Introduction

Neoadjuvant endocrine therapy (NET) has gained renewed attention in the last 20 years. NET was historically used as an alternative to chemotherapy in older and frail patients with estrogen receptor (ER)-positive breast cancer [1]. However, we have accumulating evidence that NET is an effective and well tolerated alternative to postmenopausal patients with ER-positive/HER2-negative breast cancer, not only to downstage tumors and allow less-extensive surgery, but as a scientific platform to obtain information on tumor resistance and potential biomarkers [2,3,4,5].

Residual tumor burden after neoadjuvant therapy has been validated as an important prognostic and predictive tool after neoadjuvant chemotherapy [6, 7]. Pathological complete response (pCR) can predict long-term outcomes and the benefit of additional systemic therapy after surgery, especially in triple-negative and HER2-positive breast cancer [8,9,10,11]. In ER-positive/HER2-negative breast cancer, pCR is rare either with chemotherapy (6–8%) and endocrine therapy (1%), and an association with long-term outcomes appears to valid only for high-grade tumors submitted to neoadjuvant chemotherapy [3, 9, 12, 13]. In an attempt to identify predictors of long-term outcomes after NET, a clinical multiparametric tool was developed and validated [4, 14]. Preoperative endocrine prognostic index (PEPI) takes into consideration tumor size, lymph node involvement, ER expression, and Ki67 expression on surgical specimen after 4 to 6 months of NET. Tumors with lower scores are associated with excellent prognosis [4, 14].

Randomized controlled trials (RCTs) are an ideal design for evaluating the effects of medical interventions under strictly controlled conditions. Patient selection and randomization help reduce bias and guarantee internal validity. However, these highly selected populations may not represent the more heterogeneous characteristics of patients seen in clinical practice, jeopardizing the generalizability of the results (external validity). Frequently drugs do not show the same results in clinical practice as reported in clinical trials, and this disparity of findings has been termed “efficacy-effectiveness gap” [15]. Real-world data (RWD) is being increasingly used to bridge that gap, by regulatory agencies, healthcare providers and even pharmaceutical industries. This kind of data provides useful insights into treatment access, effectiveness, toxicities, as well as the quality of care, guiding quality improvement interventions [16,17,18]. There are no RWD regarding the use of NET in ER-positive/HER2-negative breast cancer.

The aim of this study was to analyze prospectively collected data from patients submitted to NET and report RWD on patient characteristics, treatment, and surgical outcomes in clinical practice.

Methods

This was a prospective cohort study of luminal breast cancer patients treated with NET at three tertiary hospitals in Brazil (Women’s Hospital Prof. Dr. José Aristodemo Pinotti—Campinas/SP—CAISM, Pontifical Catholic University, PUC—Porto Alegre/RS—PUCRS, and Serra Gaúcha Research Center, Caxias do Sul/RS—CEPESG). Institutional Review Board approval of the study protocol was obtained in all participating institutions. Patients were included if they were postmenopausal with histological diagnosis of stage I-III ER-positive/HER2-negative breast cancer, and were submitted to NET, with treatment initiated between January 2016 and December 2018. We excluded patients with metastatic disease, who received neoadjuvant chemotherapy, refused or were unfit for surgery, and those lost to follow-up. We evaluated a total of 392 patients submitted to neoadjuvant therapy. Among them, 146 received NET and were included in the analysis. (Fig. 1) Since only postmenopausal patients were considered for NET, none of the patients received concomitant ovarian suppression.

Fig. 1
figure 1

Flow chart of patient selection process

Full size image

Data were prospectively collected from patient charts, including age at diagnosis, clinicopathological variables, type and duration of treatment, type of breast and axillary surgery (sentinel lymph node biopsy, SLNB; or axillary lymph node dissection, ALND), and pathological specimen analyses. Pathological complete response was defined as absence of invasive disease in the breast and axilla (ypT0/is ypN0) [9]. Immunohistochemistry for ER (clone 1D5, 1:1.000, Dako) and PR (clone PR 636, 1:800, Dako) expression was done according to international consensus, and positivity was considered when at least 1% of the nuclei stained [19, 20]. HER2 expression was initially evaluated with immunohistochemistry (clone PN2A, 1:1.000), and staining was scored as 0 + /1 + (negative), 2 + (equivocal), and 3 + (positive). Equivocal cases were, then, confirmed by in situ hybridization, according to the ASCO/CAP recommendations [21]. Ki67 expression was conducted by immunohistochemistry (clone MIB1, 1:500, Dako) and defined as the mean expression in the whole tumor area [22]. Preoperative Endocrine Prognostic Index (PEPI) calculation were done as described by Ellis et al.[14] Clinical and pathological stage were defined according to the American Joint Committee on Cancer (AJCC)/International Union Against Cancer (UICC) TNM staging 8th edition [23].

