Efficacy of neoadjuvant endocrine therapy compared with neoadjuvant chemotherapy in pre-menopausal patients with oestrogen receptor-positive and HER2-negative, lymph node-positive breast cancer

For hormone-responsive breast cancer, the question of which patients can safely be spared adjuvant chemotherapy has been extensively studied. For cases involving hormone receptor-positive tumours, younger patients (< 35 years old) show a poorer prognosis than older patients in terms of tamoxifen resistance [1]. However, this finding does not indicate that young patients should receive chemotherapy. The greater effects of chemotherapy on younger patients are likely to be partially explained by the endocrine effects of chemotherapy on ovarian function. Irrespective of the chemotherapy agent administered, iatrogenic amenorrhea is associated with an increased survival rate in oestrogen receptive (ER)-positive breast cancer [2, 3], and amenorrhea is a surrogate marker for effective treatment in hormone receptor-positive pre-menopausal breast cancer patients [4]. This finding suggests that irrespective of chemotherapy or endocrine therapy, amenorrhea is important for pre-menopausal patients who have an ER-positive, HER2-negative tumour. The trial assigning individualized options for treatment (TAILORx) and the EORTC 10041/BIG 03-04 (MINDACT) studies can identify patients with a low-risk profile, thereby justifying the omission of chemotherapy based on its potentially low benefit [5‐8].

Treatment decisions based on demographic characteristics and tumour burden have the potential to overtreat many individuals and undertreat others. The Oncotype DX assay has been studied for lymph node-positive breast cancer patients, and the SWOG 8818 study reported that the low-risk Oncotype DX group gained no additional benefits from chemotherapy [9]. SOFT and TEXT joint analysis showed excellent survival data that included lymph node-positive, ER-positive breast cancer patients who received endocrine therapy alone. These findings suggested that not all lymph node-positive, ER-positive, and HER2-negative tumours require chemotherapy. Apart from the tumour burden, identifying robust biological predictors of benefit is likely to be informative and useful. It has been shown that endocrine therapy, which can substitute for chemotherapy even though there is a high tumour burden in hormone-responsive breast cancer, is effective. The factors that determine chemotherapy sensitivity remain poorly understood and might not entirely overlap with negative predictors of endocrine sensitivity or prognostic determinants [10, 11].

Neoadjuvant treatment has benefits, which include reducing the tumour burden and evaluating the treatment response of the tumour. The response to neoadjuvant chemotherapy (NCT) is diverse and varies according to the intrinsic subtype. ER-positive, HER2-negative breast cancer shows a reduced pathologically complete response (pCR) rate, but better survival than other subtypes. Moreover, pCR is not related to survival in the luminal A subtype [12]. Therefore, the goal of neoadjuvant treatment for hormone-responsive tumours is to evaluate the treatment effects and reduce the tumour size prior to treatment, although it can rarely achieve pCR. Neoadjuvant endocrine therapy (NET) has been mostly studied for post-menopausal breast cancer patients in comparison with tamoxifen and aromatase inhibitors [13‐15]. Semiglazov et al. compared NET and chemotherapy for post-menopausal breast cancer patients and found a similar clinical response between the two therapies [16]. The STAGE study was conducted for pre-menopausal breast cancer patients and compared gonadotropin-releasing hormone agonist (GnRHa) with tamoxifen and GnRHa with an aromatase inhibitor [17]. However, no clinical trial has yet compared NCT with NET for pre-menopausal patients.

Therefore, we hypothesised the non-inferior response of NET compared with NCT and reported a phase III clinical trial to compare the response between NCT and NET in pre-menopausal women with ER-positive, HER2-negative, and lymph node-positive breast cancer.

A total of 187 patients from 7 participating centres were included and randomised. Seven patients in the NCT group and 5 patients in the NET group withdraw their consent. One patient in the NCT group was randomised but did not receive treatment. Therefore, a total 174 patients completed the scheduled treatment and were finally analysed (87 patients received NCT and 87 patients received NET; Fig. 1). Four patients in the NET group did not receive the surgery, of whom 3 patients showed PR and 1 patient showed stable disease (SD). The data of these 4 patients were excluded from the post-surgery data (pCR and rate of BCS) analysis (Fig. 1). Baseline characteristics were generally balanced between the treatment groups (Table 1). The median age was 42 years (range, 27–54). All patients were pre-menopausal, and 63% of patients had a normal body mass index. Sixty-nine percent of patients had a clinical T2 breast cancer, and 88.5% had a clinical N1 stage breast cancer. Ninety-four percent of patients had G1/2 breast cancer, and few patients (less than 5%) had poorly differentiated (G3) tumour. The mean Ki-67 expression was not different between the two groups (26.3 for NCT vs. 26.7 for NET, p = 0.891). Forty-eight percent of NCT patients and 61% of NET patients were expected to undergo mastectomy prior to the start of treatment (p = 0.142).

