Impact of Imaging-Guided Localization on Performance of Tailored Axillary Surgery in Patients with Clinically Node-Positive Breast Cancer: Prospective Cohort Study Within TAXIS (OPBC-03, SAKK 23/16, IBCSG 57-18, ABCSG-53, GBG 101)

This prospective study was preplanned within the randomized controlled international phase-III TAXIS trial (OPBC-03/SAKK 23/16/IBCSG 57-18/ABCSG-53/GBG 101; ClinicalTrials.gov Identifier: NCT03513614). Patients ≥ 18 years of age with histologically or cytologically proven node-positive breast cancer and adequate condition for breast cancer surgery were included. Node-positive disease was detected by palpation or imaging at the time of initial diagnosis. Clip placement in the biopsy-proven positive lymph node was required. Patients were included in both the upfront surgery setting and after NACT in case of residual nodal disease, which was confirmed by either intraoperative frozen section analysis or final histopathology. Patients with stage IV, cN3c or cN2b breast cancer, contralateral or other tumor malignancy within 3 years, prior axillary surgery (except SLN procedure), or prior axillary radiotherapy were considered ineligible. Furthermore, patients with nodal pCR, absence of clip in the specimen radiography, palpable disease left behind in the axilla after TAS, or no SLN identified in the axilla were excluded from randomization. This study was pre-planned after 500 patients were randomized (one-third of the total sample size) in the TAXIS trial. It was planned to gain relevant insight into the impact of IGL on performance of the pragmatic concept of TAS. Patients were treated from August 2018 to June 2022. Data extraction was performed after data cleaning on 30 September 2022.

This study was approved by the local ethics committees and performed in accordance with the requirements of the national regulatory authorities. Written informed consent was obtained from all patients. The study is reported according to the STROBE guidelines (Supplementary Table S1).²⁴

A total of 500 patients from 44 breast centers were included between August 2018 and June 2022. The median age at diagnosis was 57 years (IQR 48–69 years, Table 1). Of the excluded patients after registration, 72.8% (182/250) had nodal pCR after neoadjuvant treatment, in 4.0% (10/250) no SLN could be identified, in 15.2% (38/250) the clip was not removed during TAS, and 8.0% (20/250) of patients had other exclusion criteria. At the time of initial diagnosis, lymph node metastases were palpable in 51.6% (258/500), and detectable only by imaging in 48.4% (242/500) of patients. Of the 500 randomized patients, 67.0% (335/500) underwent upfront surgery and 30.2% (151/500) had residual nodal disease after NACT, while 2.8% (14/500) received neoadjuvant therapy other than chemotherapy (13 patients received neoadjuvant endocrine therapy; one patient received neoadjuvant double HER2-blockade without chemotherapy). Half of the patients (50%, 250/500) were randomized to undergo ALND—whereas 49.8% (249/500) actually received ALND—and 50% (250/500) to ART (Fig. 1). Of the 165 patients with neoadjuvant treatment, 79.4% (131/165) were randomized intraoperatively after confirmation of residual nodal disease by intraoperative frozen section analysis, whilst 20.6% (34/165) underwent randomization after primary surgery when residual nodal disease was confirmed on final histopathology.

IGL was attempted in 84.4% (422/500) of patients and was considered to have successfully localized the target node in 97.9% (413/422, Table 2). In excluded patients with known IGL status, IGL was attempted in 92.5% (210/227) and successful in 84.3% (177/210, Table S2). The IGL success rate was lowest in excluded patients in which no clip was found (34.4%, 11/32). The most common IGL technique was wire placement (53.8%, 227/422). IGL was used more often for nonpalpable than for palpable disease, with huge variation by country (p < 0.001, Table 3). The success of IGL was high with use of US (98%, 295/301). Other imaging modalities used before surgery included CT and MRI in a very limited number of patients (success rate for CT 100%, 3/3; MRI 100%, 3/3). The success of IGL did not depend on type of clip (Table 4). The type of clips used in excluded patients are presented in Table S2. The clip removal rate in all registered patients was 94.9% (712/750). In 74.4% (372/500) of randomized patients, the clipped node corresponded to a SLN. In 128 patients, in which the clipped node was not identified as a SLN, IGL was attempted in 80.5% (103/128), and was considered to have successfully localized the target node in 95.1% (98/103) of these patients.

There was no difference in any number of nodes removed by TAS with or without IGL, neither total [5 (IQR 3–8) versus 4 (IQR 3–6), p = 0.3], nor positive [2 (IQR 1–4) versus 2 (IQR 1–3), p = 0.6], nor negative [2 (IQR 1–4) versus 2 (IQR 0–4), p = 0.2, Table 5]. In multivariable linear regression models controlling for confounding variables, the effect of IGL on the number of total as well as positive nodes removed remained nonsignificant (Supplementary Table S3). However, while the number of nodes removed during TAS with IGL remained stable over time [5 (IQR 3–8 in 2019 and 5 (IQR 3–8) in 2022, p = 0.8], it decreased significantly without IGL, from 6 (IQR 4–6) in 2019 to 4 (IQR 3–4) in 2022 (p = 0.015). The difference in the number of nodes removed by use of IGL was not statistically significant at any time point, neither in 2019 (p = 0.6), nor in 2022 (p = 0.6). Size of lymph node metastases did not differ by use of IGL (p = 0.7).

This is the first prospective study investigating the role of IGL in TAS for clinically node-positive breast cancer in patients undergoing upfront surgery or NACT with residual nodal disease. While IGL was frequently used, it had no impact on the number of positive or negative lymph nodes removed and residual nodal disease after TAS, with two-thirds of patients having at least one additional positive lymph node. Without use of IGL, the number of nodes removed during TAS decreased over the study period, suggesting that surgeons went through a learning curve and sampled fewer nodes at later time points. The study was, however, not designed to assess whether this small difference in number of nodes removed over time was clinically relevant. However, due to the limited impact of IGL on all other performance characteristics of TAS, we conclude that IGL is not a key element of TAS.

