Sentinel Lymph Node Biopsy vs No Axillary Surgery in Patients With Small Breast Cancer and Negative Results on Ultrasonography of Axillary Lymph Nodes

1Sentinel lymph node biopsy (SLNB) is the standard of care for axillary node staging in patients with early breast cancer (BC). The application of this technique represented a milestone in surgical de-escalation, providing the same outcome as axillary lymph node dissection (ALND)., The results of the American College of Surgeons Oncology Group Z0011 (ACOSOG Z0011) randomized clinical trial showed that there is no advantage in performing ALND compared with not performing ALND, even when up to 2 sentinel nodes are positive, for patients receiving breast-conserving surgery, adjuvant radiotherapy, and medical treatment. These data were supported by other randomized clinical trials, and became the basis for a new standard of axillary care., In the past, SLNB was conceived as a reliable means to distinguish between patients with negative nodes who can be spared the morbidity associated with a complete ALND and those with nodal involvement who might benefit from a more extensive surgical procedure. But the absence of advantages from ALND revealed in the ACOSOG Z0111 trial raised 2 questions: first, whether it is really necessary to perform surgical staging of axillary lymph nodes, and second, whether imaging might replace surgery for reliable staging of axillary lymph nodes. The SOUND (Sentinel Node vs Observation After Axillary Ultra-Sound) trial was launched in February 2012 with the aim of evaluating the oncological safety of omitting axillary surgery in patients with BC of a diameter equal to or smaller than 2 cm and a negative result on preoperative axillary lymph node ultrasonography.

2The SOUND trial was a prospective multicenter noninferiority phase 3 randomized clinical trial conducted in 18 hospitals in Italy, Spain, Switzerland, and Chile (eTable in Supplement 2). Participants were recruited and enrolled from February 6, 2012, to June 30, 2017. Data were analyzed between October 10, 2022, and January 13, 2023. The Trial Protocol and Statistical Analysis Plan for this trial are provided in Supplement 1. Of note, the SOUND trial was unintentionally registered late with ClinicalTrials.gov on June 18, 2014, after 492 patients had already been randomized. No interval analysis was conducted, and no data were examined before clinical trial registration. The trial was conducted in accordance with the amended Declaration of Helsinki, and the protocol was approved by the institutional review board at each participating center. All patients provided written informed consent. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline for randomized clinical trials.

3Eligible patients were women of any age with invasive BC up to 2 cm in diameter, lack of involvement of axillary nodes at clinical evaluation, and a plan to undergo a breast-conserving surgery and radiotherapy. All patients were required to have preoperative axillary ultrasonography showing no lymph node involvement at imaging. In the case of a doubtful finding on ultrasonography concerning an isolated lymph node, fine-needle aspiration was performed to rule out the presence of nodal metastases by cytological examination. Axillary lymph nodes with micrometastases or macrometastases were defined as positive. Exclusion criteria were the preoperative presence of multiple doubtful or suspicious lymph nodes, extensive multifocality or multicentricity, bilateral BC, diagnosis of synchronous distant metastases, previous cancer, ongoing pregnancy or lactation, and obstacles to obtaining informed consent or undergoing regular follow-up.

4For the sample size calculation, the 5-year distant disease–free survival (DDFS) in the group randomized to receive no axillary surgery (no axillary surgery group) was assumed to be 96.5%. We calculated a target sample size of 1560 participants (780 per group) to test whether the no axillary surgery group experienced outcomes that were no worse than those of the group randomized to receive SLNB (SLNB group) given a margin of noninferiority for the 5-year DDFS of 2.5%. Between February 6, 2012, and June 30, 2017, 1463 women were recruited and enrolled in the study (93.8% of the planned sample size). The steering committee decided to close the accrual period early because the enrollment became too slow after most of the participating hospitals adopted the ACOSOG Z0011 criteria. Most of the enrolled patients (n = 1406) had negative ultrasonographic results; 57 patients who had a single doubtful node on ultrasonography were randomized after undergoing fine-needle aspiration cytological examination with negative results. Overall, 727 patients were randomized to the SLNB group and 736 to the no axillary surgery group. In the SLNB group, 19 patients discontinued intervention; in the no axillary surgery group, 39 patients discontinued intervention (Figure 1). The remaining 1405 women (708 in the SLNB group and 697 in the no axillary surgery group) were included in the intention-to-treat analysis.

