Sentinel node involvement with or without completion axillary lymph node dissection: treatment and pathologic results of randomized SERC trial

Based on results of clinical trials, completion ALND (cALND) is frequently not performed for patients with breast conservation therapy and one or two involved sentinel nodes (SN) by micro- or macro-metastases. However, there were limitations despite a conclusion of non-inferiority for cALND omission. No trial had included patients with SN macro-metastases and total mastectomy or with >2 SN macro-metastases. The aim of the study was too analyze treatment delivered and pathologic results of patients included in SERC trial. SERC trial is a multicenter randomized non-inferiority phase-3 trial comparing no cALND with cALND in cT0-1-2, cN0 patients with SN ITC (isolated tumor cells) or micro-metastases or macro-metastases, mastectomy or breast conservative surgery. We randomized 1855 patients, 929 to receive cALND and 926 SLNB alone. No significant differences in patient’s and tumor characteristics, type of surgery, and adjuvant chemotherapy (AC) were observed between the two arms. Rates of involved SN nodes by ITC, micro-metastases, and macro-metastases were 5.91%, 28.12%, and 65.97%, respectively, without significant difference between two arms for all criteria. In multivariate analysis, two factors were associated with higher positive non-SN rate: no AC versus AC administered after ALND (OR = 3.32, p < 0.0001) and >2 involved SN versus ≤2 (OR = 3.45, p = 0.0258). Crude rates of positive NSN were 17.62% (74/420) and 26.45% (73/276) for patient’s eligible and non-eligible to ACOSOG-Z0011 trial. No significant differences in patient’s and tumor characteristics and treatment delivered were observed between the two arms. Higher positive-NSN rate was observed for patients with AC performed after ALND (17.65% for SN micro-metastases, 35.22% for SN macro-metastases) in comparison with AC administered before ALND.


INTRODUCTION
The most commonly accepted prognostic factors for proposing adjuvant systemic therapy in breast cancer (BC) include patient age, tumor size, axillary lymph node status, tumor pathology including grade, lymphovascular invasion (LVI), endocrine receptor (ER) status, Her2 status, and proliferation assays such as the Ki67 labeling index [1][2][3][4][5] . Axillary lymph node involvement remains a major prognostic factor. Sentinel lymph node biopsy (SLNB) is the recommended surgical procedure for patients with BC with clinically N0 status based on results of randomized studies 6,7 without completion axillary lymph node dissection (cALND).
Axillary lymph node involvement rate in early BC is about 35-36%, with 3% pN0(i+), 8% pN1mi, and 24% pN1 8 . Since results of randomized ACOSOG-Z0011 trial, cALND is less frequently performed for patients with primary breast conservation therapy 1 and one or two involved sentinel nodes (SNs) by micro-or macrometastases without extracapsular extension with whole breast radiotherapy and systemic adjuvant treatment (endocrine therapy and or chemotherapy) 9 . Results of IBCSG 23-01 confirmed the non-inferiority of cALND omission for patients with SN isolated tumor cells (ITC) or micro-metastases 10 . However, some limitations to conclude non-inferiority of cALND omission in comparison with cALND were reported, particularly for patients with SN micrometastases and total mastectomy, related to the very low number of patients included in this situation in IBCSG 23-01 trial. Moreover, neither trial had included patients with SN macro-metastases and total mastectomy or patients with more than two involved SN by macro-metastases and or extracapsular extension.
Before results with 10-years follow-up of these two randomized trial, we started the SERC randomized trial of cALND versus no cALND for patients with SN involvement, whatever the size of metastasis, and with breast conservative surgery or mastectomy. The aim of this study was to analyze treatment delivered and pathologic results of patients included in SERC trial.

