Implementation of the Findings From the American College of Surgeons Oncology Group Z0011 Study in Axillary Management in Patients With Invasive Breast Cancer: A Cohort Study in a Brazilian Public University Hospital

Background: To evaluate and compare overall survival and locoregional recurrence between patients with invasive breast tumors and sentinel node metastasis undergoing sentinel lymph node dissection (SLND) alone and those undergoing\\ complete axillary lymph node dissection (ALND). Methods: In this retrospective cohort study, we reviewed medical records of all consecutive patients with primary invasive breast carcinoma who had undergone conservative surgery at a public university hospital in Brazil between 2008 and 2018. We evaluated the overall survival and the onset of locoregional recurrence using Kaplan-Meier and Cox regression analyses, respectively. Results: Overall, 97 participants underwent conservative breast surgery, 41 in the ALND group, and 56 in the SLND group. The mean age was 57.8 years. Only 17% of the patients in the ALND group had an additional biopsy-proven axillary disease, and 83% were treated with complete dissection unnecessarily. The 5-year survival rates were 80.1 and 87.5% in the SLND and ALND groups, respectively (p = 0.376). Locoregional recurrence was rare (1.8% and 7.7% in the SLND and ALND groups, respectively; p = 0.196). Intraoperative assessment was performed in 90.2% and 30.8% cases (p < 0.00001) before and after ACOSOG Z0011, respectively. Additionally, a second surgical procedure was performed in 3.8% of cases after ACOSOG Z0011 criteria adoption. Conclusions: Overall survival and locoregional recurrence were similar between the two groups. The de-escalation of ALND to SLND in women with metastasis in the sentinel lymph node treated with conservative surgery and radiotherapy is feasible even in developing countries. of real-world cases evaluating the effects of SLND alone in patients with early-stage breast cancer and one or two positive SLN metastases in the post-Z011 era showed equivalent survival and recurrence outcomes between those undergoing SNLD alone or ALND, demonstrating that SLND alone was safe. However, this shift in clinical practice should not occur in patients with residual lymph node disease following neoadjuvant chemotherapy. All these studies included patients who were treated with systemic adjuvant therapy.


Introduction
Breast cancer is the most common type of cancer worldwide and is responsible for 15% of cancer-related deaths among women. 1 Sociodemographic index levels are signi cant determinants of breast cancer incidence and mortality. In low-income and middleincome countries, the incidence rate of breast cancer is low. 2 However, breast cancer mortality remains high due to limitations in early diagnosis and treatment options. 3,4 In recent decades, the standard axillary management of early-stage breast cancer has changed dramatically. Axillary lymph node dissection (ALND) has been replaced by sentinel lymph node dissection (SLND) for treating patients with node-negative breast cancer. 5 The American College of Surgeons Oncology Group (ACOSOG) Z0011 was a milestone in the surgical treatment of the axilla in patients with early breast cancer that contributed signi cantly to reducing the extent of breast surgery. [6][7][8][9][10][11] Consequently, treatment of the axilla resulted in reduced morbidity and improved quality of life. [12][13][14] ACOSOG Z0011 was a randomized, and non-inferiority clinical trial, including women with invasive breast tumors sized up to 5 cm, with clinically negative axilla and up to two metastatic sentinel lymph nodes. Patients were treated with conservative surgery, breast radiotherapy, and systemic adjuvant therapy. That study concluded that complete ALND neither signi cantly improved overall survival or disease-free survival nor did it reduce locoregional recurrence in these patients. 15,16 The ACOSOG Z0011 data were updated with a 10-year follow-up, and the results con rmed the evidence, which resulted in fundamental changes in the surgical management of the axilla. 17,18 Other studies have corroborated the results of the Z0011trial. NSABP B4 found no bene t with resection of positive occult lymph nodes at the time of surgery. 19 IBCSG 23 − 01 also evaluated patients with minimal lymph node involvement and showed that ALND could be avoided. 20 The AMAROS trial con rmed that axilla treatment (surgery or radiotherapy) in patients with metastatic axillary sentinel lymph nodes provides comparable axillary control and there were no signi cant differences in overall survival between treatment groups. Additionally, axillary radiotherapy can be used as an alternative to ALND in patients with metastasis in the sentinel lymph node and ful lling the ACOSOG Z0011 exclusion criteria. 21 However, several shortcomings were identi ed in the Z0011 trial, such as enrollment not meeting the accrual goal, absence of standard testing for human epidermal growth factor receptor 2 (HER2), doubts about the radiotherapy elds, and applicability to other patient populations. 22 In Brazil, the utility of the conservative approach to the axilla in patients with positive sentinel lymph node biopsies has been challenged because survival after breast cancer remains poor in developing countries 23 due to delays in diagnosis, which occurs in more advanced stages when treating the disease is di cult. 24 In addition, there is resistance in some cancer centers to adhere to the new surgical approach. These facts motivated this study.
