Introduction

Preoperative systemic therapy is now an established part of the management of large, potentially operable, and locally advanced breast cancers as it could achieve down-staging of the primary tumor, thereby allowing breast-conserving surgery [1–4]. Neoadjuvant therapy has been shown to be equivalent to adjuvant therapy in terms of survival and relapse-free survival [5]. The achievement of pathological complete response (pCR) including nodal involvement could be a surrogate marker to predict long-term outcome and is currently the main endpoint reported by neoadjuvant chemotherapy protocols [6–9]. Clinical trials in this setting therefore require fewer patients and can be completed more rapidly. However, not all patients achieving pCR will be cured and other surrogate markers must therefore be identified [10, 11].

Cyclooxygenase-2 (COX2), which mediates the production of prostaglandins and thromboxanes from arachidonic acid, is induced in inflammation and in cancer development and progression [12–15]. COX-2 is frequently over-expressed in breast cancer, and associated with unfavorable outcomes. Over-expression of COX2 in breast cancer cell lines increases cell migration and invasion [16]. Celecoxib is a COX-2 inhibitor with anti-angiogenic and pro-apoptotic activity. In breast cancer cells celecoxib analogs were shown to be potent inhibitors of phospho-Akt-signaling pathways and to induce apoptosis [17]. In animal models, treatment with selective COX-2 inhibitors reduced the formation, growth, microvasculature, and metastases of tumors [13].

Trastuzumab, a recombinant humanized monoclonal antibody targeting the HER2 receptor, is clinically active in HER2 metastatic breast cancer when used alone or in combination with chemotherapy [18]. Large trials have established a survival benefit for trastuzumab in the adjuvant setting [19, 20]. A limited number of phase II trials have evaluated trastuzumab in combination with various chemotherapeutic agents in the neoadjuvant setting [2, 21–24]. In these studies, the pCR rates obtained ranged from 12 to 47% [23]. Three randomized clinical trials have studied the combination of trastuzumab with a sequential regimen of anthracyclines-based chemotherapy then paclitaxel or docetaxel. In the M.D. Anderson group trial, the pCR rate was 25% in the chemotherapy-only arm versus 67% in the trastuzumab arm [25, 26]. In the NOAH trial reported by Gianni et al. [27], the pCR rates were, respectively, 23% versus 43% (P = 0.002) without or with trastuzumab. In the GeparQuattro, including 453 HER2 positive patients, pCR rates were, respectively, 20 and 41.3% without or with trastuzumab [28].

The aim of the present randomized phase II study was to assess anti-tumor activity of sequential epirubicin/cyclophosphamide followed by docetaxel with the randomized addition of celecoxib in HER2 negative tumor patients or trastuzumab in HER2 positive tumor patients in terms of pathological complete response.

Patients and methods

Patients and chemotherapy regimen

The randomized phase II Remagus 02 trial included patients from four different institutions. Eligibility criteria for the study were female patients over the age of 18 and under the age of 65 with histologically proven non-metastatic invasive breast carcinoma (Stage II and III), not amenable to breast-conserving surgery (diameter > 3 cm, central) or with risk factors making neoadjuvant chemotherapy the preferred treatment (i.e., N2–N3, rapid growth rate). Inflammatory breast cancers were allowed. Availability of frozen tumor tissue for molecular studies was mandatory. Eligible patients had no history of previous malignancy other than treated in situ carcinoma of the cervix or non-melanoma skin cancer, no bilateral breast cancer, and no distant metastasis. The routine diagnostic work-up included bilateral mammography and ultrasound, breast MRI, tumor biopsy with frozen sample, chest X-rays, abdominal ultrasound, bone scan, blood sampling, and clinical examination. Patients were eligible is Left Ventricular Ejection Fraction (LVEF) assessed by MUGA scan or ultrasound was >50%. HER2 status was considered positive if the immunohistochemistry result was 3+ according to current ASCO criteria. In doubtful cases (2+), a FISH analysis was performed. HER2 status was centrally reviewed for all patients and HER2 positivity confirmed by FISH in all cases. The cut-off used to define hormone receptor positivity was 10% of stained cells.