Patient accruals and characteristics
The first 1855 randomized patients were accrued from 53 centers ( Fig. 1), of whom 929 were randomized to receive cALND and 926 SLNB alone (Fig. 2).
Of the 1855 randomized patients, the actual treatment received was documented in 1823 patients: 840 patients had SLNB followed by ALND, 983 patients had SLNB alone, and the status of ALND was missing in 32 patients. One hundred and eight patients (5.92%) did not receive the study treatment as randomized: 90 in the ALND arm did not have ALND and 18 patients in the SLNB-alone arm had ALND. The main reason for not adhering to the randomized arm was due to the patient's decision in the ALND arm (83/90), and the investigator's decision in the SLNB alone arm (11/18), Overall, these protocol deviations were in relation with patient's decision in 83 cases (11 for ITC SN, 24 for micro-metastases, 39 for macro-metastases, and 9 with unknown SN status), in relation with investigator's decision in 18 cases (1 ITC, 4 pN1mi, 12 macrometastases, and 1 with unknown SN status) and others reasons in 7 cases (1 pN1mi, 4 macro-metastases, and 2 unknown SN status).
In summary, in the ALND arm 12 ITC (   For 149 patients with ER-negative tumors, AC was administered in 144 patients (95%: 74 in the ALND group and 75 in the SLNB group) and no AC was delivered in three patients of the ALND group and five patients of the SLNB group.
Crude rates of positive NSN were 17.6% (74/420) and 26.5% (73/ 276) for patient's eligible and non-eligible to Z0011 trial, respectively, 9.8% (23/235) for patients eligible to IBCSG-23-01 trial (no patient with positive-NSN among 8 patients non-eligible). Some minor differences of patient's characteristics between center 1 and others centers is report in Supplemental Data File 5. A very few number of events (n = 102) were reported. Consequently higher follow-up is necessary to analyze the noninferiority of cALND omission. In both Z0011 and IBCSG-23-01 trials under 50% of the initially estimated population in order to demonstrate non-inferiority of omission of cALND were included. SERC trial annually included   One hundred and fourteen patients (6.04%) did not receive the allocated treatment. Ninety-five patients randomized in the ALND group did not undergo ALND and 19 patients randomized in the SLND group underwent ALND (2.0%).
These findings are consistent with results of other trials: 43 patients in Z0011 trial did not receive the allocated treatment, 32 in the ALND arm and 11 in the SLND arm; and 31 patients in IBCSG-23-01: with 17 in the ALND group and 14 in the SLND group.
Regarding SN results the following results were found: 1105 (65.97%) macro-metastasis, 471 micro-metastasis (28.12%), and 99 ITCs (5.91%). Nine hundred of our patients were eligible for Z0011 trial: 591 (65.5%) had macro-metastasis versus 430 (50.2%) in Z0011, 311 (34.5%) had micro-metastasis versus 301 (35.2%) in Z0011, and 125 (14.6%) had unknown SN status. In the Z0011 trial  Bold entries indicate statistical significant differences.  11 . There were 21.2% patients in our trial who presented with NSN involvement, with 17.62% eligible for Z0011 and 9.79% for IBCSSG-23-01. NSN involvement rates were 27.3, 7.6, and 13% for Z0011, IBCSG-23-01, and AATRM. The difference in terms of NSN involvement rates between our study and Z0011 could be attributed to the higher proportion of patients undergoing ALND after chemotherapy (21.3%: 178/835). Indeed chemotherapy significantly reduces NSN involvement for patients undergoing ALND after chemotherapy and is therefore responsible for downstaging when compared to patients receiving AC or no chemotherapy. The ACOSOG-Z1071 trial 12 reported comparable findings with a 41% downstaging after NAC. The SENTINA trial reported a 17.8% positive NSN rate for cN0 patients presenting with a negative pre-therapeutic SN 13 . Park et al. 14 reported a 40.8% downstaging attributed to NAC for patients with positive axillary nodal cytology. These crucial findings can partly explain the very low rate of axillary recurrence in patients for whom cALND was not performed 9,10,15,16 . The following chemotherapy rates were reported: 65.4% in our study, 57.9% in Z0011, 69.4% in IBCSG-23-01, and 92.1% in AATRM.
