Factors Associated With Residual Positive Lymph Nodes With Targeted Axillary Lymph Node Dissection for Breast Cancer and Accuracy of Clipped Node Retrieval in Non‐Mapping Patients

ABSTRACT

Background

The current study aimed to determine whether results could justify whether axillary lymph node dissection (ALND) could be further de‐escalated in selected patients having neoadjuvant chemotherapy (NAC).

Methods

Retrospective analyses were performed for breast cancer patients treated with NAC from 2015 to 2023. Abnormal axillary nodes evaluated by pretreatment ultrasound had core biopsies with microclip placement. Patients assigned to ALND had histologically positive SLND (ypN+), were clinically node‐positive (ycN+) post‐NAC, did not map, or did not have a clipped node retrieved.

Results

There were 274 predominantly Hispanic (78%) patients, the median age was 51 years, and the follow‐up was 38.1 months. Of 198 pN+ patients, 120 were clinically node‐negative (ycN0) post‐NAC and had wire‐directed (WD) SLND, whereas 78 were ycN+ and had planned ALND. Sixty percent of pN+ patients converted to ypN0 post‐NAC. Fifty‐eight percent of patients with ypN+ status had additional positive non‐sentinel lymph nodes in the ALND specimen.

Conclusions

ALND was avoided in approximately half of patients who presented with axillary node metastases and became ycN0 post‐NAC. Response to NAC correlated with node status at surgery. Patients who do not map may not need ALND if a pN+ clipped node is retrieved and converts to ypN0.

1. Introduction

Targeted axillary lymph node dissection (TAD) is a surgical technique performed after neoadjuvant chemotherapy (NAC), designed to decrease the rate of nontherapeutic axillary lymph node dissection (ALND) for patients presenting initially with node‐positive breast cancer. This technique was developed based on advances in systemic therapy, which led to higher rates of pathologic complete response after NAC and allowed for the de‐escalation of axillary surgery in patients who previously required ALND [1, 2]. Four prospective clinical trials established the efficacy of excising axillary sentinel lymph nodes after NAC, and a landmark study demonstrated the importance of also removing the lymph node containing metastatic disease biopsied during pretreatment axillary ultrasound [3, 4, 5, 6, 7]. National Comprehensive Cancer Network (NCCN) guidelines currently recommend the use of dual tracers (isotope and blue dye) and excision of the clipped lymph node with two or more sentinel lymph nodes [8]. A more recent prospective multicenter study reported that excision of the clipped lymph node plus one or more sentinel nodes retained a low false‐negative rate, even among patients traditionally considered ineligible for the technique [9]. Results from the prospective SenTa study further demonstrated that TAD conferred similar survival and recurrence outcomes with or without ALND [10].

Based on the compelling evidence demonstrating the safety of TAD, its regular use was integrated into clinical practice at our institution. Results from this Center on 102 patients with pretreatment biopsy‐proven axillary lymph node metastases treated with NAC from 2015 to 2020 demonstrated that pathologic complete response (pCR), node‐negative status after NAC, and triple negative biomarker subtype were associated with relapse‐free survival (RFS) [11]. In that study, 56% of patients were node‐negative on frozen section and had TAD alone (also known as wire‐directed sentinel lymphadenectomy [WD‐SLND]). Axillary recurrence was very uncommon after WD‐SLND alone, making it unlikely that those patients would have derived clinical benefit from the addition of completion ALND to WD‐SLND. These data aligned with prospective, multicenter studies, which demonstrated the efficacy of TAD and the safety of omitting ALND for patients who became histologically node‐negative post‐NAC [12, 13]. A published analysis of survival and recurrence outcomes from the same cohort of patients described in the current report demonstrated that with contemporary systemic chemo‐endocrine therapy agents and modern radiation therapy techniques, local (4%) and regional (2%) recurrence rates were low after NAC regardless of the choice of surgical management of the breast [14]. Favorable responses to NAC were significant factors associated with low local and regional recurrence and high RFS rates.

However, the question remained as to whether it was possible to further de‐escalate the use of ALND in selected patients having NAC. Thus, this study was done to specifically evaluate the accuracy of clipped axillary lymph node retrieval in non‐mapping patients, the number of histologically positive non‐sentinel axillary lymph nodes in patients who have positive sentinel lymph nodes after NAC, and the role of clinical examination of the axilla post‐NAC. The entire cohort of patients having axillary surgery after NAC was included in this analysis to evaluate the failure rates of sentinel lymph node mapping in patients who were initially clinically node‐negative versus those with biopsy‐proven axillary lymph node metastases before NAC.

