This study was designed as an analysis of retrospective data in a single-facility, the Kurume University School of Medicine (Kurume, Japan). This observational study was approved by the Ethics Committee of Kurume University and all pts provided written informed consent for the treatment and publication of the data.

Women under the age of 75 years with previously untreated clinical stage III breast cancer, who were diagnosed by mammography, ultrasonography, breast magnetic resonance imaging (MRI), core needle biopsy and positron emission tomography-computed tomography (PET/CT), were eligible for this study. Each pt had a locally advanced breast cancer with ALN involvement. The eligibility criteria also included adequate performance status [Eastern Cooperative Oncology Group (ECOG) performance 0–1], adequate hematology, renal and liver function and an ejection fraction ≥60%, confirmed by ultrasonic cardiography. Pts who had an otherwise adverse medical history, another malignancy, contralateral breast cancer and/or a severe systemic condition were excluded.

Three different NAC regimens were used: ED (60 mg/m² epirubicin and 60 mg/m² docetaxel), FEC (500 mg/m² fluorouracil, 75–100 mg/m² epirubicin, and 500 mg/m² cyclophosphamide) and EC (60 mg/m² epirubicin and 600 mg/m² cyclophosphamide). For the pts who underwent EC and most of the pts who underwent FEC, a further four cycles of D (docetaxel 70 mg/m²) were then administered. Each chemotherapy regimen was administered every three weeks for four cycles; however, this interval was prolonged by at least one week if the pt did not recover from the adverse effects. Subsequent to the completion of the four cycles of NAC, we evaluated the clinical responses and performed surgery within 2–3 weeks. The surgical methods included Patey’s procedure in three pts, mastectomy in 16 pts and lumpectomy in three pts. All pts underwent level I+II ALN dissection. In addition, the three pts who received Patey’s procedure underwent level III lymph node dissection.

Following surgery, extensional adjuvant chemotherapy was administered to 13/22 pts (59%) who had numerous ALN metastases (≥4 positive nodes) and/or poor pathological responses to NAC. Each regimen of extensional chemotherapy was selected by the clinician. Nine of the 22 pts (41%), who had a positive HER2 status, were treated with adjuvant trastuzumab (initially 8 mg/kg, followed by 6 mg/kg) for 12 months. Subsequent to the completion of adjuvant chemotherapy, whole breast irradiation of 50 Gy was performed for the pts who underwent a lumpectomy, while chest wall and regional lymph node irradiation of 50–60 Gy was performed for the majority of the pts. In addition, postmenopausal pts were treated with aromatase inhibitors for ≥5 years, whereas premenopausal pts were given tamoxifen until menopause, prior to being switched to aromatase inhibitors.

To compare the efficacy of NAC using epirubicin and/or docetaxel in stage III breast cancer, we investigated 31 pts with stage III breast cancer who were treated with adjuvant chemotherapy between 1996 and 2005.

The clinical response was assessed based on a physical examination, mammography, ultrasonography, MRI and CT according to the Response Evaluation Criteria In Solid Tumors (RECIST) version 1.1 criteria (8). A clinically complete response (cCR) was defined as the disappearance of all known lesions; a clinically partial response (cPR) was defined as a ≥30% reduction in the sum of the longest diameter (LD) of the primary lesion; progressive disease (PD) was defined as a ≥20% increase in the sum of the LD of the primary lesion and stable disease (SD) was defined as neither sufficient shrinkage to qualify for cPR nor sufficient increase to qualify for PD. The efficacy of NAC was examined in the surgical specimens, while the Ki-67 labeling index was examined in the pre-treatment biopsy specimens. The pathological response was assessed based on the histological changes in the invasive area by the Japanese Breast Cancer Society criteria (9). A pCR was defined as no residual invasive cancer in the breast tissue, regardless of the ALN status, while the grade 0 response indicated no cancerous degeneration. A grade 2 response was defined as ≥2/3 cancerous degeneration or a small amount of invasive cancer in the specimen, while a grade 1 response was defined as <2/3 cancerous degeneration in the specimen. The number of involved ALNs was confirmed in the dissected ALN specimen by the pathologist. In addition, toxicities of the NAC were graded by the ECOG common toxicity criteria.

OS and DFS were considered from the onset of NAC and from the day of breast surgery, respectively. The statistical analysis was conducted using JMP version 9.0 statistical software (SAS Institute, Inc., Cary, NC, USA). The correlation analysis was performed to compare two variables, including categorical and continuous variables. Univariate survival analyses to investigate predictive factors for OS and DFS were performed with a Cox proportional hazard model. The significant factors (P<0.05) were entered into a Cox multivariate regression model to analyze the potential simultaneous effects of the predictors of OS and DFS identified by univariate analyses. The survival analysis of the most significant factor was performed with the Kaplan-Meier method, and comparisons between the survival curves were performed with the log-rank test. Pt follow-up was performed in our hospital, from the beginning of chemotherapy either until mortality or the last visit of the pt.