5The study design is summarized in Figure 1 and has been previously described.,, Eligible patients were randomly assigned in a 1:1 ratio to undergo SLNB or no axillary surgery. Of note, in the SLNB group, patients with metastases in the sentinel lymph node had to undergo ALND. Randomization was performed online using computer-generated allocation without stratification.

6The protocol-specified primary end point of the study was DDFS. Secondary end points were the cumulative incidence of distant recurrences, the cumulative incidence of axillary recurrences, DFS, overall survival (OS), and the adjuvant treatment recommendations.

7To describe the study population, we used frequencies, percentages, medians, and IQRs. Differences in the distribution of categorical variables were assessed using the χ2 test.

8We compared women in the SLNB group with those in the no axillary surgery group using an intention-to-treat approach. The primary end point was DDFS, with distant metastases and deaths from all causes as the events of interest. Ipsilateral BC recurrences, axillary recurrences, contralateral BC, and nonbreast primary tumors were treated as censoring events. The secondary end points were DFS and OS. In the DFS analysis, all available events (ie, ipsilateral BC recurrences, axillary recurrences, distant metastases, contralateral BC, nonbreast primary tumors, and deaths from all causes) were events of interest. In the OS analysis, deaths from all causes were events of interest. In case of no events, observations were censored at last disease assessment for DDFS and DFS, while observations were censored at last vital status assessment for OS. Survival curves were estimated using the nonparametric Kaplan-Meier method. Differences in survival were assessed by means of the log-rank test. As additional secondary end points, we calculated the cumulative incidence of distant metastases and the cumulative incidence of axillary recurrences in a competing risk framework; events that were not distant metastases or axillary recurrences were treated as competing events. Differences in the cumulative incidence between groups were evaluated by means of the Gray test. For the analysis of DDFS, DFS, and cumulative incidence, only the data from the first event were used. All tests were 2-sided.

9Based on a margin of noninferiority for the 5-year DDFS of 2.5%, the statistical power was set at 80%, and the 1-sided type 1 error was set at 5%. Noninferiority was shown if the upper limit of the 2-sided 90% CI for the hazard ratio for no axillary surgery vs SLNB, calculated at 5 years (ie, right-censoring follow-up at 5 years), was less than 1.74. The corresponding 1-sided P value for noninferiority was reported. The significance threshold was 1-sided P <.05. Analyses were performed using SAS software, version 9.4 (SAS Institute Inc).

10Among 1405 women included in the intention-to-treat analysis, the median (IQR) age was 60 (52-68) years, the median (IQR) tumor size was 1.1 (0.8-1.5) cm, and 1234 patients (87.8%) had estrogen receptor (ER)–positive ERBB2 (formerly HER2 or HER2/neu), nonoverexpressing BC. Baseline characteristics were similar between treatment groups (Table 1). In the SLNB group (n = 708), 97 patients (13.7%) had positive axillary nodes (36 [5.1%] with micrometastases and 61 [8.6%] with macrometastases), and 4 (0.6%) had 4 or more positive lymph nodes.

11The recommended adjuvant systemic therapy and radiotherapy were similar in the 2 groups (Table 2). In the SLNB group, 652 patients (92.1%) had ER-positive BC; of those, 638 (97.9%) received hormone therapy. In the no axillary surgery group (n = 697), 653 patients (93.7%) had ER-positive BC; of those, 646 (98.9%) received hormone therapy. In the SLNB group, 48 patients (6.8%) had ERBB2-overexpressing BC; of those, 45 (93.8%) received trastuzumab. In the no axillary surgery group, 47 patients (6.7%) had ERBB2-overexpressing BC; of those, 46 (97.9%) received trastuzumab. Overall, 142 women (20.1%) in the SLNB group and 122 women (17.5%) in the no axillary surgery group received chemotherapy, while 694 women (98.0%) in the SLNB group and 680 women (97.6%) in the no axillary surgery group received radiotherapy. In total, 76 patients (10.7%) in the SLNB group and 75 patients (10.8%) in the no axillary surgery group received partial breast radiotherapy (intraoperative electron radiotherapy [ELIOT], 21 Gy). A total of 24 patients (3.4%) in the SLNB group and 39 patients (5.6%) in the no axillary surgery group received an intraoperative boost of ELIOT (12 Gy) followed by a hypofractionated course of whole-breast radiotherapy (37.05 Gy in 13 fractions). In total, 593 patients (83.8%) in the SLNB group and 565 patients (81.1%) in the no axillary surgery group received whole-breast radiotherapy with conventional fractionation delivered over 3 to 5 weeks according to the standard of care applied in the different institutions.