2. Methods

A retrospective analysis was performed using a single institution database that included clinical, radiographic, and pathologic parameters for all newly diagnosed breast cancer patients treated with NAC between 2015 and 2023. Patients with metastatic breast cancer or inflammatory cancer were excluded from this analysis.

Axillary lymph nodes were routinely evaluated by pretreatment ultrasound, and abnormal nodes were submitted for core needle biopsies and placement of a microclip. All patients in this study then received NAC as recommended in a weekly multidisciplinary breast conference. Patients were generally advised to have chemotherapy if they had ≥ T2 cancers, biopsy‐proven axillary lymph node metastases, or preferred breast conservation but had cancers for which mastectomy was more appropriate. The patient‐specific treatment plans adhered to regimens for NAC described in NCCN guidelines, based on the cancer stage and tumor biomarkers [15].

Wire‐directed sentinel lymph node dissection (WD‐SLND) was done for patients with pretreatment biopsy‐proven axillary lymph node metastases who received NAC and were clinically node‐negative post‐NAC (ycN0). Wire localization of the metastatic lymph node containing the microclip was done on the day before operation using a flexible radial wire. A rotated cranio‐caudal mammogram view including the axilla was then done to demonstrate the relationship between the wire and the clipped lymph node. Periareolar injection of technetium 99m sulfur colloid and lymphoscintigram were also done on the afternoon before the operation. Methylene blue dye was injected in the dermis overlying the tumor if the patient did not map by either lymphoscintigraphy or use of the gamma probe to evaluate the axilla in the operating room after the induction of anesthesia, or if the patient was enrolled in a clinical trial that required the use of dual mapping agents.

The WD‐SLND limited the resected lymphadenectomy specimen, as much as possible, to the radioactive (“hot”) and wire‐localized (“clipped”) lymph node(s). Lymph nodes removed during WD‐SLND were imaged in the operating room using a Faxitron OR specimen Radiography System (Hologic Inc, Mississauga, ON, Canada) to ensure that the node containing the microclip was removed. Lymph nodes were evaluated on frozen section by hematoxylin and eosin. Patients who were ypN0 on frozen section had WD‐SLND alone, whereas those with histologically positive nodes on frozen section (ypN+) had WD‐SLND followed by ALND. ALND was also done for patients who did not map, those whose microclip was not retrieved in the WD‐SLND, or those who had residual palpable adenopathy identified at the time of axillary surgery.

Statistical analyses were conducted using SPSS v27.0 (IBM Corp, Armonk, NY, USA) and Excel v.16 (Microsoft Corporation, Redmond, WA, USA). Clinical, radiographic, and pathologic factors known to be associated with outcomes were included in uni‐ and multivariable models. Factors with a p‐value < 0.05 in univariable analysis were submitted to a multivariable analysis, and p‐values < 0.05 in the multivariable analysis were considered statistically significant. The Health Sciences Institutional Review Board at the University of Southern California approved the study (HS‐24‐00200).

3. Results

Descriptive statistics are shown in Table 1 for 274 predominantly Hispanic (78%) patients with breast cancer receiving NAC. The median patient age was 51 years. There were 94 (34%) estrogen receptor‐positive human epidermal growth factor receptor Type 2‐negative (ER+HER2−), 72 (26%) ER−HER2−, and 108 (39%) HER2+ patients. There were 120 (43%) multifocal or multicentric cancers, and 188 (68%) patients had pretreatment clinical Stage I–II disease. The breast surgical procedure was breast‐conserving surgery in 85 (31%) and mastectomy in 189 (69%) patients.

Table 1.

Descriptive statistics for 274 patients having neoadjuvant chemotherapy for breast cancer, 2015–2023.

Patient factors	N (%)
Age at diagnosis
≤ 50 years	133 (49)
> 50 years	141 (51)
Ethnicity
Hispanic	213 (78)
Asian	25 (9)
African American	10 (4)
White	3 (1)
Other	23 (8)
Focality
Unifocal	154 (56)
Multifocal	78 (28)
Multicentric	42 (15)
Biomarkers
ER−HER2−	72 (26)
ER+HER2−	94 (34)
ER−HER2+	47 (17)
ER+HER2+	61 (22)
Pretreatment T‐category
T1	30 (11)
T2	156 (57)
T3	62 (23)
T4	26 (9)
Pretreatment clinical stage
I	9 (3)
II	179 (65)
III	84 (31)
IV	2 (1)
Surgical procedure
Breast‐conserving surgery	85 (31)
Mastectomy	189 (69)

Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor Type 2.