A total of 22 women with stage III breast cancer underwent NAC between January 2005 and May 2011. The median age was 55 years (range, 33–72 years) and the median follow-up period was 66 months (range, 9.3–90.0 months). The clinicopathological characteristics of the pts are shown in Table I. The observed intrinsic subtypes were as follows: six luminal, nine HER2-positive and seven triple-negative types. The pts’ histological types showed invasive ductal carcinoma in 19 pts, mucinous carcinoma in two pts and invasive lobular carcinoma in one pt. There were no significant differences between the pts who underwent the FEC regimen and those who underwent a non-FEC regimen (ED, EC followed by D).

Clinical efficacy of NAC was observed in 64% (cCR, 5%; cPR, 59%) of all pts (Table I). The remaining 36% had SD (27%) and PD (9%). A younger age (P=0.010) and non-FEC regimen (P=0.005) were correlated with poor clinical responses. Pathological efficacy was observed in 36% (pCR, 18%; grade 2, 18%) of all pts (Table I). The remaining 64% showed grade 1 (50%) and grade 0 (14%) responses. There was no significant factor predicting whether the pathological response was likely to be good or poor.

Based on the ECOG common toxicity criteria, the most common toxicities were grade 1 and 2 neutropenia (n=12, Table II). One pt required an admission for grade 4 neutropenia, and six pts were classified as grade 3. The other grade 3 toxicities are shown in Table II. Additional common toxicities were anorexia (n=7) and fever (n=6). The majority of the pts experienced alopecia.

The median survival time was 66 months, and 11/22 pts (50%) succumbed to refractory breast cancer. There were 12 distant metastases (five brain metastases, five lung metastases, one bone metastasis and one subclavian lymph nodes metastasis) and two local recurrences. pCR was not a prognostic factor for the success of NAC in this study. Univariate analyses showed that the triple-negative type, positive status of estrogen receptor and the number of involved ALNs were correlated with DFS, while the triple-negative type, Ki-67 labeling index (%), pathological tumor size (cm) and the number of involved ALNs were correlated with OS (Table III). Multivariate analyses showed that the number of involved ALNs [hazard ratio (HR), 1.079; 95% confidence interval (CI), 1.011–1.155; P= 0.023] was correlated with DFS, while the Ki-67 labeling index (HR, 1.109; 95% CI, 1.004–1.265; P=0.042) and the number of involved ALNs (HR, 1.087; 95% CI, 1.012–1.180; P=0.023) were correlated with OS (Table III).

We assessed the outcome of four pts with observed pCR subsequent to surgery (Table IV). The five-year survival rate of these pts was 75% (3/4 pts) and 2/4 pts (50%) suffered from a relapse of the breast cancer. These two pts relapsed with brain metastasis, having had a short DFS (2.9 and 7.9 months). For the four pts with observed pCR subsequent to surgery, initial staging, intrinsic subtype and Ki-67 labeling index were suggested as prognostic indicators.

Comparisons of the characteristics for the pts treated with NAC and those treated with adjuvant therapy (AT) are shown in Table V. The pts treated with AT included a higher proportion of stage IIIA disease and a smaller proportion of stage IIIC disease (P=0.040). The majority of the pts treated with AT underwent four cycles of anthracycline- and/or taxane-based regimens using doxorubicin, epirubicin and/or paclitaxel. DFS and OS curves for the NAC and AT groups are shown in Fig. 1. In our hospital, epirubicin and/or docetaxel-based NAC was found not contribute to enhanced survival in stage III breast cancer.

We compared DFS and OS curves between pts with NIN <4 and pts with NIN ≥4, and a greater number of NIN (≥4) was significantly correlated with poor prognoses. Tausch et al (10) observed that an increased number of involved nodes (NIN) and an increased ratio of involved to removed nodes (LNR) were significantly correlated with worse DFS and OS in univariate and multivariate analyses (P<0.001). We compared DFS and OS curves between pts with NIN <4 and pts with NIN ≥4 (Fig. 2). A high number of NINs (≥4) was a significant prognostic indicator correlated with DFS and OS (P=0.025 and P=0.024, respectively). In the current study, the Ki-67 index was indicated to be as an independent prognostic factor for OS (HR, 1.109; 95% CI, 1.004–1.265; P=0.042). However, achieving pCR subsequent to NAC was not associated with the Ki-67 labeling index (P=0.654, Wilcoxon test). Since the median percentage of Ki-67 was 21.3% (range, 7.1–55.2%), we compared the DFS and OS curves with a cut-off value of Ki-67 at 20% (Fig. 3). A high percentage of Ki-67 (≥20%) was suggested as a prognostic indicator correlated with OS (P=0.057), while there was no significant difference between the DFS curves (P=0.183).