This study aimed to validate the applicability of the Z0011 trial approach by evaluating and comparing the overall survival and locoregional recurrence in patients with invasive breast cancer who underwent either SLND as indicated by biopsy or complete ALND.

Study design, setting, and ethics
This was a retrospective cohort study based on medical records evaluating the survival in consecutive patients with primary invasive carcinoma of the breast who underwent conservative surgery at a public university hospital in Brazil (Hospital São Paulo, Universidade Federal de São Paulo, UNIFESP) between February 2008 and December 2018. We evaluated overall survival and locoregional recurrence in patients who underwent either SLND or ALND. This project was approved by the institutional review board. Patients were informed about the study objectives, and they agreed to have their clinical data used in the research without any incentive and signed an informed consent form.
Participants, sources of data, and treatments received We reviewed an electronic database of the surgical ward records to identify patients (only women) who underwent surgery during the study period for a primary breast tumor of size up to 5 cm. We excluded patients who underwent a mastectomy, complete axillary lymph node resection without a previous sentinel lymph node biopsy, or those who received neoadjuvant therapy. We considered only women who underwent conservative breast surgery and sentinel lymph node biopsy. Next, we excluded patients who had negative sentinel lymph node biopsy test ndings. Therefore, only those who underwent conservative surgery (quadrantectomy) and had positive sentinel lymph nodes were nally included. The axillae of these women were clinically negative (N0).
We divided the participants into two groups: the ALND group included patients who underwent complete ALND and SLND group included those who underwent dissection of only the axillary nodes indicated by the sentinel lymph node biopsy. Figure 1 shows the treatment protocols for patients at our hospital before and after the publication of the Z0011 trial. The only Z0011 trial criterion that was not considered at our hospital was the presence of extracapsular extension identi ed as focal or smaller than 2 mm. These patients were treated with SLND.
During the study period, ultrasonography was not routinely performed at our hospital. Sentinel lymph node biopsy was performed using patent blue dye, and some patients were treated using a combined technique with technetium-99m. Lymphoscintigraphy was not routinely performed because of the lack of resources at our hospital. Intraoperative assessment of sentinel lymph nodes could be performed during surgery at the surgeon's discretion.

Study endpoint, variables, and sample size calculation
We compared survival between the groups. The primary endpoint was overall survival, which was de ned as the time between surgery and death for any cause. Secondary endpoints were locoregional recurrence, which was determined by the return of the disease either in the breast, ipsilateral lymph nodes in the axilla, or in the internal supraclavicular, infraclavicular, or thoracic (mammary) chains and survival after the recurrence of the disease, which was considered as the time between surgery and the onset of locoregional recurrence.
We also compared sociodemographic and clinical variables such as age group (younger than 50 years or aged 51 years or older), race, educational level, histological diagnosis and grade, angiolymphatic invasion, hormonal receptors, HER-2 positivity, Ki-67, tumor size, number of positive lymph nodes, and therapy between the groups.