We conducted descriptive analyses on the data collected for this study. Categorical variables were described as frequency and percentage, while continuous variables were reported as median. Data were collected and managed using REDCap (hosted by the State University of Campinas).

Results

Characteristics of the cohort of 146 patients are described in Table 1. Median age was 67 years old (51 to 93 years old), 73 (50%) had T2 tumors, and 128 (87.8%) were stage I-II disease. Ductal invasive carcinoma comprised 75.3% of tumors (n = 110), 111 (76.1%) were histological grade II, and median Ki67 expression were 10.0% (1–80%). ER positivity with Allred scores 3–8 were found in 134 patients (91.8%), and for PR in 114 patients (78.1%). Aromatase inhibitors (anastrozole or letrozole) were used by 99.5% of patients (n = 145), and median duration of treatment were 21.0 weeks (2 to 43 weeks). We did not observe clinical progression during treatment.

Table 1 Patient and treatment characteristics
Full size table

92 patients (63.0%) were submitted to breast conserving surgery, and sentinel lymph node biopsy (SLNB) were done in 112 patients (76.7%). Among patients in whom a SLNB was performed, 73 (50.0%) had a negative result. A positive SLNB was followed by axillary lymph node dissection (ALND) in 18 patients (12.3%). Upfront ALND was performed in 34 patients (23.3%) (Table 2).

Table 2 Surgical management after neoadjuvant endocrine therapy
Full size table

As shown in Table 3, pathological complete response (pCR) was achieved in only 2 patients (1.0%). 65 patients (44.5%) had residual pathological T2 tumors, and 82 patients (56.2%) were N0. Regarding PEPI score, 43 patients (29.5%) had score 0, and 38 (26.0%) had PEPI 4–5. 7 patients (4.8%) had HER2-positive disease at residual tumor specimen. Posttreatment Ki67 expression had a median value of 3.0% (1–80%). Figure 2 represents the change in Ki67 expression for each tumor before and after NET, showing that the majority of them had a decrease in Ki67 expression, and a marked increase in five cases (3.4%).

Table 3 Pathological outcomes after neoadjuvant endocrine therapy
Full size table
Fig. 2
figure 2

Box plot of Ki67 expression before and after treatment (n = 137). Ninety-two tumors (67.1%) showed reduction in expression values, 23 (16.8%) had stable values, and 22 (16.1%) showed paradoxical increase in Ki67 expression (including 5 tumors with marked increases)

Full size image

Detailed information regarding PEPI score individual components for the whole cohort are shown in Table 4. The majority of tumors were pathological T1/T2 (91.8%), and 40.4% had positive lymph nodes. Almost half of tumors (46.5%) had low levels of Ki67 expression (0–2,7%). All tumors with available information for hormone receptor expression were ER-positive.

Table 4 PEPI score components distribution for the whole cohort
Full size table

Adjuvant chemotherapy was given to 26 patients (17.8%), and all patients (n = 7) with residual HER2-positive disease received HER2-targeted therapy. Adjuvant radiation therapy was performed in 102 patients (69.9%). (Table 5).

Table 5 Adjuvant therapies
Full size table

Discussion

Our study examined the use of NET in real-world setting, showing that it is a safe and effective neoadjuvant therapy in postmenopausal patients with ER-positive/HER2-negative breast cancer in clinical practice. Our analysis of real-world patients compared with phase 3 clinical trials of NET revealed a similar median age at treatment initiation, although slightly inferior than in IMPACT trial [3,4,5]. Half of our patients had T2 tumors, and most patients were classified as having stage I disease. Accordingly, in P024 trial approximately half of patients had T2N0 disease, and median tumor size were 3.6 cm in PROACT trial, and 3.8–4.0 cm in IMPACT trial [3,4,5]. A tumor size of at least 3.0 cm was required for enrollment in P024 and PROACT trials, but one third of our cohort patients had tumors of 2.0 cm or less [3, 5]. In our clinical practice, we offered NET event to patients with small, operable tumors while we wait for surgery, due to long waiting times, especially in the first months after inclusion of NET in our clinical protocols. Also, older patients initially deemed unfit for surgery went on to surgical management after a period of NET.

Tumor proliferation as measured by Ki67 expression was markedly reduced in the majority of our patients. Posttreatment Ki67 median expression was 70% lower than pretreatment value, and virtually all patients were treated with an aromatase inhibitor. Endocrine therapy induces cell-cycle arrest, and evaluation of tumor proliferation may be a surrogate of treatment effect [24]. Indeed, data from P024 trial showed that Ki67 expression in surgical specimen was associated with disease-free survival and breast cancer specific survival [14]. Mean reduction in Ki67 expression with aromatase inhibitors were of 87% in P024, 81.6% in IMPACT, and 78% in ACOSOG Z1031 [12, 14, 25]. Although our median pretreatment Ki67 expression was low, we had some patients with higher Ki67 expression, probably because of ineligibility for chemotherapy. Four tumors had a marked increase in Ki67 expression, and two of them revealed HER2-overexpression at surgical specimen.

