%) 37 (6%) 10 (8%) ·· detection rate. Also, we used Wilcoxon and Kruskul-
Unknown 55 (8%) 25 (7%) 44 (7%) 11 (9%) ·· Wallis testing to compare the number of detected
sentinel lymph nodes between two groups and three
Data are number of patients (%), unless otherwise stated. NACT=neoadjuvant chemotherapy. ER=oestrogen receptor.
PR=progesterone receptor. cN0/+=clinically node-negative/positive. pN0sn/pN1sn=sentinel node pathologically groups, respectively. We did multivariate logistic
negative/positive. ycN0/+=clinically node-negative/positive after NACT. regression in arm C to find factors that affected the
detection rate and false-negative rate. We did analyses
Table 1: Patients’ characteristics
with SAS 9.2, under SAS Enterprise Guide 4.3.

cortex. No uniformly accepted standard for sonographic Role of the funding source
lymph-node assessment is available. We classified lymph The sponsor of the study had no role in study design,
nodes as suspicious in case of cortex asymmetry or loss or data collection, data analysis, data interpretation, or
displacement of the hilum relation (hilum:cortex writing of the report. TK had full access to raw data in the
ratio >2:1, or total loss of hilum). To restrict our clinical study and final responsibility for the decision to submit
findings to simple and reproducible criteria, we did not for publication.
assess cN1 and cN2 status separately. Ultrasound-guided
fine-needle aspiration or core-needle biopsy was Results
recommended but not mandatory. Patients underwent Between September, 2009, and May, 2012, 2234 patients
sentinel-lymph-node biopsy in case of negative aspiration entered the trial, and of these, 1737 women from
or core biopsy. 103 institutions fulfilled criteria for the final per-protocol
The primary outcome of our study was accuracy of analysis (figure 2). A median of eight (range 1–138)
sentinel-lymph-node biopsy (measured as the false- patients were accrued per institution. Nine institutions
negative rate) in patients who presented initially with accrued more than 50 patients each.
clinically positive lymph nodes (cN+) and subsequently 1022 (59%) of 1737 patients in the per-protocol analysis
converted to a clinically negative axillary status (ycN0) presented initially with an unsuspicious clinical lymph-
after neoadjuvant chemotherapy (those in arm C). We node status and underwent sentinel-lymph-node biopsy