The sample size was calculated to detect the non-inferiority of risks (hazard ratio) with a power of 85% and a signi cance level of 5%. We admitted a non-inferiority margin of 0.30 for the risks and a follow-up of 5 years, in which we assumed an approximately exponential distribution. A minimum sample size of 43 patients for each group (total, 86 patients) was determined. We used PASS 14 software (Power Analysis and Sample Size System, NCSS) for this calculation.

Statistical analysis
We analyzed the data descriptively. For categorical variables, we present absolute and relative frequencies and, for numerical variables, summary measures (mean, quartiles, minimum, maximum, and standard deviation).
We veri ed associations between two categorical variables using a Chi-square test or in cases of small samples with a Fisher's exact test. We compared two means using a student's t-test for independent samples, with the assumption of normality veri ed using a Kolmogorov-Smirnov test.
The survival analyses in this study evaluated the time until the occurrence of death or recurrence, considering censorship (cases that did not experience the event during the analysis period). Initially, we analyzed survival functions separately for each predictor variable (univariate analysis). Kaplan-Meier models were used for categorical variables. We estimated survival functions for each level of these variables, and then compared them using a log-rank test (Mantel-Cox).
Additionally, we adjusted univariate Cox regression models for all the predictor variables considered, and a multivariate Cox model was adjusted for signi cant characteristics identi ed in the univariate models (backward method). A Cox's model assumes the existence of proportional risks, which was veri ed via a test based on Schoenfeld residuals.
All statistical tests were two-tailed, and a signi cance level of 5% was used. Statistical analyses were performed using the statistical software Statistical Package for the Social Sciences (SPSS) 20.0 and STATA 12.

Patient characteristics
During the study period, 1009 patients underwent surgery for breast cancer in our hospital. After applying the exclusion criteria, we identi ed 97 patients with positive lymph nodes who underwent conservative breast surgery. These patients were diagnosed with invasive breast carcinoma, with tumors of 2 cm or less (T1) or 2.1 cm to 5 cm (T2), all with clinically negative axilla (N0). There were 41 and 56 patients in the ALND and SLND groups, respectively ( Figure 2).
The mean age was similar between the groups ( Table 1). The mean tumor size was 1.7 cm, and signi cantly larger in the ALND group (p = 0.048). The number of positive lymph nodes was signi cantly higher in the ALND group (median, 2) than in the SLND group (median: 1). Only one lymph node was positive in 91% of the cases. In the ALND group, only 17% of patients had an additional axillary disease (as shown by biopsy), suggesting that complete dissection performed in 83% of patients was unnecessary. The ALND group had a signi cantly higher number of total lymph nodes removed (median, 14) than the SLND group did (2 nodes). The ALND group also had a signi cantly longer follow-up time than the SLND group did (median, 5.3 vs. 3.5 years).
There was no difference in race, educational level, histological diagnosis and grade, hormonal receptors, HER-2 positivity, tumor size (according to TNM pathological staging), and angiolymphatic invasion between the groups (Table 2). However, the ALND group had a signi cantly higher number of positive lymph nodes and macrometastases (p < 0.001 for both), as well as higher axillary involvement, as seen by the higher frequency of pN1, pN2, and pN3 cases, than the other group did (p < 0.001). The SLND group had less extracapsular extension (p < 0.001). Adjuvant chemotherapy was more frequent in the ALND group (p = 0.024). Furthermore, the types of radiotherapy and hormonal therapy performed were similar between the groups.
Radiation oncologists at our institution used nomograms to predict the likelihood of metastasis in non-sentinel lymph nodes and helped in clinical decision-making. The most commonly used nomograms were from the Memorial Sloan-Kettering Cancer Center 25 and MD Anderson Cancer Center. 26 Patients who had greater than 30% risk of additional lymph node involvement were treated with drainage radiotherapy, 27 although it is not recommended in the ACOSOG Z0011 protocol. We identi ed 12.5% and 4.9% (p = 0.12) cases that received drainage radiotherapy in the SLND alone and ALND groups, respectively.
