The de-escalation of axillary surgery in breast cancer has been supported by trials such as SOUND and INSEMA, which demonstrated the safety of omitting sentinel lymph node biopsy in patients with clinically and sonographically negative axillae. In this context, the performance of axillary ultrasound as a selection tool was evaluated.
Material and methodsA total of 240 patients with invasive breast cancer treated with primary surgery and preoperative axillary ultrasound were included. Lymph nodes showing suspicious features were submitted to fine-needle aspiration (FNA). Ultrasound was defined as negative in the absence of suspicious findings or in the presence of a negative FNA. Axillary histology served as the reference standard. A subgroup of patients meeting criteria similar to SOUND/INSEMA (≥50 years, tumors ≤20 mm, negative ultrasound) was also analyzed.
ResultsConsidering all cases, the diagnostic metrics were sensitivity 57.8%, specificity 99.5%, PPV 96.3%, and NPV 91.1%.
In the selected subgroup (n = 151), NPV 94%, false-omission rate: 6%.
Most false-negative cases presented low metastatic burden (1–2 nodes).
ConclusionsAxillary ultrasound demonstrated adequate diagnostic performance for selecting patients eligible for omission of axillary surgery, with a false-omission rate within the acceptable range for sentinel lymph node biopsy. In carefully selected populations, it may represent a safe alternative to surgical staging.
The surgical management of the axilla in breast cancer has undergone a profound transformation over the past two decades, marked by a sustained trend toward therapeutic de-escalation. This shift is supported by robust evidence demonstrating that, in carefully selected patients, reduced axillary intervention does not compromise oncologic outcomes.
Trials such as ACOSOG Z0011 [1,2] established that in patients with tumors up to 5 cm and up to two metastatic sentinel lymph nodes, omitting axillary lymph node dissection does not affect overall or disease-free survival. More recently, the randomized SOUND [3] and INSEMA [4] trials have taken an additional step by evaluating the possibility of avoiding even sentinel lymph node biopsy (SLNB) in patients with clinically and sonographically negative axillae, demonstrating non-inferiority in axillary recurrence and oncologic outcomes at 5 years.
In this context, axillary ultrasound assumes a central role as a preoperative staging method. It is essential to emphasize that in this clinical framework, “negative axillary ultrasound” refers to both the morphological evaluation of axillary nodes and the diagnostic sampling (FNA or core biopsy) when suspicious findings are present. The increasing adoption of ultrasound as a decisive tool demands diagnostic performance not inferior to that of SLNB, which remains the standard of care for initial axillary staging.
Given the operator-dependent nature of axillary ultrasound, radiologists must understand the relevance of their role within current surgical decision-making and quantify its performance in clinical practice.
The objective of this study was to evaluate, in our experience, the diagnostic performance of axillary ultrasound in terms of sensitivity, specificity, and predictive values. A critical objective is to determine whether a negative ultrasound result can be considered non-inferior to a negative SLNB result regarding the negative predictive value and false-omission rate (false negatives among ultrasound-negative cases), a key metric for supporting omission of axillary surgery.
Material and methodsThis work consists of a review of the work team's experience regarding the ultrasound evaluation of the axilla in patients with breast cancer and includes a database of 370 consecutive patients treated between January 2019 and October 2025. A total of 130 cases were excluded based on predefined criteria (diagnosis of carcinoma in situ without invasive component, neoadjuvant treatment, absence of axillary surgery, or incomplete data). The final cohort included 240 patients with invasive breast carcinoma who underwent primary surgery and had documented preoperative axillary ultrasound.
Ultrasound evaluations were performed using a Toshiba Aplio 300 unit with an 8-MHz linear transducer. The ipsilateral axilla was systematically examined using vascular and muscular anatomy as reference. Lymph nodes were considered suspicious when cortical thickening ≥3 mm (focal or diffuse) was identified. This includes narrowing or obliteration of the fatty hilum of the lymph node. Lymph node enlargement was not considered suspicious unless accompanied by cortical thickening. All suspicious lymph nodes underwent FNA.
Ultrasound was considered negative in the absence of suspicious nodes or when FNA was negative.
In all cases where the axillary ultrasound was negative, a sentinel lymph node biopsy was performed. If this procedure was negative, no further axillary surgery was performed. In cases where the sentinel lymph node biopsy was positive, axillary lymphadenectomy was performed.