All patients were planned to receive epirubicin (75 mg/m2)–cyclophosphamide (750 mg/m2) intravenously every 3 weeks for four cycles followed by docetaxel (100 mg/m2) every 3 weeks for four cycles with or without trastuzumab (8 mg/kg at first infusion then 6 mg/kg) every 3 weeks, according to randomization, for HER2-positive tumor patients and with or without celecoxib 400 mg BID orally for HER2-negative tumor patients. Surgery (lumpectomy or modified radical mastectomy associated to axillary clearance) was performed 21–45 days after cycle 8, according to initial and post-chemotherapy assessment (MRI evaluation was not mandatory). Conservative surgery was acceptable in the absence of multifocality provided clear margins were obtained. Surgery was followed by local and regional radiotherapy when indicated (in all cases of breast conserving surgery and/or in case of nodal involvement). The administration of adjuvant trastuzumab for a total of 18 three-weekly infusions was mandatory for all HER2-positive cancer patients, to be started after surgery for those randomized to control. Indeed, all patients but eight received it. All patients with hormone receptor-positive tumors received adjuvant tamoxifen or aromatase inhibitors according to their menopausal status and the current guidelines. Adjuvant chemotherapy according to centers preferences in patients with residual axillary nodal involvement (pN+) could be delivered based on 5-fluorouracile and vinorelbine combination, concomitantly or not with radiotherapy (four cycles).

Clinical evaluation of tumor response was performed before each subsequent cycle, while mammogram, ultrasound and eventually MRI were repeated before cycle 5 and immediately before surgery.

Pathologic tumor response (primary tumor and axillary nodal status) was evaluated according to Chevallier criteria; a pCR was defined by Chevallier grade 1 or 2 criteria [29]: no residual invasive carcinoma in the breast and axillary nodes. All pathological responses were reviewed by the panel of pathologists blinded to the center and the treatment arm.

Safety evaluations

Safety evaluation procedures included clinical evaluations and blood tests initially and before each subsequent chemotherapy cycle. Patients with HER2-positive tumors had LVEF evaluation through the same procedure before initiation of treatment, before cycles 4 and 6, plus before and after cycle 8 in the group receiving trastuzumab and subsequently every 3 months during adjuvant trastuzumab.

Statistical methods

The main criterion was pathologic tumor response according to Chevallier grade 1 or 2 [29]. A two-stage modified Fleming approach was proposed for each treatment groups, allowing stopping the study only in case of insufficient antitumor efficacy. In both strata, randomization was stratified on treating center.

For each group in Stratum A (HER2 negative), the percentage defined in null hypothesis was 15% and the expected percentage was 25% (a level of pathologic tumor response inferior or equal to 15% was considered as unacceptable leading to its discontinuation, whereas a level significantly above 15% as acceptable); inclusion of 110 patients by arm was needed to access the acceptable rate of 25% with a power of 90% and with α equal to 9%. Concerning the decisions rules, if 22 pCR or more were observed, evidence was provided that the strategy has to be continued. For each group in Stratum B (HER2 positive), the percentage defined in null hypothesis was 15% and the expected percentage was 30% (a level of pathologic tumor response inferior or equal to 15% was considered as unacceptable, whereas a level significantly above 15% as acceptable); inclusion of 60 patients by arm was needed to access the acceptable rate of 30% with a power of 90% and with α equal to 7%. Concerning the decisions rules, if 14 pCR or more were observed, evidence was provided that the strategy has to be continued.

An intermediate analysis was to be conducted when half of the planned patients had been assessed in the groups in order to stop study arm(s) in case of insufficient efficacy. The study was not stopped at this stage (results not showed) and was conducted up to the end.

Differences between categorical variables were analyzed by χ2 tests or Fisher’s exact test. No comparisons tests were made between the groups within each stratum: this study was a stratified randomized phase II conducted to evaluate the primary efficacy of each therapeutic strategy and not to compare the groups between them because of a lack of power in such a case.

Secondary end-points were clinical responses at cycles 4 and 8, tolerance, disease-free survival, biomarkers studies, and analysis of markers predictive for pathological complete response in each group. Translational studies will be reported elsewhere. Survival data are not yet available. We therefore report here mature data on the main end-point (pathological complete response) together with data on tolerance and prediction of response in both strata.

Analyses were performed using R software 2.5.0 version.

Results

From May 2004 to October 2007, 341 patients were randomized in this trial. One patient was excluded for a major violation of the protocol and 340 were analyzed in intent to treat (Fig. 1, flow chart). Patients’ characteristics are shown in Table 1. Within each stratum, groups were overall well balanced. Half of patients had T2 while half had T3–T4 tumors. It is important to highlight that tumors were large with 14% of T4 including T4d. None of patients with HER2 positive tumor had grade 1 carcinoma but 59% had expression of hormonal receptors.