The optimal duration of NET was not defined. In our cohort, treatment duration varied from 2 to 43 weeks, with a median time of 21 weeks. Shorter treatment periods occurred mostly during the first months of protocol implementation and is part of the learning curve. Our goal is to maintain NET for at least 16 weeks, as recommended for PEPI score calculation. In clinical trials of NET, treatment was recommended for a period of 3–4 months.[3,4,5] However, several studies have suggested that a longer duration of treatment are required for maximum effect, and approximately one third of patients achieve maximal response after 6 months.[26,27,28,29] Our median NET duration approaches 6 months and, although longer therapy duration may be associated with the development of resistance, none of our patients progressed during treatment.

The majority of patients in our cohort were submitted to breast-conserving surgery. We did not collect data on surgery conversion rates, but our rate is somewhat higher than was reported for AI in clinical trials [3,4,5]. This may be explained by the inclusion of a large proportion of T1 tumors and the longer treatment duration in our cohort. Trials that compared NET to neoadjuvant chemotherapy reported breast-conserving surgery rates of 24 to 54.6% with chemotherapy [30,31,32]. For example, NEOCENT trial compared neoadjuvant letrozole to anthracycline-based chemotherapy, reporting breast-conserving surgery rates of 68.2% vs 54.6%, respectively [32]. Most patients in our cohort were submitted to SLNB, and less than one fourth had a positive sentinel node. In ACOSOG Z1031 the rate of SLNB ranged from 31.4 to 61.3% in tumors with a week 2 Ki67 > 10% or ≤ 10%, respectively.

Our pCR rate is consistent with previous studies on NET [3]. We observed a PEPI 0 rate of 29.5% and PEPI 1 of 17.0%. In ACOSOG Z1031, approximately 17% of patients had PEPI 0 [12]. This difference could be due to the higher proportion of luminal A-like tumors in our cohort. Endocrine therapy-resistant disease (PEPI 4–5) comprised 26% of our patients, and some of these patients (39%) received adjuvant chemotherapy, because of the perceived high risk of recurrence and mortality in these group of patients. None of our patients with PEPI 0 received adjuvant chemotherapy, in contrast to 11% in ACOSOG Z1031. On the other hand, 25.4% of our PEPI > 0 patients received adjuvant chemotherapy, in comparison to 41.2% in ACOSOG Z1031 [33]. This reflects the fact that the best approach after NET is still undefined. This issue is being evaluated in ALTERNATE trial (NCT01953588) [34]. A small subset of patients with low PEPI scores were submitted to adjuvant chemotherapy due to HER2-overexpression in surgical specimens. The rate of conversion to HER2-positive disease was of 4.8%, in accordance with what was previously published with neoadjuvant chemotherapy [35].

The strengths of our study include the inclusion of a more diverse group of patients, when compared to randomized clinical trials, and the fact that it reflects real-world treatment patterns, outcomes and decision-making process. Our study is limited by its nonrandomized nature, and the short-term follow-up. However, patients are still being followed and data on long-term outcomes will be available in the future. To the best of our knowledge this is the first prospective cohort to report on patients submitted to NET outside of a clinical trial.

Furthermore, one potential concern regarding NET is the risk of disease progression, in part due to the cytostatic, and not cytotoxic, effect of these agents [24]. Several translational studies have sought to evaluate potential biomarkers to identify patients who would show better responses to NET and those with primary resistance, such as the incorporation of algorithms based on the early effect on tumor proliferation (Ki67 expression after 2–4 weeks of treatment) and genomic profiling as in the 21-gene recurrence score assay (Oncotype DX) [34, 36, 37]. It is noteworthy that none of the patients in our real-world cohort had disease progression during NET, even though they were selected based solely on traditional clinical and pathological data, without the addition of modern molecular biology data. Our data highlights the ideal patient profile for NET (postmenopausal patients, tumors with luminal A-like features), and suggests that progression during NET is probably associated with its use in non-ideal patients (eg, luminal B-like tumors, and premenopausal patients). These findings are clinically relevant and should provide support for the adoption on NET in the clinical practice.

In conclusion, our prospective real-world data on postmenopausal patients with ER-positive/HER2-negative breast cancer shows that NET is a safe and effective alternative to chemotherapy, with similar outcomes than what was reported in the literature. These results may help increase awareness and adoption of NET in clinical practice.