As demonstrated near the top of Figure 1, before NAC, there were 76 (28%) patients who had a negative axillary ultrasound or core biopsy (cN0) and 198 (72%) who had core biopsy‐proven histologically positive (pN+) axillary lymph nodes. Of the 198 pN+ patients, 120 (61%) were clinically node‐negative post‐NAC (ycN0) and had WD‐SLND, whereas 78 were clinically node‐positive post‐NAC (ycN+) and had planned ALND. WD‐SLND was successfully performed in 94 of the 120 clinically node‐negative patients, of which 38 (40%) were histologically node‐positive (ypN+) and 56 (60%) were histologically node‐negative (ypN0). Thus, ALND was avoided in 56 (47%) of 120 pN+ patients. Of the 76 patients who were clinically and radiographically node‐negative pretreatment (cN0), successful SLND was done in 67, and only 7 (10%) were ypN+. As shown near the bottom of Figure 1, 116 (42%) patients with negative sentinel lymph nodes on frozen section had SLND or WD‐SLND only. There were 5 (7%) of 76 patients who were cN0 or pN0 before NAC who had SLND, did not map, and had ALND, 3 of whom were ypN0. There were 80 patients who had either positive sentinel lymph nodes on frozen section (N = 45) or clinical findings in the operating room (N = 35) that led to ALND.

Figure 1.

Process map for axillary management of 274 patients receiving neoadjuvant chemotherapy 2015–2023. Adenop in OR, adenopathy identified during SLND; ALND, axillary lymph node dissection; cN0, clinically node‐negative and ultrasound node‐negative pre‐NAC; NAC, neoadjuvant chemotherapy; pN0, pre‐NAC core biopsy of lymph node was histologically node‐ negative; pN+, pre‐NAC core biopsy‐proven axillary lymph node metastases; SLN+, sentinel lymph node‐positive on frozen section; SLN−, sentinel lymph node‐negative on frozen section; SLND, sentinel lymph node dissection; WD‐SLND, wire‐directed sentinel lymph node dissection; ycN0, clinically node‐negative post‐NAC; ycN+, clinically node‐positive post‐NAC.

3.1. Accuracy of Clipped Node Retrieval in Non‐Mapping Patients

Referring to the middle left of Figure 1, of the 120 patients having WD‐SLND, 11 (9%) did not map, and in 12 (10%), the clip was not retrieved. Eight of these 11 non‐mapping patients had ALND and the clipped node evaluated, six of whom had negative frozen section, and five were ypN0. Of the three whose clipped nodes were not evaluated, for one patient, the clip could not be found, and for the other two, they were very early in the series before clipped nodes in non‐mapping patients were routinely evaluated by frozen section at our institution.

3.2. Residual Positive Nodes in ALND After NAC

Of the 38 patients with histologically positive sentinel lymph nodes at the time of WD‐SLND, 22 (58%) had additional positive non‐sentinel lymph nodes in the ALND specimen. Final lymph node status (ypN) was ypN1 for 24 (63%) patients, ypN2 for 12 (32%), and ypN3 for 2 (5%). As demonstrated in Table 2, biomarker subtype, pathologic T‐category post‐NAC, and radiographic response to NAC were significantly associated with lymph node status for the 94 patients having WD‐SLND, whereas patient age and focality of tumor were not. The biomarker data were consistent with the observation that only 10% of ER+HER2− patients achieved a pCR after NAC [14]. Thirty (91%) of 33 ypT0 patients, 32 (84%) of 38 ER− patients, and 24 (83%) of 29 patients with radiographic complete response to NAC (rCR) converted to histologically negative sentinel nodes (ypN0) at the time of WD‐SLND.

Table 2.

Factors associated with lymph node status for 94 patients with successful WD‐SLND after neoadjuvant chemotherapy, 2015–2023.

Patient factors	N	N (%) node‐negative	p‐value
Total	94	56 (60)
Biomarkers
ER+	56	24 (43)	Ref.
ER−	38	32 (84)	< 0.0001 *
HER2+	33	23 (70)	Ref.
HER2−	61	33 (54)	0.168
ER−HER2−	27	23 (85)	Ref.
ER+HER2−	34	10 (29)	0.002 *
ER−HER2+	11	9 (82)	0.900
ER+HER2+	22	14 (64)	0.051
Post‐Treatment Pathologic T‐category
ypT0	33	30 (91)	< 0.0001 *
ypTis (noninvasive)	8	7 (88)	0.004 *
ypT1–4 (invasive)	53	19 (36)	Ref.
Radiographic response to NAC
rCR	29	24 (83)	Ref.
No rCR	65	32 (49)	0.002 *

Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor Type 2; NAC, neoadjuvant chemotherapy; rCR, radiographic complete response; Ref., reference.