There was a signi cant reduction in the intraoperative assessment of sentinel lymph nodes after the publication of the ACOSOG Z0011 clinical trial and its adoption as a guideline at our hospital. Of the 415 patients evaluated, 90.2% (46 of 51 patients operated) were subjected to the exam before and 30.8% (112 of 364) after the publication. The rate of patients undergoing a second surgery as a result of the anatomopathological result was 3.8%, and the main indications for the second surgical procedure were the presence of three or more positive lymph nodes and gross capsular extravasation.

Overall survival
The mean overall survival was 9.18 years (95% con dence interval [CI], 8.47-9.90). There was no difference in survival between the two groups; the 5-year overall survival was 80.1% and 87.5% in the SLND and ALND groups (p = 0.376), respectively ( Figure 3). Only hormonal therapy had a signi cant effect on survival; patients who had positive hormone receptors lived longer (Table 3). Patients with negative receptors for estrogen (ER) and progesterone (PR) had shorter survival, but only two patients were classi ed as such; therefore, they were not considered ( Table 1). All patients positive for HER2 received targeted therapy, and there was no difference in survival between patients with positive and negative HER2 status.
In the survival model, only hormonal therapy had a signi cant effect (p = 0.018, Table 4), and patients receiving this therapy had a 78% lower risk of death. As in the previous analysis, the two patients who were ER-and PR+ were not considered.
The multivariate Cox regression model included groups, hormonal therapy, tumor size (pT), the number of resected lymph nodes, capsular extension, the size of axillary metastasis, and adjuvant chemotherapy ( Table 5). We did not consider the number of affected lymph nodes because some staging categories were present in only one group. In this analysis, there was no difference in survival between the groups (p = 0.536) ( Table 5). However, the risk of death was 91% lower in patients undergoing hormonal therapy in the nal model (p = 0.005) and 86% lower in those treated with adjuvant chemotherapy (p = 0.027). The Schoenfeld residuals testing showed that the hazards were proportional in the initial model (p = 0.453) and at the end (p = 0.194), indicating the absence of violation of this assumption.

Locoregional recurrence
Locoregional recurrence was a rare event, with only four patients having disease recurrence: 7.7% of patients in the ALND group and 1.8% in the SLND group. Recurrence occurred in all cases within 18 months of follow-up. There was no signi cant difference in survival after disease recurrence (p = 0.196) between the groups (Table 6). Survival after recurrence was 10.1 years (95%CI 9.62-10.40). Figure 4 shows the accumulated survival per group.

Discussion
This study showed that survival was not improved in patients with invasive carcinoma when we performed complete ALND. The similar overall survival between the two groups provides evidence that ALND is unnecessary in patients with metastatic sentinel lymph nodes treated with conservative surgery and radiotherapy. This nding suggests that even in countries such as Brazil, where the overall survival in patients with breast cancer is lower than that in patients in developed nations, 23 conservative surgical treatment of the axilla is possible. This result corroborates the data from the ACOSOG Z0011 trial. 15 Patients who underwent SLND alone had a lower survival rate, although this was not statistically signi cant. This is probably because, in our study, patients in the SLND group received adjuvant chemotherapy less frequently (75% vs. 92.7% in SLND alone and ALND, respectively, p = 0.024, Table 1). Additionally, they were administered adjuvant chemotherapy because of comorbidities, and these data were not evaluated in the study.