4 www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9

 Articles

before neoadjuvant chemotherapy. Of these 1022 patients, institution (defined as ≥50 patients or <50 patients) did not
662 (65%) had a tumour-free sentinel lymph node affect the detection rate. However, we noted a difference in
(included in arm A) and 360 (35%) had pN1sn status the detection rate for patients who underwent a combined
(included in arm B). 715 (41%) of 1737 patients presented detection procedure (radiocolloid and blue dye) compared
with initially suspicious nodes and underwent with use of radiocolloid alone (figure 3). No difference was
neoadjuvant chemotherapy as their first treatment. Of
these, 592 (83%) converted from a clinically positive to a Arms A and B, Arm B, after Arm C, after Total (n=1974)
negative axillary status after neoadjuvant chemotherapy before NACT NACT (n=360) NACT
(included in arm C), whereas 123 (17%) had persistent (n=1022) (n=592)
suspicious nodes and underwent primary axillary-lymph- Tracer
node dissection alone after neoadjuvant chemotherapy Radiocolloid alone 580 (57%) 238 (66%) 389 (66%) 1207 (61%)
(included in arm D). Pre-treatment nodal status was Blue dye alone 7 (1%) 5 (1%) 7 (1%) 19 (1%)
confirmed in 149 (25%) of 592 women in arm C by fine- Combined 401 (39%) 105 (29%) 164 (28%) 670 (34%)
needle aspiration or core-needle biopsy. Unknown 34 (3%) 12 (3%) 32 (5%) 78 (4%)
Patients’ characteristics showed, as expected, different Injection site, multiple choice
risk profiles between groups (table 1). Individuals with Periareolar 606 (59%) 230 (64%) 396 (67%) 1232 (62%)
positive lymph nodes had more advanced tumours and Peritumoral 342 (33%) 125 (35%) 207 (35%) 674 (34%)
less favourable biological features than patients with Subcutaneous 433 (42%) 128 (36%) 276 (47%) 837 (42%)
uninvolved nodes Median injected dose (MBq) 89·8 95·0 100 94·0
Some variations in sentinel-lymph-node biopsy Protocol
procedures occurred (table 2). Exclusive use of a
1 day 334 (33%) 97 (27%) 143 (24%) 574 (29%)
radiocolloid was preferred by surgeons in 1207 (61%) of
2 days 676 (66%) 227 (63%) 415 (70%) 1318 (67%)
1974 biopsy procedures, whereas a combined detection
Unknown 12 (1·2%) 36 (10%) 34 (6%) 82 (4%)
technique with additional use of a blue dye was used in
670 (34%). The injection site varied between institutions, Data are number of patients (%), unless otherwise stated. NACT=neoadjuvant chemotherapy.
but was well balanced within groups. 76 institutions did
Table 2: Technical aspects of sentinel-lymph-node biopsy procedures in the trial
multiple injections, to various sites. The injected dose of
radiocolloid was similar in all groups.
Table 3 presents rates of detection of sentinel lymph Arms A and B Arm B Arm C p
nodes. Lymphoscintigraphy indicated a hot spot in Hot spot on 1014/1022 (99%) 236/360 (66%) 476/592 (80%) <0·0001
1014 (99·2%) of 1022 patients in arms A and B (before lymphoscintigraphy
neoadjuvant chemotherapy), 236 (65·6%) of 360 in Overall surgical detection 99·1% (1013/1022; 60·8% (219/360; 80·1% (474/592; <0·0001
arm B (after neoadjuvant chemotherapy), and 476 (80·4%) rate (n/N; 95% CI) 98·3–99·6) 55·6–65·9) 76·6–83·2)
of 592 in arm C. At surgery, one or more sentinel lymph Overall surgical detection 98·8% (573/580; 52·9% (126/238; 77·4% (301/389; ··
nodes were detected in 1013 (99·1%) of 1022 patients rate with radiocolloid 97·5–99·5) 46·4–59·4) 72·9–81·4)
alone
in arms A and B (before neoadjuvant chemotherapy), in
Overall surgical detection 99·5% (399/401; 76·2% (80/105; 87·8% (144/164; ··
219 (60·8%) of 360 patients in arm B (after neoadjuvant rate with radiocolloid and 98·2–99·9) 66·9–84·0) 81·8–92·4)
chemotherapy), and in 474 (80·1%) of 592 patients in blue dye
arm C. The median number of sentinel lymph nodes Sentinel lymph nodes
removed was two, which was similar between groups. removed
When sentinel-lymph-node biopsy was done before 0 9/1022 (1%) 141/360 (39%) 118/592 (20%) ··
neoadjuvant chemotherapy (arms A and B), no difference 1 284/1022 (28%) 96/360 (27%) 142/592 (24%) ··
in the detection rate was recorded between the combined 2 294/1022 (29%) 56/360 (16%) 131/592 (22%) ··
(radiocolloid and blue dye) and single-agent (radiocolloid 3 186/1022 (18%) 22/360 (6%) 81/592 (14%) ··
alone) detection techniques (99·5% [399 of 401] vs 98·8% 4 114/1022 (11%) 20/360 (6%) 59/592 (10%) ··
[573 of 580]; table 3). When sentinel-lymph-node biopsy >4 135/1022 (13%) 25/360 (7%) 61/592 (10%) ··
was done after neoadjuvant chemotherapy, in arms B At least one sentinel node
and C, the additional use of blue dye was associated with removed
a significant increase in the detection rate (76·2% All patients Mean 2·7, median 2·0 Mean 2·4, median Mean 2·7, median <0·0001
2·0 2·0
[80 of 105] vs 52·9% [126 of 238] in arm B, and 87·8%
[144 of 164] vs 77·4% [301 of 389] in arm C) and a higher Radiocolloid alone Mean 2·6, median Mean 2·3, median Mean 2·6, median 0·012
number of nodes were detected in arm C (median three 2·0 2·0 2·0

vs two in arm B; p=0·0059). Radiocolloid and blue Mean 2·8, median Mean 2·6, median Mean 2·9, median 0·059
dye 2·0 2·0 3·0
In arm C, we did a multivariate analysis of factors that
could potentially affect the detection rate (figure 3). Data are n/N (%), unless otherwise stated.
Morphological or biological features, tumour involvement
Table 3: Detection of sentinel lymph nodes, according to selected factors
of the sentinel lymph node, or the case load of the