^* p‐values < 0.05 are in bold.

3.3. Clinical Evaluation of the Axilla After NAC

As demonstrated in the middle right of Figure 1, there were 78 patients with persistent clinically positive axillary lymph nodes after NAC (ycN+) who had planned ALND. However, from 2015 through 2018, 21 (38%) of 56 patients having planned ALND were ypN0 at surgery, at which time the authors switched from reliance on physical exam of the axilla post‐NAC to a more aggressive use of targeted axillary dissection using wire localization of previously clipped pN+ lymph nodes (WD‐SLND). From 2019 through 2023, 10 (45%) of 22 patients having a planned ALND were ypN0, so in the latter time period the percentage of patients with ypN0 status after planned ALND remained relatively high, but the number of patients having a planned ALND was markedly reduced. It should also be noted that the increase in the number of ALND in 2021 was due to an increased number of patients having persistent ypN+ nodes on WD‐SLND post‐NAC during that year, which coincided with the treatment of those patients during the COVID pandemic. However, in 2021, compared to other years, there was not a significant increase in identifiable poor‐risk clinicopathologic features, which might have accounted for a higher number of ypN+ WD‐SLND, such as the percentage of ER+HER2− (vs. ER−HER2−) or pretreatment clinical Stage III (vs. Stage II) cancers. However, the change in surgical strategy to more aggressive use of targeted axillary dissection resulted in a reduction in the percentage of ALND, from 84% (76/90 patients) in 2015–2018 to 70% (74/106 patients) in 2019–2023 (p = 0.14). The temporary increase in ALND in 2021 contributed to the lack of statistical significance in the reduction in the percentage of ALNDs from 2019 to 2023 versus 2015 to 2018.

4. Discussion

This retrospective study of 274 patients receiving NAC between 2015 and 2023 demonstrated that ALND may be avoided in approximately half of patients who present before NAC with axillary node metastases and become clinically node‐negative post‐NAC. Over one‐half of patients with ypN+ status at WD‐SLND had additional positive non‐sentinel lymph nodes in the ALND specimen, and the response to NAC correlated with lymph node status at surgery. Biomarker subtype, pathologic T‐category post‐NAC, and radiographic response to NAC were significantly associated with lymph node status at the time of WD‐SLND.

4.1. Failure of Sentinel Lymph Node Mapping After NAC

In this study, technetium 99m sulfur colloid was routinely used as the mapping agent for SLND. The non‐mapping rate after NAC was 7% in patients in patients who were initially node‐negative and 9% in those who were proven by pretreatment core biopsy to have axillary lymph node metastases. These rates in the neoadjuvant setting can be compared with a prior study by the same authors using the same technique in the upfront surgery setting, in which the failure rate of sentinel lymph node mapping was 5.7% [16]. A similar phenomenon was observed during the validation phase of SLND, where the non‐mapping rate in the neoadjuvant setting in ACOSOG Z1071 was 7.1% compared with the upfront surgery setting in NSABP B‐32, where the non‐mapping rate was 3% [17]. These technical difficulties with sentinel lymph node mapping in the neoadjuvant setting led to the development of a technique by Caudle et al. to incorporate the removal of the clipped lymph node into the SLND, which significantly improved the accuracy and became known as targeted axillary dissection [7]. Removal of the clipped node was incorporated early into the SLND technique described in the current report, named herein as WD‐SLND. The fact that five of the eight non‐mapping patients in this study who had a clipped node evaluated converted to ypN0 status suggests that patients who do not map may not need ALND if a core biopsy‐proven metastatic axillary clipped node is retrieved and is histologically node‐negative at targeted ALND, a finding which should be further evaluated with a larger sample size.