Fig. 1
figure 1

Flow chart of the REMAGUS 2 trial

Full size image
Table 1 Patients characteristics
Full size table

Compliance to treatment was good: only four patients stopped for toxicity (1.2%), 10 for tumor progression (3%), and 10 (3%) for various reasons (patient’s decision, protocol violation, etc.). Altogether 212/220 patients in Stratum A were operated (104/108 in group A1, 108/112 in group A2), 214 had axillary sampling (106/112 in group A2); 119/120 patients in Stratum B were operated and all had axillary clearance. Therefore, 329 patients were evaluable for pathological response. For both strata (Table 2), the first step analysis at 110 and 60 patients in total indicated that the second step was to be conducted; the second step analysis after the recruitment of twofold the number of patients was performed until 340 eligible patients had been analyzed. In the HER2 negative group complete pathological rates were 11% without celecoxib versus 13% with celecoxib. According to sequential analysis and the decisions rules, at the second stage, as the number of pCR was below 21, the null hypothesis of a pCR rate with celecoxib ≤15% could not be rejected. There was no apparent increase in response rate with the addition of celecoxib to sequential neoadjuvant chemotherapy.

Table 2 Pathological response according to Chevallier criteria
Full size table

For the group of patients with HER2 positive tumor, at the second step, the number of pCR was superior to 14, the hypothesis of an increased efficacy of the combined arm trastuzumab + chemotherapy with pCR rate of 26% (95% confidence interval 16–37%) was confirmed providing evidence that this was a strategy to be continued. pCR rate was 19% (95% confidence interval 9–29%) in the chemotherapy alone arm. Grade 1 response (residual in situ carcinoma of the breast) rate was 7% (4/58) in the trastuzumab arm and 19% (12/62) in the arm without trastuzumab.

There were 78 triple negative breast cancers (23%) in this study (Table 3). This subgroup of patients experienced the highest pCR rate, 29.5% (95% confidence interval 19.7–40.9%) compared to 11.4% in all the other subgroups or to 2% in the HER2 negative, hormonal receptors positive subgroup.

Table 3 Pathological response (Grade 1 and 2) according to hormonal receptor status and according to treatment arm in the HER2 positive tumor stratum
Full size table

Clinical complete response rates between the four arms were: 26.9, 23.2, and 22.4%, respectively, in the arms without trastuzumab and 33.9% in the arm with trastuzumab and clinical response rates (more than 50% tumor size reduction) were 77.8, 72.3, 86.2, and 77.4%, respectively. In this population with large tumor, not eligible for primary breast conservative surgery 50% (170/340) of the patients had lumpectomy with axillary dissection after neoadjuvant chemotherapy avoiding mastectomy. There was no suggestion of difference according to HER2 status and or therapeutic arm (Table 1).

There was no treatment-related death and no Sudden Unexpected Serious Adverse Event. Use of celecoxib in randomization in Stratum 1 (HER2 negative tumor patients) for a new patient was suspended by AFFSAPS (National French Health authorities) from December 2004 to September 2005 as part of the European assessment of cardiac side effects of coxib, and was thereafter authorized with revision of the informed consent form. Tolerance according to treatment arm is shown in Table 4.

Table 4 Treatment-related toxicity
Full size table

Five (4.6%) HER2 negative tumor patients without celecoxib had five episodes of skin rash compared to seven receiving celecoxib who had nine episodes of skin rash.

There was no obvious difference for other toxicities as shown in Table 2. Reduction of LEVF between 10 and 20% but with an absolute value >50% was observed in 30 (16.3%) of the evaluated patients. One patient had a reduction of LVEF below 50% (45%) in the arm receiving trastuzumab (B2).

Discussion

With pathological complete response being the primary endpoint, the addition of celecoxib to standard sequential epirubicin/cyclophosphamide—docetaxel in HER2 negative patients does not differ from the null hypothesis (≤15% pCR rate). Indeed the group of HER2 positive tumor patients with trastuzumab added to docetaxel with 26% (95% confidence interval 16–37%) pCR, was the only one for which the null hypothesis could be eliminated. The pCR rate for the whole population is in the low range of reported values with similar sequential anthracyclines cyclophosphamide then taxanes chemotherapy regimen [30]. Among the possible explanations there is the fact that all cases were reviewed by the panel of pathologists with very strict criteria of pCR definition, using Chevallier classification including the nodal status. Also, the median tumor size was large and none of the patients was eligible for a breast conservative surgery before chemotherapy.