www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9 5

 Articles

Odds ratio (95% CI) p

recorded in the detection rate in patients with lymph-node
Lobular tumour 2·46 (0·609–9·98) 0·206
involvement confirmed by fine-needle aspiration or core-
G3 1·37 (0·766–2·46) 0·287
needle biopsy before neoadjuvant chemotherapy in arm C,
Unifocal tumour 1·26 (0·670–2·36) 0·475
compared with those who underwent an exclusively
L0 1·42 (0·703–2·85) 0·331
clinical assessment of axillary nodal status (82·6% [123 of
V0 2·57 (0·586–11·3) 0·211
149] vs 79·2% [351 of 443]; p=0·41).
No extracapsular extension 2·11 (0·888–5·00) 0·091
False-negative rates of sentinel-lymph-node biopsy
ER/PR negative 0·863 (0·467–1·60) 0·639
were compared between arms B and C (table 4). For
HER2-negative 1·31 (0·741–2·33) 0·351
patients in arm A, the false-negative rate could not be
pN0 1·39 (0·694–2·78) 0·353
ascertained, because these patients did not undergo
Large centre 1·22 (0·695–2·12) 0·494
axillary-lymph-node dissection. Of 219 patients in arm B
Radiocolloid and blue dye 2·13 (1·01–4·46) 0·046
who had a successful second sentinel-lymph-node
No pCR 1·20 (0·600–2·41) 0·604
biopsy procedure after neoadjuvant chemotherapy,
155 (70·8%) had ypN0 status and 64 (29·2%) were
0·25 0·50 1·00 2·0 4·0 8·0 16·0 ypN1. Sentinel-lymph-node biopsy was false-negative in
33 (51·6%) of these 64 patients. Of 474 patients in
Detection rate lower Detection rate higher
arm C who converted from clinically positive to negative
Figure 3: Multivariate regression analysis for detection rate (arm C) axillary status after neoadjuvant chemotherapy and
G3=grade 3. L0=no lymphovascular invasion. V0=no vascular invasion. ER/PR=oestrogen receptor/progesterone had a successful sentinel-lymph-node procedure,
receptor. pN0=pathologically node-negative. pCR=pathological complete response. 248 (52·3%) had ypN0 status and 226 (47·7%) were
ypN1. Sentinel-lymph-node biopsy was false-negative in
Arm B (n=64) Arm C (n=226) 32 (14·2%) of these 226 patients.
Overall false-negative rate (n/N; 95% CI) 51·6% (33/64; 38·7–64·2) 14·2% (32/226; 9·9–19·4) The median number of resected non-sentinel lymph
False-negative rate, according to number of sentinel nodes removed nodes was 11 (range 0–40) in arm B and 13 (0–44) in
1 66·7% (16/24) 24·3% (17/70)
arm C. Lymph-node involvement was restricted to the
2 53·8% (7/13) 18·5% (10/54)
sentinel lymph node (or nodes) in 45 (70·3%) of
3 50·0% (5/10) 7·3% (3/41)
64 patients after neoadjuvant chemotherapy in arm B
and 131 (58·0%) of 226 patients in arm C.
4 50·0% (3/6) 0·0% (0/28)
For patients in arm C who converted from clinically
5 18·2% (2/11) 6·1% (2/33)
positive disease to negative axillary status after
False-negative rate, according to detection technique
neoadjuvant chemotherapy, we did a multivariate
Radiocolloid alone 46·2% (18/39) 16·0% (23/144)
analysis of factors that could potentially affect the false-
Radiocolloid and blue dye 60·9% (14/25) 8·6% (6/70)
negative rate (figure 4). A significant relation was seen
Data are rate (number of patients), unless otherwise stated. between the number of resected sentinel lymph nodes
and the false-negative rate (figure 4). Patients who had
Table 4: False-negative rate of sentinel-lymph-node resection in patients with positive nodes, according
to selected factors
one sentinel node removed had a false-negative rate of
24·3% (17 of 70) and those who had two removed had a
false-negative rate of 18·5% (ten of 54). The false-
Odds ratio (95% CI) p negative rate was consistently less than 10% for patients
Lobular tumour 0·123 (0·006–2·37) 0·165 who had three or more sentinel lymph nodes removed.
G3 1·75 (0·553–5·53) 0·342 The false-negative rate was 8·6% (six of 70) for patients
Unifocal tumour 1·06 (0·335–3·35) 0·921 who underwent the combined detection method (with
L0 1·83 (0·591–5·67) 0·295 radiocolloid and blue dye) compared with 16·0%
V0 2·16 (0·058–80·4) 0·678 (23 of 144) for those who received radiocolloid alone;
No extracapsular extension 0·430 (0·110–1·68) 0·225 however, this difference was not significant after
ER/PR negative 0·980 (0·299–3·20) 0·973 multivariate analysis (table 4, figure 4). We did not
HER2-negative 1·78 (0·511–6·19) 0·366 identify a difference in the false-negative rate for
Large centre 0·653 (0·222–1·92) 0·437 patients with lymph-node involvement confirmed by
Number of sentinel nodes (per 1 sentinel node) 0·487 (0·287–0·825) 0·008 either fine-needle aspiration or core-needle biopsy
Radiocolloid and blue dye 0·353 (0·087–1·43) 0·145 before neoadjuvant chemotherapy in arm C compared
No pCR 1·92 (0·323–11·5) 0·472 with individuals with an exclusively clinical assessment
of lymph-node status (19·0% [12 of 63] vs 12·3%
0·001 0·01 0·1 1·0 10 100
[20 of 163]; p=0·21).
Lower false-negative rate Higher false-negative rate