In cases where the axillary ultrasound was positive, axillary lymphadenectomy was performed.
Axillary histology (SLNB or axillary dissection) served as the reference standard. Only cases with macrometastases were classified as positive; micrometastases and isolated tumor cells were considered negative due to their lack of therapeutic impact in this cohort.
A second analysis included a subgroup of patients meeting criteria similar to SOUND and INSEMA (age ≥ 50 years, tumors ≤2 cm, negative axillary ultrasound). In this subgroup, the false-omission rate was calculated, and the number of metastatic nodes in false-negative cases was recorded.
ResultsThe cohort included 240 patients with invasive tumors; histological and immunophenotypic characteristics are shown in Table 1.
Histological and immunophenotypic characteristics.
| Histología | n° | % |
|---|---|---|
| IDC | 85 | 35 |
| IDC/DCIS | 119 | 50 |
| IDC/ILC | 9 | 4 |
| ILC | 27 | 11 |
| Immunohistochemistry | n° | % |
| Luminal | 209 | 87 |
| HER 2-positive | 21 | 9 |
| Triple negative | 10 | 4 |
IDC = invasive ductal carcinoma; ILC = invasive lobular carcinoma; DCIS = ductal carcinoma in situ.
Axillary ultrasound was negative in 213 cases (89%) and positive in 27 (11%). Axillary histology was negative in 195 patients (81%) and positive in 45 (19%).
Performance metrics for axillary ultrasound are shown in Table 2.
Diagnostic performance of axillary ultrasound.
| Metric | % | 95% CI |
|---|---|---|
| Sensitivity | 57.8 | 43–71 |
| Specificity | 99.5 | 97–100 |
| Positive predictive value | 96.3 | 82–99 |
| Negative predictive value | 91.1 | 87–94 |
| Positive likelihood ratio | 115.6 | 16–814 |
| Negative likehood ratio | 0.42 | 0.3–0.6 |
| False-omission rate | 8.9 | 5.8–13.4 |
From the 240 patients, those meeting the criteria of age ≥ 50 years, tumor size ≤20 mm, and negative axillary ultrasound (similar to SOUND/INSEMA) were selected, yielding 151 cases. Among them, 9 had histologically positive axillae (false negatives). Performance metrics for this subgroup are shown in Table 3.
In false-negative cases within the subgroup, the distribution of metastatic nodes is shown in Table 4. The table shows that in 78% of cases, there were up to two metastatic lymph nodes, and only in 2 cases were more than two lymph nodes affected.
DiscussionOur results regarding the performance of axillary ultrasound in identifying patients with truly N0 axillae are consistent with those reported in several series in the literature, although with some variations depending on the cohort characteristics and the diagnostic methods used. Retrospective and prospective studies have documented a wide range of sensitivity for axillary ultrasound, frequently between approximately 50% and 80% when using ultrasound alone or combined with guided fine-needle aspiration techniques, with consistently high specificities in most series (>90%) [5].
The SOUND and INSEMA clinical trials provide the highest-level methodologically rigorous, prospectively controlled evidence for omitting axillary surgery in patients with early-stage breast cancer and negative axillary ultrasound. Their results have been pivotal in the debate on de-escalation of axillary treatment. These trials represent a paradigm shift in modern axillary surgery, demonstrating that omission of axillary surgery is safe in patients selected based on a negative ultrasound.
In the multicenter randomized SOUND trial [3] (1405 patients), omission of all axillary surgery was compared with SLNB in patients with tumors ≤2 cm and negative axillary ultrasound. The primary endpoint, 5-year distant disease-free survival, was 97.7% in the SLNB group versus 98.0% in the no-surgery group, confirming non-inferiority. The 5-year cumulative incidence of axillary recurrence was 0.4% in both arms, and no clinically relevant differences were observed in adjuvant treatment recommendations. In the SLNB arm, nodal involvement was found in 13.7% of cases (8.6% for macrometastasis).
In the prospective randomized INSEMA trial [4], axillary surgery omission was evaluated in patients with T1–T2 invasive breast cancer and clinically and sonographically negative axilla undergoing breast-conserving surgery. Among 4858 patients, 5-year invasive disease-free survival was equivalent between those who did and did not undergo SLNB (91.7% vs. 91.9%).
