Patients and Samples
Of the 247 patients who had undergone mastectomy or breast-conserving surgery for invasive ductal carcinoma of the breast at the National Defense Medical College (NDMC) Hospital between 1995 and 1999, 150 patients with ER-positive and/or progesterone-receptor (PgR)-positive localized breast carcinomas were selected based on immunohistochemical reevaluation of ER and PgR expression. Tissue microarray (TMA) blocks of the tumors from these 150 patients were constructed as previously described . Briefly, double tissue cores 2 mm in diameter were taken from each donor block, and these core specimens were transferred to a recipient block using a Tissue Microarrayer (Beecher Instruments, Silver Spring, MD, USA). The use of the tissue blocks was internally reviewed and approved by the NDMC Ethics Committee.
The 150 patients had been followed up for a median of 82 months (range, 1–151 months), during which time there were 30 relapses and 15 deaths. In most cases, the patients were prescribed adjuvant endocrine therapy (for example, tamoxifen, toremifene, fadrozole, or LHRH analogues). Forty-nine patients with large tumors and/or 4 or more lymph node metastases had received adjuvant chemotherapy (cyclophosphamide-epirubicin-5-fluorouracil (CEF), cyclophosphamide-adriamycin-5-fluorouracil (CAF), cyclophosphamide-methotrexate-5-fluorouracil (CMF), or oral fluoropyrimidines), and 12 patients with locally advanced breast cancer had received preoperative chemotherapy (for example, CAF or CEF). One hundred forty-eight patients were females and 2 were males. The clinical stage of the patients was determined based on the TNM classification according to general rules of the Japanese Breast Cancer Society . Clinicopathological data were obtained from the medical records and pathology reports, but ER, PgR and HER2 status were examined in our previous study .
Immunohistochemistry was performed on a TMA composed of 150 breast cancer tissue specimens. The antibodies used were mouse monoclonal anti-p53 antibody (DO-7, Dako, Glostrup, Denmark), mouse monoclonal anti-Ki67 antibody (MIB-1, Dako), mouse monoclonal anti-FOXA1 antibody (2D7, Abnova, Taipei, Taiwan), mouse monoclonal anti-GATA3 antibody (HG3-31, Santa Cruz, Santa Cruz, CA, USA) mouse monoclonal anti-CK5/6 antibody (D5/16 B4, Dako), mouse monoclonal anti-CK14 antibody (LL002, NeoMarkers, Fremont, CA, USA), mouse monoclonal anti-P-cadherin antibody (56C1, Novocastra, Newcastle, UK), and a mouse monoclonal anti-EGFR antibody included in an EGFR pharmDX kit (Dako).
Sections (4-um-thick) were cut from the formalin-fixed, paraffin-embedded TMA blocks. Antigens were retrieved by microwave heating for 30 min in 10 mM sodium citrate (pH 6.0) for CK5/6 and GATA3 or by autoclaving for 20 min in 10 mM Tris–HCl (pH 9.0) for Ki67, p53, CK14, FOXA1, and P-cadherin. To block endogenous peroxidase activity, the sections were treated for 5 min with 100% methanol containing 3% H2O2. Non-specific binding was blocked by incubation in 1% normal swine serum (Dako) in phosphate-buffered saline. The slides were incubated with primary antibodies at 4°C overnight and then reacted with a dextran polymer reagent combined with secondary antibodies and peroxidase (Envision Plus; Dako) for 30 min at room temperature. Specific antigen-antibody reactions were visualized with 0.2% diaminobenzidine tetrahydrochloride and hydrogen peroxide. Immunostaining for EGFR was performed in accordance with the package inserts of the EGFR pharmDX Kit. The sections were counterstained with Mayer’s hematoxylin.
Evaluation of immunohistochemistry
Although there is no universal cut-off value for Ki67 LI, Cheang et al. showed that, using the cases which were subtyped by gene expression profile, the best Ki67 LI cut-off value to distinguish luminal B from luminal A was 13% . Furthermore, similar to the 10% cut-off value was used in several reports [21, 24–28]. So, in this study, Ki67 LI greater than 10% was classified as high. The Ki67 LI was calculated as the percentage of positive tumor nuclei divided by the total number of tumor cells examined on the basis of a manual count of 500 or more cells under high power (400×).
For p53, FOXA1, and GATA3, cells with immunostaining in the nucleus were defined as positive, while for CK5/6, CK14, and P-cadherin, cells with immunostaining along the cellular periphery and/or in the cytoplasm were defined as positive. For p53, positive staining of fewer than 10% of the tumor cells was defined as negative tumor expression and staining of 10% or more of the tumor cells as positive tumor expression . For P-cadherin, membrane staining of fewer than 50% of the tumor cells was defined as negative tumor expression and staining of 50% or more of the tumor cells as positive tumor expression. P-cadherin positive tumors were further divided into “weakly” and “strongly” expressing tumors based on staining intensity. Finally, negative and weakly P-cadherin-staining tumors were classified as “low” and strongly P-cadherin-staining tumors as “high”. A tumor expression cutoff point of 10% of cells stained was used for GATA3, CK5/6, and CK14 and a cutoff of 70% of cells stained for FOXA1, regardless of staining intensity. An EGFR score of 0–3+ was assigned according to the manufacturer’s package insert, and scores of 1–3+ were classified as positive. CK5/6, CK14, and EGFR were considered basal phenotype markers.
ER and PgR were examined immunohistochemically as described in the previous study , using mouse monoclonal anti-human ER (clone 1D5, Dako) and mouse anti-human PgR (clone PgR636, DAKO). ER and PgR were defined as positive if the nuclear staining was seen in 10% or more of tumor cells. Hormone receptor positive was defined as at least one of ER or PgR positive, and hormone receptor negative was defined as ER and PgR negative. HER2 was evaluated by IHC using rabbit polyclonal anti-HER2 antibody (HercepTest kit, Dako) and FISH (in case of IHC 2+) using Path Vysion kit (Abbott Park, IL, USA). HER2 was defined as positive if the IHC score was “3+” according to the standard procedure, or gene amplification (HER2:CEP17 ratio > 2.0) was detected by FISH .
The immunohistochemistry results were evaluated independently by 2 observers (TK and KI), and cases with disparate scores were re-evaluated and discussed until a consensus was reached. Ki67-positive cells were counted and the labeling index calculated by TK alone.
Comparisons between groups were evaluated with the chi-squared test or Fisher’s exact test. Patient survival curves were drawn using the Kaplan-Meier method and analyzed by the log-rank test. The hazard ratios and corresponding 95% confidence intervals (CIs) were calculated with Cox’s proportional hazards model. Univariate and multivariate Cox’s proportional hazards models were used to explore the associations of variables with disease-free and overall survival. For all tests, differences at P < 0.05 were considered statistically significant. All analyses were performed using the software JMP 6.0 for Windows (SAS Institute Inc., Cary, NC, USA).