Figure 4: Multivariate regression analysis for false-negative rate (arm C)

Discussion
G3=grade 3. L0=no lymphovascular invasion. V0=no vascular invasion. ER/PR=oestrogen receptor/progesterone Our findings show that sentinel lymph nodes were
receptor. pCR=pathological complete response. detected in almost all patients who were clinically

6 www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9

 Articles

node-negative and underwent sentinel-lymph-node A successful mapping procedure (ie, identification of

biopsy before neoadjuvant chemotherapy (ie, those a hot spot on lymphoscintigraphy) and a near 100%
women in arms A and B), whereas the detection rate was sentinel-lymph-node detection rate confirms the
80·1% for patients who converted after chemotherapy feasibility of sentinel-lymph-node biopsy before neo-
from clinically positive to negative axillary status (ie, adjuvant chemotherapy (ie, in women in arms A and B),
those in arm C). The overall false-negative rate was even in patients with advanced tumours scheduled for
14·2% for patients who converted, and it was around neoadjuvant chemotherapy. However, for women whose
20% if only one or two sentinel nodes were harvested. initial sentinel-lymph-node biopsy revealed a positive
Use of additional blue dye with radiocolloid lowered the histological finding and who underwent a second biopsy
false-negative rate. In patients with clinically negative procedure after chemotherapy (ie, those in arm B), no
disease but a histologically positive sentinel lymph node more than two-thirds of sentinel nodes were detected on
before systemic treatment (ie, those in arm B), fewer lymphoscintigraphy or at surgery. These results contrast
than two-thirds of sentinel lymph nodes were detected with data obtained in fewer patients by Khan and
after a second sentinel-lymph-node biopsy procedure colleagues,21 of a detection rate of 97% for patients in
undertaken after neoadjuvant chemotherapy, and the this setting.
false-negative rate was 51·6%. In women who converted from a clinically positive to
A strength of our study was the strict standardisation negative axillary status after neoadjuvant chemotherapy
of conditions for the sentinel-lymph-node biopsy (those in arm C), rates of identification of a hot spot and
procedure and the use of predefined chemotherapy surgical detection of a sentinel node were both around
regimens. Sentinel-lymph-node biopsy was undertaken 80%, significantly lower than in patients who underwent
according to interdisciplinary consensus of the German sentinel-lymph-node biopsy before systemic treatment.
Society for Senology, an organisation that has defined Data for the rate of detection of sentinel-lymph-node
evidence-based standards and scopes of action for the biopsy after neoadjuvant chemotherapy in patients who
technical performance of sentinel-lymph-node biopsy present initially with a positive axillary status are
and the histopathological assessment of sentinel lymph conflicting.14–19 Our findings confirm results from smaller
nodes.22 Use of a radiocolloid is an objective and series, in which less favourable detection rates were
measurable technique with highly reproducible success noted after neoadjuvant chemotherapy compared with
rates. This technique is, therefore, defined as a sentinel-lymph-node biopsy at primary surgery;14
minimum standard in the German consensus. Use of furthermore, Classe and colleagues20 reported a detection
blue dye is judged optional because success rates in rate of 81·5% in a subgroup analysis of a small
scientific literature are less reliable and the classification prospective trial that also addressed the feasibility of
of a lymph node as blue is subjective and less sentinel-lymph-node biopsy after neoadjuvant chemo-
reproducible. 19 patients had blue dye exclusively in our therapy in clinically downstaged patients.
study. These patients were not treated according to the The application of a radiocolloid with subsequent
protocol but we included this subgroup (<1%) in the lymphoscintigraphy is a reproducible and objective
final analysis. procedure. Our data suggest that tracer uptake in the
We could not compare false-negative rates directly sentinel lymph node is less favourable in patients who
before or after neoadjuvant chemotherapy in our study are downstaged after neoadjuvant chemotherapy from
because patients in arm A were excluded from axillary positive to negative axillary status, compared with
dissection. For women who had successful detection of sentinel-lymph-node biopsy in primary surgery. In 7·3%
sentinel lymph nodes during the biopsy procedure, the of patients in arm C, surgical detection failure was
median number of removed sentinel nodes was two, and recorded despite a hot spot on lymphoscintigraphy,
this number did not differ between arms. Most previous suggesting that sentinel-lymph-node biopsy after neo-
trials that have addressed the feasibility and accuracy of adjuvant chemotherapy might be technically more
sentinel-lymph-node biopsy in primary surgery or in the challenging in this scenario. Lymph-node involvement
neoadjuvant setting have yielded similar results. did not affect the detection rate in univariate and
Clinical assessment of axillary status was an important multivariate analyses, suggesting that impairment of
issue in our study. Fine-needle aspiration or core-needle sentinel-lymph-node detection is instead related to the
biopsy of axillary lymph nodes are not clinical standards effect of chemotherapy (fibrosis, obstruction of
in international guidelines and, thus, are not widely lymphatic channels, etc) than to the tumour burden in
used. We, therefore, restricted lymph-node assessment the lymph nodes.
to palpation and ultrasound and recommended fine- In our study, a second sentinel-lymph-node biopsy
needle aspiration or core-needle biopsy as optional. Our procedure after neoadjuvant chemotherapy in women
study clearly showed the limitations of exclusively clinical with a positive sentinel node before systemic treatment
lymph-node assessment and provides a rationale to (ie, those in arm B) resulted in a false-negative rate of
promote cytological or histological confirmation of 51·6%, indicating that this procedure is not a useful
clinically suspicious lymph nodes. option. We could not confirm the results of Khan and