These prospective studies not only support the usefulness of axillary ultrasound as a tool for selecting N0 patients but also justify a shift in the therapeutic paradigm: from using sentinel lymph node biopsy as a mandatory diagnostic tool, toward a strategy in which a negative axillary ultrasound, under standardized criteria, may be sufficient to omit invasive surgical procedures without compromising oncological safety in low-risk groups.
Compared to our series, where we observed moderate sensitivity and very high specificity, the SOUND/INSEMA data emphasize that, beyond isolated diagnostic metrics, the clinical utility of ultrasound as a screening strategy for N0 selection lies in its ability to identify a subgroup of patients at very low risk of axillary recurrence with oncological outcomes comparable to SLNB when axillary surgery is omitted. This positions axillary ultrasound as a key component within less invasive diagnostic algorithms, where its high specificity supports the decision to avoid surgical procedures without sacrificing patient safety.
In our study, considering all cases, ultrasound showed moderate sensitivity (58%) and very high specificity (99.5%). The positive predictive value indicates that a patient with a positive ultrasound had a 96.3% probability of true metastasis. The negative predictive value was also high (91.1%), leaving 8.9% false negatives—within the 5–10% false-negative range reported for SLNB in clinically node-negative patients [6–11].
In a second stage of our study, patients aged ≥50 years with tumors ≤2 cm and negative axillary ultrasound (criteria comparable to SOUND) were analyzed. The aim was to determine the likelihood that a negative ultrasound might be incorrect (presence of metastasis).
This is clinically relevant because if ultrasound is to replace SLNB, its error rate must not exceed the acceptable range for SLNB (≤10%).
Among studies evaluating axillary ultrasound with biopsy of suspicious nodes, classical false-negative rates (false negatives / all true positives) range from 15% to 25%.
However, in the current clinical application—where only ultrasound-negative cases are considered—the key metric is the false-omission rate (false negatives among negative ultrasound cases). This directly quantifies the risk of missing metastatic nodes when ultrasound is negative and is crucial when determining whether an invasive procedure can be safely omitted.
Some analyses using INSEMA-like criteria report false-omission rates of 10.5% [12]. In a prospective study of 227 patients, Rukanskienė et al. [13] reported 8.5%. In a previous study from our group [5], the rate was 9%. In the current study, the false-omission rate was 6%, with an NPV of 94%, which falls below the commonly accepted false-negative rate for SLNB (5–10%).
Importantly, most false-negative cases had low metastatic burden (1–2 nodes), consistent with ACOSOG Z0011 criteria that would not require axillary dissection, and consistent with findings from SOUND and INSEMA, where limited metastatic burden in missed cases did not worsen oncologic outcomes.
These results support that, in strictly selected populations, axillary ultrasound can provide diagnostic safety comparable to the current standard.
Some limitations of this study should be acknowledged.
First, its retrospective, single-center design may limit the generalizability of the results, as institutional experience, operator expertise, and local protocols could influence diagnostic performance. This aspect is particularly relevant given the operator-dependent nature of axillary ultrasound, which may lead to variability in sensitivity across different clinical settings despite the use of standardized morphological criteria in our practice.
Second, the relatively low number of axillary-positive cases resulted in wide confidence intervals for the positive likelihood ratio, reflecting statistical imprecision rather than reduced clinical relevance.
Finally, the study lacks long-term oncologic follow-up, including axillary recurrence and survival outcomes.
ConclusionOur findings reinforce that in patients with early breast cancer, clinically negative axilla, and negative axillary ultrasound, omission of axillary surgery is a feasible and safe strategy when strict selection criteria are applied. Axillary ultrasound demonstrated diagnostic performance aligned with international standards and a false-omission rate within acceptable limits to replace sentinel lymph node biopsy.
Implementation of this strategy requires: rigorous ultrasound technique, standardized criteria for suspicion, adequate operator training, and structured long-term clinical follow-up.
Axillary surgery de-escalation appears to be advancing toward a new standard, in which high-quality axillary ultrasound may play a central role.
FundingNo funding was received for this study.
Ethical considerations and informed consentIt is confirmed that all participants, where applicable, gave informed consent and that the study has been approved by the institutional ethics committee.
The authors declare no conflicts of interest.
