The cohort includes a total of 263 cases of invasive breast cancer, diagnosed and surgically treated between January and December 1993 at the Department of Breast Cancer, Tianjin Medical University Cancer Institute and Hospital 27. Patients had not received prior radiotherapy or neoadjuvant therapies. All patients received conventional postoperative treatment, depending on the extent of the disease. Patients without axillary lymph node involvement were treated with surgery alone, while patients with axillary lymph node involvement received six courses of adjuvant chemotherapy with cyclophosphamide/methotrexate/fluorouracil regimen. Patients with positive nodes or tumor size ≥ 5 cm received postoperative radiation. The patients with ER⁺/PR⁺ tumor were treated for 2–5 years with tamoxifen.

The patient characteristics including age (median: 50.3 years), menopausal status, clinical stage (TNM classification defined by the International Union against Cancer, UICC, 2003) were assessed by the surgical pathologists. These data as well as the postoperative follow-up information including local (regional) recurrence and overall survival have been previously described 27. Briefly, among the 263 cases, 132 women were premenopausal, while 131 were postmenopausal. At the time of surgery, 68 cases (25.9%) were Grade I tumors, 119 (45.25%) cases were grade II tumors, and 76 cases (28.90%) were grade III tumors. We used the commonly used grading standard to assign the scores of histological grades of breast cancer. Briefly, the tumor grading combines nuclear grade, tubule formation and mitotic rate. Grade I is assigned to well-differentiated tumors. Grade II is assigned to moderately differentiated tumors. Grade III is assigned to poorly differentiated tumors. Among the 263 cases, 45 cases (17.1%) were classified of T1, 154 cases (58.6%) were T2, 59 cases (22.4%) were T3 and 5 (1.9%) cases were T4. A total of 128 cases (48.%) were lymph node negative, 57 cases (21.7%) were N1, 26 cases (9.9%) were N2, and 52 case (19.8%) were N3. Local recurrence happened in 8 cases (3%). Metastases were identified in 54 cases (20.5%), including 13 cases to the liver, 12 cases to the bone, 11 cases to the lung, 7 cases to ipsilateral or the contralateral suprascapular lymph nodes, 5 cases to the contralateral breast, 2 cases to the brain, 2 cases to the ovary, 1 case to the pericardium, and 1 case to the pleura. All patients, unless deceased, were followed up for at least 36 months, up to 173 months. The cases lost for follow-up (due to patients' death or other reasons) had been marked as censored data. Patient outcome was defined by the months of post surgery overall survival (OS). This study was performed under a protocol approved by the Tianjin Medical University Administrative Panel on Human Subjects in Medical Research and the Institutional Review Boards of Tianjin Cancer Research institution.

TMA were constructed with a Beecher Instrument Tissue Array (Beecher Instruments, Silver Spring, MD) as previously described 27. Briefly, archival paraffin blocks of invasive breast cancer cases were employed. The recipient blocks were cut into tissue cores of 0.6 mm in diameter. Two cores of each specimen, representing the tumor areas, were transferred to the donor blocks.

Consecutive 4 μm-thick sections were cut from the recipient blocks to a poly-L-lysine-coated slide for IHC analysis. IHC was performed using a method described previously 27. A modification of antigen retrieval (5 min high-power microwave followed by 10 min low-power microwave in phosphate-buffered saline (PBS, PH 7.0) was made to detect MMP-2, MMP-9, MMP-13, TIMP-1 and TIMP-2. The antibodies and the dilution factors were as follows: MMP-13 (1: 150 diluted Clone VIIIA2 from Lab Vision Corp.), ER (1:450 diluted clone ID5 from DAKO), PR (1:200 diluted clone IA6 from DAKO); Her2/neu (1:1000 diluted polyconal from DAKO), MMP-2 (1:150 diluted polyclonal from Lab Vision Corp), MMP-9 (1:150 diluted polyclonal from Lab Vision Corp), TIMP-1 (1:150 diluted polyclonal from Lab Vision Corp), and TIMP-2 (1:150 diluted polyclonal from Lab Vision Corp).

Human placenta tissues known to express high levels of MMP-2, MMP-9 TIMP-1 and TIMP-2 2829 were used as positive controls for these proteins,. Breast cancer cell lines MCF-7 and T47D were used as positive control for ER and PR, respectively. Ovarian cancer cell line SK-OV-3 was used as a positive control for Her-2/neu staining 30. PBS was used in the place of the primary antibodies in all negative controls of IHC.