Tumour size decreased over 24 weeks in both treatment groups (p > 0.05). The clinical response rate measured using MRI was 83.7% (72 of 87 patients) with NCT and 52.9% (45 of 87 patients) with NET (estimated difference 30.8%, 95% CI 17.6–44.0, p < 0.001). On callipers, the clinical response rate was 83.9% (73 of 87 patients) with NCT and 71.3% (62 of 87 patients) with NET (estimated difference 12.6%, 95% CI 0.4–24.9, p = 0.046) (Table 2).

The conversion rate of breast-conserving surgery (BCS) with neoadjuvant treatment in patients who planned to have total mastectomy was 13.8% in the NCT group and 11.5% in the NET group, which was not significantly different (p = 0.531). Three NCT patients (3.4%) and one NET patient (1.2%) showed pCR (ypT0ypN0M0) (p < 0.005). Twelve NCT patients (13.8%) and 4 NET patients (4.9%) had complete lymph node response (p < 0.005). However, Ki-67 expression change during the treatment period was not significantly different between the two groups (p = 0.114; − 13.56 for NCT and − 7.49 for NET).

Nineteen NCT patients (21.8%) had treatment-related grade 3 or worse adverse events, with no adverse effects in the NET group.

Exploratory subgroup analysis was performed. There was a difference of tumour response according to Ki-67 expression. For patients with a low Ki-67 expression (≤ 20%), the clinical response rate was 83.7% (41/49) in NCT, 60.4% (29/48) in NET with MRI, and 69.4% (34/49) in NCT, 83.3% (40/48) in NET with calliper. Among the high Ki-67 expression group (> 20%), the response rate was 81.1% (29/36) in NCT, 40.5% (15/37) in NET by MRI, and 77.8% (28 /36) in NCT, 78.4% (29/37) in NET by calliper (Table 3, Fig. 2).

During 24 weeks of neoadjuvant treatment, in pre-menopausal, hormone receptor-positive, HER2-negative, and lymph node-positive breast cancer patients, NCT achieved a significantly better clinical response rate than NET. The difference in response was even higher in patients with a highly proliferating tumour (Ki-67 expression > 20). Neoadjuvant therapy was comparatively more effective in patients with low Ki67 (low, 60.4%, vs. high, 40.5%, by MRI) while chemotherapy was equally effective irrespectively of Ki67 (low, 81.1%, vs. high, 83.7%, by MRI).

This study is the first to compare NCT with NET (tamoxifen plus ovarian function suppression) in pre-menopausal breast cancer only. This study is also unique because the patients were all ER-positive/HER2-negative and lymph node-positive.

The primary finding of this study was that the response to NET was inferior to NCT in pre-menopausal ER-positive/HER2-negative subtype breast cancer. The pCR rate was also higher in the NCT group than in the NET group. However, this does not imply that NCT patients will have a better long-term survival outcome than NET patients. First, the changes in Ki-67 expression were not different between the two groups. In addition, NET patients could receive adjuvant chemotherapy in most cases especially when the response was poor (data not shown). Long-term follow-up of the two patient groups is warranted.

Considering pCR and/or clinical response is not a reliable surrogate endpoint for survival, and the most important role of neoadjuvant systemic therapy in ER-positive/HER2-negative breast cancer is expanding the pool of potential BCS candidates by downstaging tumours and permitting BCS in patients who would otherwise require mastectomy [18]. In our study, although NCT led to a better clinical response, the BCS rate was not different between the two groups. Therefore, for the purpose of enabling BCS for mastectomy candidates, NCT might not be a better option than NET.

