Purpose: In the present study, we investigated the expression and prognostic value of matrix metalloproteinase (MMP)-2 and MMP-9 in breast cancer as well as their relation to transcription factor activator protein (AP)-2 and HER2 oncogene. The role of invasion and metastasis-promoting MMPs and their potential regulators, AP-2 and HER2, is currently still unclear in breast cancer.
Experimental Design: MMP-2 and MMP-9 expressions were analyzed immunohistochemically in a large prospective series of 421 breast cancer patients diagnosed and treated between 1990 and 1995 at Kuopio University Hospital (Kuopio, Finland). The relation of MMP-2 and MMP-9 expressions to AP-2, HER2, clinicopathological data, and survival was investigated.
Results: Both MMP-2 and MMP-9 were expressed in the cytoplasm of malignant and stromal cells. High expression of MMPs in carcinoma cells was related to small tumors (T1, stage I), whereas positive stromal expression of MMPs was associated with aggressive factors. High expression of MMP-2 and MMP-9 in carcinoma cells, but not in stromal cells, was related to high AP-2 expression. Positive stromal MMP-2 expression was associated with HER2 overexpression in the whole patient group and in the node-negative patient subgroup. Positive stromal MMP-9 expression was related to HER2 overexpression in estrogen receptor (ER)-positive disease. In the univariate survival analysis, positive stromal MMP-9 predicted shorter recurrence-free survival (RFS; P = 0.0389) and breast cancer-related survival (BCRS; P = 0.0081) in ER+ disease, especially in the subgroup of ER+ tumors of ≤2 cm in diameter (T1; P = 0.0031 for RFS, and P = 0.0089 for BCRS). High MMP-9 expression in cancer cells predicted longer RFS (P = 0.0351) in the whole patient group. In the multivariate analysis of the whole patient group, the independent predictors of shorter RFS were reduced MMP-9 expression in carcinoma cells (P = 0.0248), HER2 overexpression (P = 0.0001), and advanced-stage disease (P = 0.0002). Shorter BCRS was predicted by advanced-stage disease (P < 0.0001).
Conclusions: Expression of MMP-2 and MMP-9 in breast cancer seems to be partly related to expression of AP-2 and HER2. Positive stromal MMP-9 expression predicts poor survival in the hormone-responsive small tumors, whereas MMP-9 expression in carcinoma cells favors survival. Evaluation of MMP-9 expression seems to add valuable information on breast cancer prognosis.
Matrix metalloproteinase (MMP)-2 and MMP-9 (gelatinase A and B) are zinc-dependent endopeptidases of a large MMP family (1 , 2) . MMP-2 and MMP-9 are related to tumor invasion and metastasis by their capacity for tissue remodeling via extracellular matrix as well as basement membrane degradation and induction of angiogenesis (1 , 2) . These gelatinases are secreted as zymogens and cleaved to the active form, and their function is tightly regulated by several different mechanisms (1 , 2) . In breast cancer, both gelatinases seem to be expressed in cancer tissue, although the results have not been consistent (3, 4, 5, 6) . To date, only a few studies have investigated the prognostic value of MMP-2 and MMP-9 in breast cancer (5 , 7, 8, 9, 10) .
Aberrant expressions of transcription factor activator protein (AP)-2 and HER2 oncogene are related to disease progression and increased invasive capacity in breast cancer (11, 12, 13) , which may be due in part to increased MMP activity (14 , 15) . Indeed, AP-2 has been shown to be critically responsible for MMP-2 transcription (16, 17, 18, 19, 20, 21, 22) and to enhance MMP-9 promoter activity (23 , 24) . In melanoma cells, inactivation of AP-2 leads to increased invasiveness via increased MMP-2 expression and activity (14) , whereas in breast cancer, the relationship between AP-2 and gelatinase expression has not been studied. The effect of HER2 overexpression on the invasion capacity of tumor cells is related, at least in part, to the up-regulation of MMP-2 and MMP-9 expression and proteolytic activity (15 , 25, 26, 27, 28) . Furthermore, c-ErbB ligands are able to up-regulate several MMPs (29 , 30) . To the best of our knowledge, this is the first study to investigate the relation of MMP-2 and MMP-9 to AP-2 and HER2 expression in a large prospective clinical breast cancer study.