12The median (IQR) follow-up for disease assessment was 5.7 (5.0-6.8) years in the SLNB group and 5.7 (5.0-6.6) years in the no axillary surgery group. The median (IQR) follow-up for vital status assessment was 5.8 (5.0-6.9) years in the SLNB group and 5.8 (5.0-6.8) years in the no axillary surgery group. During the follow-up period, 12 (1.7%) locoregional relapses, 13 (1.8%) distant metastases, and 21 (3.0%) deaths occurred in the SLNB group, while 11 (1.6%) locoregional relapses, 14 (2.0%) distant metastases, and 18 (2.6%) deaths occurred in the no axillary surgery group (Table 3). The 5-year DDFS was 97.7% in the SLNB group and 98.0% in the no axillary surgery group (log-rank P =.67) (Figure 2A). When evaluating the primary hypothesis of noninferiority for 5-year DDFS, we found that the omission of axillary surgery was noninferior to SLNB (hazard ratio, 0.84; 90% CI, 0.45-1.54; noninferiority P =.02). The 5-year DFS was 94.7% in the SLNB group and 93.9% in the no axillary surgery group (log-rank P =.30) (Figure 2B). The 5-year OS was 98.2% in the SLNB group and 98.4% in the no axillary surgery group (log-rank P =.72) (Figure 2C).

13The 5-year cumulative incidence of distant metastases was 2.3% in the SLNB group and 1.9% in the no axillary surgery group (Gray P =.69) (eFigure 1 in Supplement 2). The 5-year cumulative incidence of axillary recurrences was 0.4% in both groups (Gray P =.91) (eFigure 2 in Supplement 2).

14In the SOUND trial, the omission of axillary surgery was noninferior to surgical staging performed by SLNB when evaluating DDFS at 5 years in patients with BC up to 2 cm and a negative result on preoperative ultrasonography of axillary lymph nodes. Of note, in the no axillary surgery group, the cumulative incidence of lymph node recurrences in the axilla was very low (0.4% at 5 years), despite a 13.7% rate of nodal involvement in the SLNB group.

15Other prospective randomized clinical trials,, conducted in the pre-SLNB era have shown that ALND did not improve outcomes compared with no surgery in the axilla, confirming that removal of axillary lymph nodes had no therapeutic effect by itself and was performed mainly as a staging procedure. Thus, a relevant research aim addressed as a secondary end point was to evaluate possible differences in terms of adjuvant treatment recommendations. Data from the current trial indicated that adjuvant treatments were not significantly different in the 2 study groups, regardless of whether the pathological information from SLNB was available. These data confirm the increasing pattern of guiding adjuvant treatment mostly through the use of biological parameters rather than clinicopathological variables.,,

16The data in this trial were consistent with guidelines from the Choosing Wisely campaign of the Society of Surgical Oncology, which recommends omitting SLNB in patients older than 70 years with small ER-positive ERBB2-negative BC when the adjuvant treatment plan is clear and does not include the addition of chemotherapy to endocrine treatment. However, the information provided by nodal status is currently not being completely ignored when selecting postoperative treatment for younger patients, and the absence of the pathological information acquired from SLNB might still create challenges in the management of ER-positive ERBB2-negative BC. Even with the wide availability of genomic testing, chemotherapy can be prescribed or at least considered in addition to endocrine treatments for women with endocrine-responsive disease and axillary lymph node involvement, especially for premenopausal patients. In the Rx-PONDER (A Clinical Trial Rx for Positive Node, Endocrine Responsive Breast Cancer) trial, the advantage of adding cytotoxic agents could not be ruled out in patients younger than 50 years with nodal metastases, even when the recurrence score was low or intermediate. In addition, the duration of adjuvant endocrine therapy can be adapted according to the risk estimate and usually prolonged after 5 years in patients with node-positive disease.,, On the other hand, the absence of pathological nodal involvement might allow de-escalation of the hormonal treatment, both in terms of drug choice (tamoxifen vs aromatase inhibitors or ovarian suppression vs no ovarian suppression) and duration, especially in the case of adverse effects that substantially affect patient quality of life. Furthermore, in the subset of patients with ERBB2-positive disease undergoing upfront surgery, information on nodal status is relevant to properly tailor adjuvant treatment, which in node-negative disease might be restricted to paclitaxel and trastuzumab. In addition, in patients with small triple-negative BC undergoing upfront surgery, pathological staging of nodal status might be relevant to modulate the postoperative treatment plan. Moreover, nodal radiation fields are frequently adapted for women with nodal involvement as a complement to breast radiotherapy after breast conservation. In contrast, some patients 65 years and older with node-negative disease might even be spared from undergoing any radiotherapy with a limited number of locoregional events and no detrimental effect on OS.,