The tendency towards reducing the use of intraoperative lymph node evaluation after the publication of the ACOSOG Z0011 trial was similar to that in other studies. 6,7,28,29 Intraoperative assessment of the sentinel lymph node can be associated with a shorter average time of surgery, 6 a reduction in perioperative costs, 30,31 and a signi cant increase in the proportion of patients in whom complete dissection can be avoided. 32 According to van der Noordaa et al., 33 intraoperative assessments of the sentinel lymph node should be performed only in patients with a restricted indication of lymph node dissection in the presence of metastasis in sentinel lymph node biopsy. Thus, the intraoperative assessment of the sentinel lymph node is not necessary for patients who meet the ACOSOG Z0011 criteria, and the surgical re-approach resulting from the de nitive anatomopathological result of the axilla is rare (3.8%). This is an important nding in our study that can promote the practice of avoiding the intraoperative assessment of sentinel lymph nodes.
Locoregional recurrence was rare and similar between the groups. We believe that we had adequate follow-up time to evaluate recurrence. Long-term follow-up data from the NSABP trial 19 showed that recurrence usually occurred early, at 14.8 months on average. In the ACOSOG Z0011 trial, 18 recurrences occurred in 3.1 years, a bit shorter than our average follow-up time of 3.7 years.
The SLND group included women who were postmenopausal, with small tumors (pT1), positive hormone receptors, and small axillary involvement (35.7% with micrometastasis in the sentinel lymph node biopsy). These characteristics were similar to those of the same arm in the Z0011 trial. 15,18 However, in the Z0011 study, 15 the arm undergoing complete axillary dissection also had a high prevalence of micrometastases (37.5%), different from our study.
For patients with HER2 overexpression, triple-negative tumors and those aged below 50 years, using the ACOSOG Z0011 trial criteria can be discussed. Chung et al. 34 reported no bene t of performing ALND in this subgroup. In our study, the groups were homogeneous in terms of these three variables. The underrepresentation of this group in the ACOSOG Z0011 trial may be due to the local demographic characteristics of patients with breast carcinoma. Nevertheless, it was assumed that the distribution of HER2 positive tumors was balanced between the two arms of the trial.
Several studies around the world have identi ed increasing acceptance of the Z0011 results and a change in clinical practice in relation to the standard treatment of axillary lymph nodes in patients with breast cancer. [35][36][37][38] .
A meta-analysis comparing SLND/radiotherapy only with ALND in early-stage breast cancer with limited sentinel node metastasis estimated that overall survival and disease-free survival were higher in the SLND group than in the ALND group and observed a greater axillary recurrence in the SLND/radiotherapy group. In conclusion, the omission of ALND in patients with one or two sentinel lymph nodes (SLNs) is indicated. 39 Another meta-analysis of real-world cases evaluating the effects of SLND alone in patients with early-stage breast cancer and one or two positive SLN metastases in the post-Z011 era showed equivalent survival and recurrence outcomes between those undergoing SNLD alone or ALND, demonstrating that SLND alone was safe. 40 However, this shift in clinical practice should not occur in patients with residual lymph node disease following neoadjuvant chemotherapy. 41 All these studies included patients who were treated with systemic adjuvant therapy.
Complete ALND might be an overtreatment for many patients with capsular extravasation in the dissected sentinel lymph nodes. The Z0011 trial excluded patients with gross capsular extravasation and did not analyze the effect of microscopic capsular extravasation on recurrence or survival, making the management of these patients uncertain. 15,18 The extension of capsular extravasation is directly associated with the burden of axillary disease. 42 However, the rates of local, regional, or distant recurrence or mortality were similar between patients with and without capsular extravasation of ≤ 2 mm, 43 and regional recurrence was rare and equal to that in patients without capsular extravasation even in the absence of nodal radiotherapy. Capsular extravasation is not the only reason for complete ALND. 44 In our study, we identi ed ve patients with capsular extravasation of ≤ 2 mm who were treated with SLND alone, avoiding the morbidity associated with complete axillary resection. However, these patients received regional radiotherapy at our hospital.