www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9 7

 Articles

colleagues,21 who reported a false-negative rate of 4·5% indicate that sentinel-lymph-node biopsy is less accurate
in this group of patients. The low detection rate and high in the specific setting of initially node-positive patients
false-negative rate for repeat sentinel-lymph-node biopsy who undergo sentinel-lymph-node biopsy after
after neoadjuvant chemotherapy can be explained by the neoadjuvant chemotherapy, particularly if only one or
notion of lymphatic mapping, which is based on an two sentinel nodes are identified.
orderly lymphatic spread of tumour cells. Possibly, In our study, addition of blue dye to radiolabelled
additional lymph-node involvement cannot be detected tracer could not improve further the detection rate for
once the primary drainage has been destroyed. This sentinel-lymph-node biopsy before surgery because
situation is, however, different to a second sentinel- radiocolloid alone already had a good detection rate. In
lymph-node biopsy procedure in recurrent disease. By patients with initially suspicious nodes who underwent
contrast with the setting investigated in our study, sentinel-lymph-node biopsy after neoadjuvant chemo-
tumour spread from a de-novo lesion happens after therapy (those in arm C), the additional blue dye resulted
preceding axillary surgery. Our results, therefore, cannot in a significant improvement in detection rate, from
be transferred to this clinical setting. 77·4% to 87·8%. This result is similar to data from
In patients whose axillary status converted from Mamounas and colleagues,27 who reported an increase in
clinically positive to clinically negative after neoadjuvant the detection rate after neoadjuvant chemotherapy, from
chemotherapy (those in arm C), we noted a false-negative 78·1% to 87·6%, if a blue dye was used in addition to
rate of 14·2% for sentinel-lymph-node biopsy. Compared radiocolloid. In arm C, we noted a non-significant
with our large prospective study, findings of smaller association between the false-negative rate and mapping
retrospective series show a wide range of false-negative technique (8·6% for combined tracer vs 16·0% for
rates (5–35%) for this group of patients.14–18 Our findings radiocolloid alone). Similar findings were reported from
accord with a subgroup analysis of the only prospective ACOSOG Z1071.26 Use of a combined tracer was
trial we could identify, by Classe and colleagues,20 who associated with a false-negative rate of 10·8%, whereas
reported a false-negative rate of 15%. the single tracer technique gave false-negative rates of
In several trials addressing the accuracy of sentinel- 22·2% for use of blue dye and 20·0% for radiocolloid
lymph-node biopsy at primary surgery for breast cancer, alone. We also noted a significantly higher lymph-node
false-negative rates of less than 10% have been noted yield for patients who underwent the dual-tracer
consistently.1,8,25 The accuracy of sentinel-lymph-node technique (median three sentinel nodes vs two for
biopsy after neoadjuvant chemotherapy for patients with radiocolloid alone), thus, whether the lower false-
initially positive lymph nodes is, therefore, less negative rate associated with the combined tracer is the
favourable compared with sentinel-lymph-node biopsy effect of the mapping technique or an effect of the
before systemic treatment. additional lymph-node harvest is unclear.
The accuracy of sentinel-lymph-node biopsy was Since sentinel-lymph-node biopsy has good feasibility
closely related to the number of sentinel nodes removed before systemic treatment, this procedure remains an
in our study. This association has been shown in several attractive option for clinically node-negative patients
trials of primary surgery and in the neoadjuvant scheduled for neoadjuvant chemotherapy. About two-
setting.6,17,22 In the National Surgical Adjuvant Breast and thirds of patients do not need axillary-lymph-node
Bowel Project (NSABP) protocol B32 study,8 a false- dissection if sentinel-lymph-node biopsy is undertaken
negative rate of 9·8% was reported for all patients who before neoadjuvant chemotherapy and shows uninvolved
underwent sentinel-lymph-node biopsy in primary nodes. Even in some cases of nodal involvement, patients
surgery. However, women with only one detected might be spared from axillary dissection, according to
sentinel node had a false-negative rate of 17·7%, the results of ACOSOG Z0011.28 In this trial, no benefit of
whereas those with at least two removed sentinel lymph axillary-lymph-node dissection was seen in patients with
nodes had a false-negative rate of 10·0% or better. In T1/T2 tumours and one or two involved sentinel lymph
our study, we recorded a false-negative rate of 14·2% for nodes who underwent breast-conserving treatment and
all patients who converted axillary status from positive radiotherapy to the breast. Compared with sentinel-
to negative after neoadjuvant chemotherapy (ie, those in lymph-node biopsy in primary surgery, the feasibility and
arm C). The group of women who had only one sentinel accuracy of sentinel-lymph-node biopsy after neoadjuvant
node resected had a false-negative rate of 24·3% and the chemotherapy is less favourable if patients are
group with two removed sentinel nodes had a false- downstaged from an initially positive to a negative
negative rate of 18·5%. Data presented recently from axillary status. However, this group of patients might
the American College of Surgeons Oncology Group derive significant benefit from sentinel-lymph-node
(ACOSOG) Z1071 study26 support our findings. For cN1 biopsy after neoadjuvant chemotherapy, because more
patients who underwent sentinel-lymph-node biopsy than 50% of them might be spared from either axillary
after neoadjuvant chemotherapy, the false-negative rate dissection or regional radiotherapy, or both. By contrast,
was 31·5% and 21·1%, respectively, if only one or two patients with one or two removed sentinel lymph nodes
sentinel lymph nodes were harvested. These results have an unacceptable false-negative rate, so this approach