Tangent radiation fields have an important effect on axillary control with a 10-year axillary recurrence rate of 0.08% for WBI and 0.754% for partial breast radiation (HR 0.25: 0.08-0.75) 13 . In all, 86.3% patients received breast or chest wall radiotherapy, and 89.3%, 89.7%, and 80.6% after breast conservative surgery in Z0011, AATRM, and IBCSG-23-01 trails. Within a very large population of 14,095 patients we observed a very low (0.51%) rate of axillary recurrence. The following factors were significantly associated with a higher risk of axillary recurrence regardless of size of nodal involvement with no difference between patients with or without ALND: high SBR grade tumors, HR-negative, or HER2-positive tumors 16 .
Eighty-six patients underwent total mastectomy in the in IBCSG-23-01 trial, 18 in the AATRM trial, and 381 in our study. There is currently no information from previous randomized trials on mastectomy with SN macro-metastasis. Hopefully results from SERC, BOOG 2013-07 (ref. 18 ), and SENOMAC trials 19 should provide sufficient evidence to support omission of in case of SN macro-metastasis after mastectomy. Unfortunately due to insufficient inclusion the BOOG trial was closed.
In all, 27.7% patients (522) included in this study were not eligible for Z11; this population had an increased rate of NSN involvement particularly when cALND was performed prior to chemotherapy. These patients were not included in previous randomized trails and received adjuvant treatments more frequently.
As a result of lacking evidence to support omitting cALND due to underpowered previous trials 20 , there are currently several ongoing non-inferiority randomized trials investigating the possibility of avoiding cALND in case SN macro-metastatic involvement [21][22][23] . However, recent contributions are in line with the tendency to avoid cALND according to the Z0011 trial's criteria 24,25 . The AMAROS trial concluded that there was no significant difference in terms of local control between ALND and axillary dissection for patients with SN involvement; however, considering the sample size the authors could not demonstrate equivalence of both techniques 26 . Due to insufficient evidence to support the omission of ALND in case of SN micro-metastasis after a mastectomy, SERC trial evaluated macro-metastatic as well as micro-metastasis and ITCs with a planned stratification based on metastasis size 27 . Moreover, we have recently confirmed the external validity of SERC trial population in comparison with others populations studies for BC patients with SN micrometastasis 28 .
In conclusion, we reported involved SNs by ITC, micrometastases, and macro-metastases for early BC in 5.91%, 28.12%, and 65.97%, respectively, without significant difference between two arms for all criteria but with a significant difference between the two groups of patients according to treatment realized. AC administration rates, post-operative radiotherapy rates, and ET rates were not different between the two arms. Higher positive NSN rate was observed for patients with AC performed after ALND (17.65% for SN micro-metastases and 35.22% for SN macrometastases). In contrast, when AC was administered before ALND, NSN involvement rates were low: 5.71% for SN micro-metastases and 20.31% for SN macro-metastases. These results explain in part the low rate of axillary recurrence for patients without cALND reported in trials and others studies. When cALND is required, AC could be administered before cALND so as not to delay systemic treatment. Of interest, crude rates of positive NSN were higher for patients non-eligible to Z0011 trial (26.45% versus 17.62%).
Due to the evolution of clinical practice since 10-year results of ACOSOG-Z0011 and IBCSG 23-01 trials, an amendment was performed in July 2018 with restriction of inclusions for noneligible patients to Z11 trial: patients with involved SNs by microor macro-metastases and mastectomy, patients with more than two involved SN or extracapsular extension, and patients with SLNB performed before NAC with SN involvement whatever the size of SN involvement. Moreover, results according to tumors subtypes, for non-eligible patients to Z0011 trial who had the higher positive NSN rate and particularly for mastectomy, could potentially lead to practice changes.