We acknowledge that translating the Z0011 results into clinical practice is complicated by the inconsistent use of radiotherapy elds in their study. In a prospective study of 793 patients with sentinel lymph node metastasis, using the ACOSOG Z0011 eligibility criteria resulted in the avoidance of ALND in 84% of patients, and the 5-year cumulative regional recurrence rate was 1%, which did not differ between radiotherapy elds. The authors concluded that even without the routine use of nodal radiotherapy, complete dissection could be avoided with excellent regional control. 45 Hopefully, we will have answers about the real in uence of radiotherapy in regional control with the results of the ongoing trials. 46-49 .
This was a retrospective study based on the medical records, which did not allow us to evaluate costs and surgical times after the change in the clinical approach in our hospital after the publication of the ACOSOG Z0011 study. Studies that evaluated cost reduction associated with the elimination of complete axillary dissection, 30,31 did not consider the risk of surgical re-approach due to the presence of more than two sentinel lymph nodes compromised with macrometastasis or capsular extravasation. The cost of a second surgery remains to be evaluated. Even the ACOSOG Z0011 trial did not report the rate of surgical re-approach in the group subjected to SLND alone. The rate of surgical re-approach in this study was too low to answer this question. Nevertheless, this was the rst study in our country to address the implementation of the ndings of Z0011 and was important to encourage conservative surgical treatment of the axilla in our country and other developing countries, with the aim of disseminating this practice and bene ting patients.
The preliminary internal evaluation of the results of this study prompted major changes in our hospital's clinical approach, with more conservative surgeries being performed and the elimination of ultrasonography, ndings of which would often cause patients to undergo radical lymphadenectomy in the absence of sentinel lymph node biopsy results in the past.

Conclusions
The overall survival and locoregional recurrence in patients with metastatic axillary sentinel lymph nodes treated with SLND were similar to those in patients treated with complete ALND. The elimination of routine axillary lymphonodectomy and the implementation of sentinel node biopsy at our hospital bene ted the patients who could be treated with less aggressive surgery. The de-escalation of ALND to SLND in women with metastasis in the sentinel lymph node treated with conservative surgery and radiotherapy is possible. Our study showed that it is possible to apply the ACOSOG Z0011 recommendation in developing countries.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
The rst author (VMS) was supported by a doctoral scholarship from Conselho Nacional de Desenvolvimento Cientí co e Tecnológico. No other funding was received for this study.
Authors' contributions VMS designed the study, performed the experiments, analyzed the data, and wrote the manuscript; SE revised the manuscript; GF revised the manuscript; ACPN designed the study, analyzed the data and revised the manuscript. All authors read and approved the nal manuscript.   However, she was later classi ed clinically as T1a. ER = estrogen receptor; HER2 = human epidermal growth factor receptor 2; PR = progesterone receptor Table 3. Kaplan-Meier survival analysis in patients with breast cancer undergoing ALND or SLND ALND = complete axillary lymph node dissection. (-) not shown due to lack of precision. HR = hazard ratio. Tis = ductal carcinoma in situ; T1= tumor size T2 = tumor size 20-50 mm. N0i+ = the area of cancer spread contains fewer than 200 isolated tumor cells and is smaller than 0.2 mm (cancer cells seen in routine stains or immunohistochemistry); N1mi = micrometastasis to lymph node; N1 = 1-3 lymph nodes affected; N2 = 4-9 lymph nodes affected; N3 = 10 or more lymph nodes affected. ER = estrogen receptor; PR = progesterone receptor; HER2 = human epidermal growth factor receptor 2; SLND = sentinel lymph node dissection; ALND = complete axillary lymph node dissection  Flow diagram of patient inclusion procedure used in the study Figure 3 Overall survival over time in patients with breast cancer undergoing ALND or SLND alone Legends: SLND = sentinel lymph node dissection; ALND = complete axillary lymph node dissection Survival over time in patients with breast cancer and locoregional recurrence Legends: SLND = sentinel lymph node dissection; ALND = complete axillary lymph node dissection