8 www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9

 Articles

chemotherapy can be directly compared and whether the

Panel: Research in context clinical outcome is similar is also unclear. Patients who
Systematic review undergo sentinel-lymph-node biopsy in the setting of
We searched PubMed from January, 1997, to March, 2013, primary surgery will receive systemic treatment
with the terms “breast cancer”, “sentinel lymph node biopsy”, afterwards, which might erase undetected lymph-node
and “neoadjuvant chemotherapy”. We did not restrict our metastases, whereas systemic treatment is accomplished
search by language. We analysed 252 articles that matched in patients who receive sentinel-lymph-node biopsy after
our query. We did not identify a prospective study that had as neoadjuvant chemotherapy (apart from eventual
its primary endpoint the detection rate and false-negative additional endocrine treatment).
rate of patients who converted during neoadjuvant In conclusion, our findings show that sentinel-lymph-
chemotherapy from clinically positive to clinically negative node biopsy provides less reliable results if it is
axillary status. Furthermore, we did not find prospective undertaken after neoadjuvant chemotherapy in patients
comparative evidence for the detection rate of who convert from a positive to a negative axillary status,
sentinel-lymph-node biopsy before and after neoadjuvant compared with women who undergo sentinel-lymph-
chemotherapy, and we did not retrieve any prospective node biopsy in primary surgery (panel). The clinical
studies of the feasibility and accuracy of a second significance of these restricted success rates is, however,
sentinel-lymph-node biopsy procedure. unclear. Patient selection and, eventually, use of a
combined tracer might improve the false-negative rate.
Interpretation Selection of patients who can be spared from further
The findings of SENTINA suggest that detection rate and regional treatment after an initially positive lymph-node
accuracy (false-negative rate) of sentinel-lymph-node biopsy status remains a clinical challenge. Further trials are
are inferior for patients who convert during neoadjuvant needed that address the clinical effect of the increased
chemotherapy from an initially positive axillary status to false-negative rate after neoadjuvant chemotherapy. Our
clinically negative disease, compared with those who undergo findings confirm sentinel-lymph-node biopsy as a
sentinel-lymph-node biopsy in primary surgery. The accuracy reliable diagnostic method in the context of neoadjuvant
of sentinel-lymph-node biopsy is especially unfavourable in chemotherapy.
patients with only one or two harvested sentinel lymph
Contributors
nodes after neoadjuvant chemotherapy. Addition of blue dye TK was principal investigator and designed the study. TK, GH, and CL
might improve the accuracy of sentinel-lymph-node biopsy. contributed to the literature search, data collection, data analysis, and
A second sentinel-lymph-node biopsy after neoadjuvant data interpretation. IB, TF, PS, and MU contributed to data collection,
data analysis, and data interpretation. BF, AL, GvM, and AS contributed
chemotherapy, in patients with histologically proven
to data analysis and data interpretation. MH and SS contributed to data
sentinel-lymph-node involvement before systemic collection. VN did the sample size calculation and statistical analyses
treatment, is not a good clinical option. and designed figures and tables. All authors wrote the report and have
approved the final version of the manuscript.