4.2. Residual Positive Nodes in ALND

The finding in the current study that over one‐half of patients with ypN+ status at WD‐SLND had additional positive non‐sentinel lymph nodes in the ALND specimen is consistent with recently presented findings by Boughey et al. from the Alliance 011202 Clinical Trial, in which the rate of additional positive axillary nodes with ALND was 46% in patients with histologically positive sentinel lymph nodes post‐NAC [18]. The likelihood of additional histologically positive (non‐sentinel) axillary lymph nodes with ALND was influenced by the number of positive sentinel lymph nodes, ypT3 category, and the number of lymph nodes removed at ALND. ALND led to upstaging of N‐category in 25% of patients, most prominent in patients with the ER+HER2− biomarker subtype. Johnson et al. evaluated non‐sentinel node‐positivity at ALND after NAC in 122 patients and demonstrated that non‐sentinel lymph nodes were histologically positive in 43%. pretreatment clinical adenopathy (cN1) and ER+HER2− biomarker subtype were factors associated with the presence of additional positive lymph nodes after a node‐positive SLND [19]. Sanders et al. found that factors associated with non‐SLN positivity in cN+ patients included HER2 status, multicentricity/multifocality, number of positive SLNs, and size of SLN metastasis, with number of positive SLNs and size of SLN metastasis having the greatest influence [20]. Moore et al. aimed to identify factors associated with additional positive lymph nodes at ALND and the impact of residual cancer burden (RCB) in patients who were pretreatment biopsy‐proven node‐positive [21]. The study demonstrated that 25% of patients had additional nodal disease at ALND and that greater RCB post‐NAC was associated with non‐sentinel node‐positivity. As RCB is essentially a surrogate marker for the response to NAC, these findings were comparable to those reported in the current study. Finally, another study by Moore et al. found that when the clipped node on TAD was a non‐sentinel lymph node but histologically positive, half of the time it was the only positive node and only site of residual disease, contributing to the importance of localizing the clipped node [22].

4.3. Clinical Evaluation of the Axilla After NAC

A striking observation early in this experience was the relatively high incidence of patients having planned ALND who were ypN0. This observation reinforced the inadequacy of clinical examination to identify persistent histologically positive axillary lymph nodes after NAC. Further evaluation of this same patient cohort was done to address whether axillary lymph node imaging was accurate enough to correctly identify the final axillary node status at the time of WD‐SLND [23]. The positive predictive value of abnormal post‐NAC axillary imaging was 48% for ultrasound and 53% for MRI. The negative predictive value for normal post‐NAC axillary imaging was 67% for ultrasound and 68% for MRI. The results of axillary imaging were not adequate to accurately identify lymph node status after NAC, even when done by dedicated breast radiologists. These observations led in 2019 to the more aggressive use of targeted axillary dissection in this cohort of patients using wire localization of clipped axillary lymph nodes and WD‐SLND. Even with the restriction of planned ALND after making the switch in strategy towards more aggressive surgical staging of the axilla, the false‐positive rate of clinical examination of the axilla post‐NAC remained unacceptably high. The imaging findings from this cohort of patients described above are consistent with the results of the recently presented prospective AXSANA registry trial, which evaluated the diagnostic performance of axillary ultrasound after NAC in 3260 patients with initially node‐positive breast cancers [24]. In that trial, the positive predictive value of ultrasound was 64.2% and the negative predictive value was 64.3%. Our conclusion is that virtually all patients should have a sentinel lymphadenectomy after NAC, regardless of clinical findings in the axilla. The results of axillary staging will clarify the roles of ALND and radiation therapy to regional lymph nodes, with the publication of results from the Alliance 011202 and NSABP B‐51 trials [18, 25].

This study had several potential limitations. It was a retrospective, single‐institution study with potential for selection and observer bias. Although there was a consistent rationale over the course of this study for systemic treatment, and adjuvant radiotherapy treatments were standardized by institutional treatment regimens, individual treatment plans differed by patient, and these differences were not controlled for in data analysis. Chemotherapy regimens also changed over time to include adjuvant chemotherapy and immunotherapy agents based on completion of recently reported clinical trials and FDA approval of drugs. In addition, differentiation between anthracycline and non‐anthracycline regimens was not considered in the analysis. Finally, blue dye was not routinely used in sentinel node mapping unless patients did not map with technetium 99m sulfur colloid or were enrolled in a clinical trial that required its use.

5. Conclusion

ALND was avoided in approximately half of patients who presented with axillary node metastases and became ycN0 post‐NAC. Response to NAC correlated with node status at surgery. Approximately half of the patients with a histologically positive sentinel lymph node after NAC had residual positive non‐sentinel lymph nodes at ALND. Patients who do not map after NAC may not need ALND if a pN+ clipped node is retrieved and converted to ypN0.

Synopsis

Axillary lymph node dissection (ALND) can be avoided in approximately half of patients who present before neoadjuvant chemotherapy (NAC) with axillary node metastases and become clinically node‐negative post‐NAC. Patients who do not map may not need ALND if a core biopsy‐proven metastatic axillary clipped node is retrieved and is histologically node‐negative at targeted ALND.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.