A potential advantage of neoadjuvant therapy is avoidance of axillary lymph node dissection in patients who have had negative conversion of tumour-positive lymph nodes. NCT downstages axillary nodes in 20 to 40% of patients, and these rates are even higher (> 50%) in HER2-positive patients given anti-HER2 therapy [19‐21]. Two recent prospective trials (ACOSOG Z1071 and SENTINA) reported that the false-negative rate was reasonably low when a dual tracer was used and 3 or more sentinel nodes were harvested [22, 23]. Kang et al. showed that in breast cancer patients who had axillary lymph node conversion from clinically positive to negative following NCT, a sentinel lymph node biopsy-guided axillary operation had similar rates of axillary and distant recurrence with axillary lymph node dissection without sentinel node biopsy [24]. Our study showed that for the purpose of avoiding lymph node dissection, NCT can be better than NET because the CR in lymph nodes was significantly higher in the NCT group than in the NET group (13.8% vs. 4.6%). However, further study is needed as to whether complete axillary lymph node dissection has survival benefit and should be a mandatory procedure for patients who have metastasis in a limited number of axillary lymph nodes after NCT or NET.

Most NET studies in breast cancer have focused on post-menopausal women [12, 13, 25, 26]. Few data are available on NET in pre-menopausal women. Some studies have shown that NET could be effective in well-selected pre-menopausal patients [17, 27]. GEICAM has reported randomised phase II results of chemotherapy versus exemestane in pre- and post-menopausal women [28]. Although the sample size was small, the response rate was higher for chemotherapy than for endocrine therapy in pre-menopausal patients, which was consistent with our study. The two therapies were comparable in post-menopausal women and those with a low baseline Ki-67.

The optimal duration of NET is not well defined. Most previous NET studies applied from 3 to 6 months of therapy. In a study by Llombart-Cussac et al. [26], 37% of patients achieved the maximal response beyond 6 months. Carpenter et al. showed that median time to achieve BCS (in those who responded) was 7.5 months [29]. Notably, 62% of panellists at St Gallen 2013 were in favour of continuing NET until a maximal response [30]. In our study, the patients received 6 months of NET. If we had treated to the maximal response for these patients, the response might have been better.

For ER-positive/HER2-negative subtype breast cancer, objective response determination is an issue. pCR, which is the only definite response criteria in neoadjuvant therapy, is highly uncommon; furthermore, it is not a surrogate endpoint for survival in this subtype [31]. We evaluated the clinical response using MRI and callipers. Even though MRI is a useful method for measuring residual tumour size after neoadjuvant treatment, it is not well correlated with pathologic size in ER-positive, HER2-negative tumours [32]. In our study, there was also a large discrepancy between MRI size and pathologic tumour size in the NCT group (data not shown). The difference in response pattern in the tumour between endocrine therapy and chemotherapy makes an objective response determination and comparison difficult and somewhat imbalanced.

There are some limitations in this study. First, the small sample size which was not satisfied with pre-defined number was included. Second, clinical response or pCR is not a predictor of prognosis in ER-positive and Her2-negative subtypes. This data cannot demonstrate the survival benefit of NCT compared with NET. Third, we did not include the arm of aromatase inhibitor as a NET treatment. In a study that compared neoadjuvant GnRH analogue plus tamoxifen and GnRH analogue plus anastrozole in pre-menopausal women, the clinical response was better in the anastrozole group [17]. Because we did not consider an aromatase inhibitor, a question remains concerning the response comparison between chemotherapy versus ovarian function suppression with aromatase inhibitor in pre-menopausal women. Lastly, multigene assays were not evaluated.

Although, in our result for the primary endpoint, NET showed a worse clinical response than NCT, NET cannot be excluded as a clinical option, because (1) the BCS rate was comparable, (2) no difference was observed in Ki-67 change with the therapies, and (3) the adverse event was much smaller in the NET group. This finding indicates that we can recommend NET in clinical practice in ER-positive/HER2-negative breast cancer with low Ki-67 expression, intermediate/low grade tumour, clinically lymph node-negative patients, and for patients with large tumours where BCS is not feasible or equivocal. It is recommended to perform a multigene assay (e.g. Oncotype DX, Mammprint) prior to the start of NET to confirm whether chemotherapy is needed for the individual patient.

In pre-menopausal breast cancer patients, 24 weeks of NCT showed a better clinical response than NET except for in a subgroup with low Ki-67 or low grade cancer. However, the breast conservation rate and the Ki-67 change were not different. There may be an issue concerning accurate size measurements with MRI in both groups. Clinicians should be aware of the potential benefits and disadvantages of both treatments. Use of NET in pre-menopausal women requires further discussion.