should be restricted to individuals with more than two Conflict of interest

We declare that we have no conflicts of interest.
sentinel lymph nodes removed. This restriction, however,
would have excluded 66% of patients within our trial Acknowledgments
We received financial and logistical support from Arbeitsgemeinschaft
from this procedure. Use of a dual tracer might für Gynäkologische Onkologie–Breast, the German Breast Group, and
moderately increase the false-negative rate, although Brustkrebs Deutschland. We thank Renate Haidinger (Brustkrebs
data in scientific literature are controversial. Classe and Deutschland) for general support; Angelika Jursik (study coordinator,
colleagues20 reported a 15% false-negative rate, although Interdisciplinary Breast Center, Esslingen, Germany), Jana Ulbrich
(Interdisciplinary Breast Center Esslingen, Germany), and
a combined detection method was used in all patients. Susanne Seynaeve (Interdisciplinary Breast Center Gifhorn, Germany)
Development of a specific nomogram could be an option for data management; and Malgorzata Banys (Marienkrankenhaus
to identify patients with a high risk of further involvement Hamburg, Germany) and Mustafa Anjari (Department of Neurosurgery,
St George’s Hospital, London, UK) for help with the English text.
of non-sentinel lymph nodes.
A specific difficulty is the clinical rating of a particular References
1 Kim T, Giulliano AE, Lyman GH. Lymphatic mapping and sentinel
false-negative rate after neoadjuvant chemotherapy. On lymph node biopsy in early stage breast carcinoma: a metaanalysis.
one hand, whether inaccurate sentinel-lymph-node biopsy Cancer 2006; 106: 4–16.
translates into an increased recurrence rate is unclear. For 2 Haid A, Kuehn T, Konstantiniuk P, et al. Shoulder-arm morbidity
following axillary dissection and sentinel node only biopsy for
instance, the NSABP-B32 trial included patients with only breast cancer. Eur J Surg Oncol 2002, 28: 705–10.
one resected sentinel node, who had a false-negative rate 3 Helms G, Kühn T, Moser L, Remmel E, Kreienberg R.
of 17·7%. The unfavourable accuracy in this subgroup did Shoulder-arm morbidity in patients with sentinel node biopsy and
complete axillary dissection: data from a prospective randomised
not translate into a clinically relevant regional recurrence trial. Eur J Surg Oncol 2009; 35: 697–701.
rate. This observation could support the idea that a much 4 Kaufmann M, von Minckwitz G, Bear HD, et al. Recommendation
higher false-negative rate than the generally accepted rate from an international expert panel on the use of neoadjuvant
of 10% would be justifiable. On the other hand, whether (primary) systemic treatment of operable breast cancer: new
perspectives, 2006. Ann Oncol 2007; 18: 1927–34.
false-negative rates before and after neoadjuvant