Chemotherapy with celecoxib was not able to demonstrate a promising pCR rate in our study. There are few results of anti-COX2 treatment in breast cancer. Further explorations with hormonal treatment have been reported without significant improvement in overall or progression free survival [31–33]. One phase II trial in 42 patients combined chemotherapy (capecitabine) and celecoxib in metastatic breast cancer patients showing an increased time to progression in case of COX2 over-expression [34]. Our study is the first randomized trial of combination of chemotherapy and COX2 inhibitor for localized breast cancer. However, the lower pCR rates in this study may underestimate the benefit of celecoxib. Tolerance was good with no cardiac event and a small trend for more frequent skin rashes. While this first analysis does not suggest increase pCR, however, longer follow-up will be necessary to ascertain reduction in the metastatic rate. As suggested by Lucci et al., COX2 produced in primary breast cancer cells may be vital to the initial development of bone marrow micrometastasis that may subsequently lead to osteolytic bone metastases in patients with breast cancer, and COX2 inhibitors may be useful in halting this process [16].

The pCR rate in patients increased to 26% (95% confidence interval 16–37%) in patients with HER2 positive tumor receiving chemotherapy and trastuzumab [23]. Higher response rates from 41 to 60% have been reported in the MD Anderson, NOAH and GeparQuattro trials [26–28]. In these three studies, one should notice that trastuzumab was given upfront, for a longer period of time and was administered concurrently with an anthracyclines-based chemotherapy regimen without so far overrated cardiac consequences. However, caution must be advised regarding the concomitant use of trastuzumab with anthracyclines [19]. In our study, one should pay attention to the absence of cardiac event. Our results with a low response rate despite trastuzumab addition substantiates a very early introduction of trastuzumab in the sequence of the neoadjuvant chemotherapy and of the possible higher efficacy of the combination with anthracyclines, targeting also topoisomerase II which can be more often amplified in HER2 over-expressing tumor .

Interestingly, grade 2 response (residual ductal carcinoma in situ DCIS only) was reduced in the trastuzumab receiving arm, 6.5% versus 12% without trastuzumab in the subgroup of HER2 positive tumor patient. This has to be put in perspective with the description of higher rate of HER2 over-expression in DCIS around 60% and the possible impact of trastuzumab on DCIS; clinical trials by the MD Anderson and the NSABP are ongoing in this setting [35].

Response rate according to hormonal status in HER2 positive tumor patients did not show important variations. We observed a trend for a higher response rate with the addition of trastuzumab predominantly in the estrogen receptor negative subgroup of patients. This is in contradiction with Peinteiger et al. [36] reporting a higher pCR rate in the HR+ subgroup with the addition of trastuzumab in neoadjuvant setting. In our study, patients with triple negative breast cancer had exquisite sensitivity to the chemotherapy regimen with a pCR rate of 30% compared to 2% in the RH + HER negative subgroup. Sensitivity of ER negative breast to neoadjuvant chemotherapy has been previously reported [30, 37, 38]. Despite fewer available data with triple negative this phenotype appears to be a strong predictor of chemosensitivity [39, 40]. The relation of the phenotype with gene expression profiling, P53 mutational status as suggested with a dose dense regimen has been reported [41].

Although pCR is an accepted endpoint in neoadjuvant studies, it does not always correlate to an overall survival or disease-free survival benefit, e.g., NSABP 27 showed no improvement in OS/DFS with taxanes despite a doubling in pCR [30]. Those patients who do not achieve pCR may still have a good outcome. However, a recent report of Southern Italy Cooperative Oncology Group (SICOG) randomized trial 9908 shows that a significant increase in pCR rate after neoadjuvant chemotherapy for locally advanced breast cancer can be translated into RFS and OS improvement after median follow-up of 74 months [42].

Conclusions

There is no evidence of an increase in pCR with the addition of anti-COX2 treatment to neoadjuvant chemotherapy in HER2 negative breast cancer. There is an increase in pCR with the addition of trastuzumab to neoadjuvant CT in HER2 positive tumor bearing patients but with a lower rate than reported in the literature when trastuzumab is used concomitantly with anthracyclines. This could suggest that the interest of an earlier introduction of trastuzumab with or without anthracyclines in the neoadjuvant chemotherapy sequence to obtain a higher response rate. Longer follow-up is mandatory to prove if the increased pCR rate translates in gain in survival. Results of the expression microarrays study on all initial frozen biopsies are awaited in order to predict more accurately response to treatment.