Study design
SERC trial (SERC: "Sentinel Envahi et Randomisation du Curage") is a multicenter prospective randomized non-inferiority phase-3 trial comparing no cALND with cALND in patients with BC and metastasis in the SN, with a stratification planned between SN disease burden (macrometastases versus ITC or micro-metastases).
Ethics approval were obtained from the Institutional Review Board of Paoli Calmettes Institute (SERC-IPC 2012-001) and the National Ethics Committee (2012-A00379-34) and a written informed consent was obtained from all subjects.
The primary objective is to demonstrate that the hazard ratio (SLNB versus cALND) for disease-free survival is significantly less than the noninferiority margin set to 1.25.
Patients randomized were recruited from 53 institutions over an accrual period of 73 months from July 2012 to July 2018. Since August 2018, we have proposed to continue inclusions in this trial, only for patients who do not meet Z0011 trial criteria. Women eligible for registration could be any age ≥18 years, provided they had no previous or concomitant malignancy, pure ductal carcinoma in situ, previous systemic therapy before SLNB, distant metastases, palpable axillary nodes.

Patients
Patients with one or more positive SN, multi-centric tumors, ≤cT2 cN0, ITC, or micro-metastases or macro-metastases with or without extracapsular extension, mastectomy or breast conservative surgery, NAC with SLNB before chemotherapy were allowed to participate. Patients with NAC or hormone therapy before SLNB were excluded.
Patients underwent whole breast radiation (WBR) with a sequential boost to the tumor bed in case of breast conservative surgery or post mastectomy radiotherapy (PMRT). Indications for PMRT were based on current guideline in use in each institution (usually all patients with axillary macro-metastasis and patients without axillary macro-metastases but several prognosis factors, i.e., lympho-vascular invasion, young age, pT size >3 or 5 cm). Radiotherapy started 4-8 weeks after surgery or at the end of AC. Indications followed each institution's guidelines. A total of 50 Gy at ICRU (International Commission on Radiation Units) point were delivered in 25 fractions of 2 Gy over a period of 5 weeks; patients did not receive axillary radiotherapy and the treatment included two tangential fields for both chest wall radiotherapy and WBR. Indications for AC and endocrine therapy were based on current guideline in use in each institution. A permuted block randomization stratified on each participating center and SN disease burden was used.
SN was detected using isotopic detection alone or a combination of colorimetric and isotopic detection after a peri-areolar or tumoral injection.
A preoperative axillary sonography was systematically performed: cN0 patients with suspicious axillary node and axillary positive biopsy could be included. In case of metastatic SN, could be performed, when randomly assigned to the ALND group, either during the initial surgery after an extemporaneous examination of the SN or after definitive examination during a second surgery. For patients with second surgery for cALND, AC was always planned before cALND which was performed before or after AC.

SN analysis
SNs were sectioned using a 50-200 microinterval between each section and all sections were examined by the pathologist using hematoxylin and eosin staining (HES). In case of negative result using HES, cytokeratin immune staining was employed. A one-step nucleic acid amplification technic was used to examine SNs 29,30 . A CK19 mRNA copy number/μl ranging between 250 and 5000 was defined as micro-metastases and above 5000 as macro-metastases.

Statistical analysis
The cut-off date for data collection was July 31, 2018. Graphical display of cumulative numbers of accruals since study start and total accruals per participating centers were presented. Patient and tumor characteristics (age, SBR grade, tumor histology, tumor size, lLVI, ER, HER2 status, tumor subtypes), SN biopsy (SN status, SN involvement), and treatment (type of surgery, systemic chemotherapy, endocrine therapy, trastuzumab) were summarized in all randomized patients. Descriptive statistics are presented, both in the ALND and SLNB randomized groups (ALND and SLNB arms) and in the actual ALND and SLNB treatment groups (ALND and SLNB groups), as mean ± standard deviation, median (interquartile range) for continuous data, and frequency (percent) for categorical data. We detailed the deviations from the randomized treatment in each arm. We used univariate analyses and multivariate logistic regression analyses to identify the factors among the patient and tumor characteristics, SN biopsy data, and treatment options significantly associated with AC administration in ER-positive patients and NSN involvement rates in patients receiving radiation therapy. Prior to analysis, individual data were first categorized using predefined thresholds. Only factors significant or of borderline significance in univariate analyses were included as exploratory variables in regression analyses. Statistical analyses were carried out using SAS release 9.4 (SAS-Institute, Inc., Cary, NC). The level of statistical significance was set to 0.05, with no adjustment for multiplicity.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.