www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9 9

 Articles

5 Fisher B, Brown A, Mamounas E, et al. Effect of preoperative 17 Shen J, Gilcrease MZ, Babiera GV, et al. Feasibility and accuracy of
chemotherapy on local-regional disease in women with operable sentinel lymph node biopsy after preoperative chemotherapy in
breast cancer: findings from National Surgical Adjuvant Breast and breast cancer patients with documented axillary metastases.
Bowel Project B-18. J Clin Oncol 1997; 15: 2483–93. Cancer 2007; 109: 1255–63.
6 Kuerer HM, Sahin AA, Hunt KK, et al. Incidence and impact of 18 Takahashi M, Jinno H, Hayashida T, Sakata M, Asakura K,
documented eradication of breast cancer axillary lymph node Kitagawa Y. Correlation between clinical nodal status and sentinel
metastases before surgery in patients treated with neoadjuvant lymph node biopsy false negative rate after neoadjuvant
chemotherapy. Ann Surg 1999; 230: 72–78. chemotherapy. World J Surg 2012; 36: 2847–52.
7 Krag DN, Anderson SJ, Julian TB, et al, for the National Surgical 19 Alvarado R, Yi M, Le-Petross H, et al. The role for sentinel lymph
Adjuvant Breast and Bowel Project (NSABP). Technical outcomes of node dissection after neoadjuvant chemotherapy in patients who
sentinel-lymph-node resection and conventional axillary-lymph- present with node-positive breast cancer. Ann Surg Oncol 2012;
node dissection in patients with clinically node-negative breast 19: 3177–84.
cancer: results from the NSABP B-32 phase III trial. Lancet Oncol 20 Classe JM, Bordes V, Campion L, et al. Sentinel lymph node biopsy
2007; 8: 881–88. after neoadjuvant chemotherapy for advanced breast cancer: results
8 Krag DN, Anderson SJ, Julian TB, et al. Sentinel-lymph-node of Ganglion Sentinelle et Chimiotherapie Neoadjuvante, a French
resection compared with conventional axillary-lymph-node prospective multicentric study. J Clin Oncol 2009; 27: 726–32.
dissection in clinically node-negative breast cancer: overall survival 21 Khan A, Sabel MS, Nees A, et al. Comprehensive axillary evaluation
findings from the NSABP B-32 randomised phase 3 trial. in neoadjuvant chemotherapy patients with ultrasonography and
Lancet Oncol 2010; 11: 927–33. sentinel lymph node biopsy. Ann Surg Oncol 2005; 12: 697–704.
9 Rouzier R, Extra JM, Klijanienko J, et al. Incidence and prognostic 22 Kuehn T, Bembenek A, Decker T, et al, for the Consensus
significance of complete axillary downstaging after primary Committee of the German Society of Senology. A concept for the
chemotherapy in breast cancer patients with T1 to T3 tumors and clinical implementation of sentinel lymph node biopsy (SLNB) in
cytologically proven axillary metastatic lymph nodes. J Clin Oncol breast cancer patients with special regard to quality assurance.
2002; 20: 1304–10. Cancer 2005; 103: 451–61.
10 Smith BD. Using chemotherapy response to personalise choices 23 Lyman GH, Giuliano AE, Somerfield MR, et al. American Society of
regarding locoregional therapy: a new era in breast cancer Clinical Oncology guideline recommendations for sentinel lymph
treatment? J Clin Oncol 2012; 30: 3913–14. node biopsy in early stage breast cancer. J Clin Oncol 2005;
11 Xing Y, Foy M, Cox DD, Kuerer HM, Hunt KK, Cormier JN. 23: 7703–20.
Meta-analysis of sentinel lymph node biopsy after preoperative 24 Sobin L, Gospodarowicz M, Wittekind C. UICC TNM classification
chemotherapy in patients with breast cancer. Br J Surg 2006; of malignant tumours, 7th edn. New York: Wiley-Liss, 2009.
93: 539–46. 25 Kuehn T, Vogl FD, Helms G, et al. Sentinel-node-biopsy is a reliable
12 Kelly AM, Dwamena B, Cronin P, Carlos RC. Breast cancer sentinel method for axillary staging in breast cancer: results from a large
node identification and classification after neoadjuvant prospective German multi-institutional trial. Eur J Surg Oncol 2004;
chemotherapy: systematic review and meta analysis. Acad Radiol 30: 252–56.
2009; 16: 551–63. 26 Boughey JC, Suman VJ, Mittendorf EA, et al. The role of sentinel
13 Van Deurzen CH, Vriens BE, Tjan-Heijnen VC, et al. Accuracy of lymph node surgery in patients presenting with node positive
sentinel node biopsy after neoadjuvant chemotherapy in breast breast cancer (T0-T4, N1-N2) who receive neoadjuvant
cancer patients: a systematic review. Eur J Cancer 2009; 45: 3124–30. chemotherapy: results from ACOSOG Z1071 trial. Cancer Res 2012;
14 Lee S, Kim EY, Kang SH, et al. Sentinel node identification rate, but 72 (24 suppl): abstr 94s.
not accuracy, is significantly decreased after preoperative 27 Mamounas EP, Brown A, Anderson S, et al. Sentinel node biopsy
chemotherapy in axillary node-positive patients. after neoadjuvant chemotherapy in breast cancer: results from
Breast Cancer Res Treat 2007; 102: 283–88. NSABP protocol B-27. J Clin Oncol 2005; 23: 2694–702.
15 Newman EA, Sabel MS, Nees AV, et al. Sentinel lymph node biopsy 28 Giuliano AE, McCallL, Beitsch P, et al. Locoregional recurrence
performed after neoadjuvant chemotherapy is accurate in patients after sentinel lymph node dissection with or without axillary
with documented node-positive breast cancer at presentation. dissection in patients with sentinel lymph node metastases.
Ann Surg Oncol 2007; 14: 2946–52. Ann Surg 2010; 252: 426–33.
16 Le Bouëdec G, Geissler B, Gimbergues P, et al. Sentinel lymph
node biopsy for breast cancer after neoadjuvant chemotherapy:
influence of nodal status before treatment. Bull Cancer 2006;
93: 415–19 [in French].

10 www.thelancet.com/oncology Published online May 15, 2013 http://dx.doi.org/10.1016/S1470-2045(13)70166-9