IHC was scored independently using a semi-quantitative scoring system as described below by two pathologists (WC and YL), who were both blinded to patients' clinicopathologic parameters and outcomes. Discordant scores were re-evaluated by the investigators and the consensus scores were used for further analyses. Both the intensity and extent of IHC were assessed 27. The intensity of the immunostaining was defined by the negative and positive controls as four categories. No brown particle staining: 0; light brown particle in cytoplasm: 1; moderate brown particle: 2; and 3: dark brown particle. The percentage of positive cells, as the extent of immunostaining, was quantified under microscope and classified into four groups. 1: <25% positive cells; 2: 25% to 50% positive cells; 3, 51% to 75% positive cells and 4, >75% positive cells. The staining index (SI), the product of the intensity and the percentage of positive staining, was used to define high (SI ≥ 6) or low (SI < 6) expression of MMP-13, MMP-2, MMP-9, TIMP-1 and TIMP-2.

The criterion of Herceptest/Pathway system was followed to score Her-2/neu Briefly, cases with strong complete membranous staining in more than 10% of the tumor cells were considered strongly positive (+3). Cases with weak to moderate complete membranous staining in more than 10% of the tumor cells were considered moderately positive (+2) and were subsequently confirmed by fluorescence in situ hybridization. Cases with little or no membranous staining were considered negative (0 or +1) 31. The criteria for scoring of ER and PR are similar. Cases with strong/moderate complete nuclear staining in more than 15% of the tumor cells were considered positive, whereas cases with little or no nuclear staining were considered negative 2731.

Statistical analyses were performed using the SPSS software package 11.0 (SPSS, Inc. Chicago, IL, USA). The correlations between MMP-13 expression and clinicopathologic variables were analyzed using Pearson Chi-square analysis. The same method was used to test the associations of MMP-13 with ER, PR, and Her-2/neu, respectively. Kaplan-Meier curves were constructed for univariate and multivariate analyses using a Cox proportional hazards model to examine the potential prognostic variables on OS. No corrections for multiple comparisons were made. All of the statistical tests were two-sided. p values of less than 0.05 were considered statistically significant.

Distinct IHC staining of MMP-13 was obtained from all 263 cases. The tumor or fibroblast-specific staining was semi-quantitatively scored by the SI scales and assigned into high and low categories. As shown in Figure 1, MMP-13 protein expression was detected mainly in cytoplasm of in malignant cells and the peritumoral fibroblasts in 127 (48.29%) and 144 (54.75%) cases, respectively. One hundred and eleven out of 127 specimens with high levels of tumor expression of MMP-13 (87.40%) featured high levels of MMP-13 in peritumoral fibroblasts. Consistently, 106 out of 144 specimens with low levels of tumor expression of MMP-13 were accompanied by peritumoral fibroblasts that expressed MMP-13 at low levels (Table 1). Overall, the detection of cytoplasmic MMP-13 in tumor cells, correlated significantly with the cytoplasmic MMP-3 in the peritumoral fibroblasts (p < 0.0001).

In order to test the potential value of MMP-13 as a breast cancer biomarker, we performed Pearson Chi-square analysis to evaluate the correlation of MMP-13 expression with clinical and histopathological features (tumor size, tumor grade, and lymph node status) that are further stratified as shown in Table 2. p values of less than 0.05 was considered statistically significant. High levels of MMP-13 expression, both in cancer cells and in peritumoral fibroblasts, correlated with lymph node metastases (p < 0.001), but not with tumor size and histological grade. MMP-13 did not appear to be regulated by sexual steroid hormones, since it was not correlated with the menopausal status of the patients. Nor was it correlated with ER or PR. Interestingly, a strong correlation between MMP-13 and Her-2/neu nuclear staining was observed both in cancer cells (p = 0.015) and peritumoral fibroblasts (p < 0.001).

The mean follow-up of the 263 patients in our cohort is 92.1 months. Kaplan-Meier analyses showed a clear stratification of the overall survival (OS) between the patients with high levels of MMP-13 expression and the patients with low levels of MMP-13 expression (Table 3). High levels of MMP-13 expression both in the cancer cells (p = 0.0008) and the neighboring fibroblasts (p = 0.0001) negatively correlated with OS (Figure 3Aand 3B). Tumor expression of MMP-13 negatively correlated with patients' OS regardless of the Her-2/neu status. It was noted that MMP-13 detected in peritumoral fibroblasts correlated with the OS of the lymph node positive subgroup (p = 0.047) and with the Her-2/neu negative subgroups (p = 0.045) (Figure 3C–F). When the patients' survival was stratified by lymph node status, MMP-13 expression was negatively correlated with the OS of the lymph node positive subgroups (p = 0.0041), but not with the OS of the lymph node negative subgroups (p = 0.344).

In the Cox regression model, univariate survival analyses showed that tumor size, lymph node positivity (>N2), PR, Her-2/neu, TIMP-1 and MMP-13 were each associated with worse prognoses, while, multivariate survival analyses showed that lymph node positivity (>N2), tumor size, Her-2/neu, MMP-13 and TIMP-1 expression by cancer cells were independent adverse prognostic factors for overall survival (Table 4). Parallel analyses showed that ER and Tumor Grades were not correlated with OS.