TAS consists of palpation-guided selective removal of obvious nodal disease, thereby tailoring the extent of axillary surgery to the extent of axillary disease. It further includes the SLN procedure to reduce the volume of microscopic disease, while IGL of clipped or suspicious nodes is optional.²¹ The underlying hypotheses are that ART can cure microscopic disease left behind in the axilla after TAS in both the neoadjuvant and adjuvant setting, and that the combination of TAS with ART is less harmful for patients compared with standard ALND in the context of extended regional nodal irradiation. Therefore, the new concept of TAS combines several established surgical techniques that are commonly used to stage the axilla in the adjuvant and neoadjuvant setting. While there is no new surgical technique per se involved in TAS, the novel concept is to use a targeted and limited approach to the axilla to treat cancer by selectively removing it, not to rule it out, as is the goal of TAD or the SLN procedure when used in today’s practice.

In the adjuvant setting, the main difference between TAS and the SLN procedure is that TAS allows IGL of nodes, and that the selective removal of palpably suspicious disease is a regular key component. Currently, since palpable disease is the main contraindication to the SLN procedure when used for staging purposes, it is rarely encountered. In the neoadjuvant setting, the main difference between the therapeutic concept of TAS, which serves as final surgical treatment of residual nodal disease, and the diagnostic concept of TAD is that the latter by definition requires IGL and is routinely followed by ALND in case of residual nodal disease.

When planning this study, we were interested to see how the pragmatic concept of TAS would be translated into clinical practice at the participating TAXIS study sites. Therefore, we extensively documented key aspects of TAS including details on preoperative workup (e.g., type of clips and imaging) and the intervention (e.g., number of lymph nodes removed, SLN- and localization techniques). Indeed, significant variation by country was observed. In Switzerland, for example, IGL was omitted in one-quarter of patients, whereas in Austria, Germany, and Hungary, IGL was used in almost all cases (Table 3). International expansion of study sites is currently ongoing. The prespecified evaluation of TAS after randomization of the total TAXIS sample size of 1500 patients will therefore further broaden its global applicability. When use of IGL was attempted, it was successfully performed in 98%. This was not depending on type of clip used (Table 4). However, patients with failed clip removal showed the lowest IGL success rate, suggesting that failed IGL may have increased the risk of missing the targeted node. In our cohort, the clipped node corresponded to an SLN in 74.4% of patients, a finding similar to data published by Kuemmel et al., in which the SLN and targeted lymph node were identical in 64.8%.¹³ Interestingly, IGL was also frequently used for upfront surgery of palpable disease, which reflects the popularity of IGL at some study sites. A prior TAXIS substudy was prespecified after the first 200 patients were randomized and included all details on the 96 patients who were excluded from the TAXIS trial, totaling 296 registered patients.²¹ It showed that the clip removal rate was 94.3% and the false-negative rate of TAS in patients undergoing completion ALND was 1.8%. Since the present substudy focused on use of IGL and its impact on TAS in the first 500 randomized patients, data collection on the 250 excluded patients was limited, and hence, no false-negative rate of TAS could be evaluated. However, since other relevant performance characteristics of TAS remained almost identical, such as the clip removal rate of 94.9%, we have no reason to assume that any of the key findings of the initial substudy have changed after expansion of the patient population.

To the authors’ knowledge, TAXIS is the only RCT addressing the omission of ALND in patients with palpable disease who undergo upfront surgery to date. However, several authors questioned the justification for ALND in this setting by showing low axillary tumor load in many of these patients, and a prospective observational study is currently investigating the selective use of ALND.^28‐30 We expect that further RCTs will soon be developed to investigate the omission of ALND during upfront surgery in light of the results of the RxPONDER and Mindact trials that questioned the routine use of chemotherapy in all patients with luminal node-positive breast cancer.^31,32

In the neoadjuvant setting, the only RCT besides TAXIS to investigate omission of ALND for residual nodal disease after NACT is the Alliance A011202 trial (NCT01901094). While we eagerly await the primary analysis of this trial, ALND remains standard care, apart from, perhaps, the lowest volume of residual nodal disease, where ART may be considered on an individual basis.¹⁸ Analyses from the SEER database of patients with residual nodal disease having undergone SLNB and RNI compared with ALND showed conflicting results.^33‐35 Recently, an analysis from the MARI trial, including also patients with residual nodal disease, showed a 3-year axillary recurrence-free interval of 98.2%. Residual disease in the MARI-node independently predicted disease recurrence in multivariable analysis. In this trial, four of five axillary recurrences occurred in patients with residual nodal triple-negative disease and the corresponding 3-year axillary recurrence rate was 3.4%.³⁶ We purposefully refrained from performing intrinsic subtype-specific analyses of IGL in TAS for the present substudy due to the limited sample size of non-HR+/HER2− breast cancer (n = 103). However, we prespecified the corresponding subgroup analyses in the TAXIS substudy that is planned to assess the performance of TAS after the full TAXIS sample size of 1500 patients is reached. Therefore, the relevance of finding positive nodes in the majority of patients after TAS with 29.3% of patients undergoing ALND having ≥ 4 additional positive nodes, currently remains unclear. Hence, we recommend waiting for oncologic outcome analysis of TAXIS before adopting the much less radical approach of TAS in clinical practice.

TAS targeted positive nodes with or without IGL. IGL proved to be feasible inasmuch as it could be successfully performed in almost all cases, irrespective of type of clip and localization method. However, IGL did not significantly change the performance of TAS, which left ≥ 2 positive nodes behind in 47.6% of patients.