17We were also interested in evaluating the capacity of ultrasonography to detect nodal involvement to understand whether imaging might eventually replace surgery for reliable staging. It is well known that ultrasonography of axillary nodes has several limitations in detecting lymph node involvement, with a sensitivity ranging from 24% to 94%. However, in the current study, the use of ultrasonography was able to rule out the presence of relevant nodal burden, which might not have been identified with clinical evaluation alone. In the SLNB group, the presence of micro- and macrometastases was limited (13.7%) and much lower than the rate reported in previous trials,, likely due to the screening effect of the negative preoperative axillary ultrasonography result required to enter the trial. Given the limited number of patients with macrometastases, the very low number of patients with extensive nodal involvement (0.6% with 4 or more positive nodes) in the axillary surgery group, and the extremely low cumulative incidence of axillary lymph node recurrence in the no axillary surgery group (0.4% at 5 years), the performance of ultrasonography can be considered clinically meaningful. Despite the need for further research to improve imaging methods, the multi-institutional nature of our study supported the wide reproducibility of ultrasonography as a simple and inexpensive method that can be routinely applied in the preoperative workup of all patients with BC.

18The results of this trial support the safety of omitting axillary surgery in older postmenopausal women with ER-positive ERBB2-negative BC who met the SOUND eligibility criteria. This subset of women represents approximately 25% of the whole population of women with BC., Considering that an estimated 2.3 million women are diagnosed with BC every year, approximately 500 000 patients might be able to take advantage of the total omission of axillary surgery, which has been shown to improve arm function in the early postoperative period.,, Data from the SOUND trial should be considered in the multidisciplinary decision-making process of the individual patient to identify those who might be able to omit SLNB without affecting the postoperative treatment plan. Moreover, the incorporation of these data in future guidelines might lead to a substantial decrease in health care costs due to the reduced involvement of human resources and savings in terms of materials and time.

19This study has several limitations. The inclusion criteria of the study led to enrollment of patients who could be considered to be at low risk of recurrence in the short term. Thus, we cannot exclude the possibility that differences in outcome might appear over a longer follow-up period because the curve of event onset is expected to occur later in patients with ER-positive ERBB2-negative BC than in patients with triple-negative or ERBB2-positive disease., Therefore, we have planned to continue the follow-up with a formal analysis after 10 years. We also highlight that the analysis of adjuvant treatments was not the primary end point and that sample size calculations were not performed for this purpose. Therefore, this trial might be underpowered to detect small differences in the details of medical or radiotherapy treatment recommendations.

20The SOUND trial was designed in 2011, immediately after publication of the ACOSOG Z0011 trial and as its natural continuation. At that time, ALND was the standard of care in the presence of sentinel lymph node metastases because data from the ACOSOG Z0011 trial had not yet been incorporated into guidelines. Because the objective of the SOUND trial was to evaluate the oncological safety of omitting axillary surgery, we decided to compare the experimental group with the most standardized and radical approach comprising ALND in the presence of nodal macrometastases. It was only between 2016 and 2017 that most of the participating centers embraced the ACOSOG Z0011 approach, slowing down the recruitment pace of our enrollment. Furthermore, all available options of radiotherapy were allowed, even partial breast radiotherapy. Of note, 114 patients (16.3%) randomized to the no axillary surgery group received ELIOT as a full dose or an intraoperative boost.

21We also highlight the unintentional late registration of the SOUND trial in ClinicalTrials.gov even though we specify that no data were examined and no interval analysis was conducted before the trial was registered. Late registration occurred without any intention to bias the reporting, which is confirmed by the fact that study design, inclusion criteria, exclusion criteria, end points, and sample size were published in a peer-reviewed journal immediately after the study started. Another article was later published reporting the number of patients included in the SOUND trial up to that time, which provided a forecast on the conclusion of patient enrollment.

22This randomized clinical trial found that omission of axillary surgery was noninferior to SLNB in women with small BC and negative results on ultrasonography of the axillary lymph nodes. These results suggest that patients with these features can be safely spared any axillary surgery when the lack of pathological information does not affect the postoperative treatment plan.