PATIENTS AND METHODS
The present study is part of the Kuopio Breast Cancer Project, a prospective long-term clinical study involving 520 breast cancer patients (31, 32, 33, 34) diagnosed among the 2,500 women who were referred to Kuopio University Hospital because of a clinical breast lump, suspicious mammographic finding, or a breast symptom (e.g., pain, nipple discharge) between April 1990 and December 1995. Women willing to participate in the project were interviewed and examined by a trained study nurse before any diagnostic procedures. The participation rate of patients with diagnosed breast cancer was 98%. Thus, the patient series represents unselected typical breast cancer cases from the University Hospital catchment area. Altogether, 479 invasive and 41 noninvasive carcinomas were diagnosed. After surgical treatment, the patients were offered adjuvant chemotherapy and/or hormonal therapy and radiotherapy, depending on the mode of the surgery, the patient’s menopausal status, and the stage of the disease, according to the national guidelines (35) . In brief, postoperative radiotherapy was given to all patients treated with breast-conserving surgery and to all patients with axillary node-positive (pN+) status irrespective of the mode of surgery. All premenopausal patients with pN+ status and some with axillary node-negative (pN−) status presenting with other adverse prognostic factors, such as estrogen receptor (ER)/progesterone receptor (PR) negative or poorly differentiated tumor, were given adjuvant chemotherapy (intravenous cyclophosphamide, mitoxantrone, methotrexate, and 5-fluorouracil) for six cycles, which was the standard therapy at the time. All postmenopausal women with ER- and/or PR-positive tumors were adjuvantly treated with either tamoxifen or toremifene for 3 years within a clinical study protocol. Thus, within a stage, the postoperative treatment was rather uniform, with only a few exceptions due to concurrent conditions. Stage was assessed by using the UICC classification (36) . Patients with noninvasive carcinomas, a previous history of breast cancer, metastatic disease (stage IV), or insufficient tumor material were excluded from the present study. Thus, 421 patients with sufficient primary tumors and complete clinical histories were available for MMP-2 analyses, 415 patients were available for MMP-9 analyses, 417 patients were available for AP-2 analyses, and 419 patients were available for HER2 analyses. The mean age of the 421 patients was 59.2 years (median, 56.8 years; range, 23.3–91.6 years). The mean follow-up time was 55.0 months (median, 57.5 months; range, 1.2–115.1 months). During the first 5 years of follow-up, 75 of the 421 patients (18%) had a recurrence, 49 patients (12%) died of breast cancer, and 37 patients (9%) died of other causes. The 5-year recurrence-free survival (RFS) rate of the patients was 79%, the breast cancer-related survival (BCRS) was 86%, and the overall 5-year survival rate was 77%. The 5-year survival of excluded stage IV patients (n = 19) was 21%. None of the patients were lost to follow-up. The clinicopathological data of the patients are summarized in Table 1⇓ .
The tumor samples were fixed in 10% neutral buffered formalin, processed routinely, and embedded in paraffin. The histologic diagnosis was confirmed by reviewing one to four original sections of the primary tumor. All tumors were simultaneously reevaluated for histologic type and grade by two senior pathologists, who were unaware of the clinical data. The most representative blocks were selected for cutting into new 5-μm–thick sections for immunohistochemical analyses.
The MMP-2 and MMP-9 stainings were demonstrated using Sequenza Immunostaining Center (Shandon Scientific Limited, Astmoor, United Kingdom). In brief, the sections were deparaffinized in xylene, rehydrated in EtOH, and washed twice with distilled water. For better antigen retrieval, the samples were boiled three times for 5 minutes in a microwave oven in a citrate buffer (pH 6.0). Endogenous peroxidases were blocked by 5% hydrogen peroxidase treatment for 5 minutes. The samples were washed with PBS (pH 7.2) and incubated in 1.5% normal horse serum for 35 minutes to prevent nonspecific antigen binding. The primary antibody for MMP-2 was a mouse monoclonal antibody toward a human proform of MMP-2 (Ab-1; clone CA-4001; Neomarkers, Fremont, CA), used at a working dilution of 1:25. The primary antibody used for detection of MMP-9 was a mouse monoclonal antibody, which recognizes both the pro- and active forms of human MMP-9 (MAB3309; clone 56-2A4; Chemicon, Temecula, CA), used at a dilution of 1:2,500. The samples were incubated with the primary antibody overnight at 4°C. Before applying the secondary antibody, the samples were washed twice with PBS. The slides were incubated for 45 minutes with the biotinylated secondary antibody, followed by a wash and a 50-minute incubation in an avidin-biotinylated peroxidase complex reagent (Vectastain Rabbit ABC Elite Kit; Vector Laboratories, Burlingame, CA). Expressions were visualized with a 5-minute diaminobenzidine tetrahydrochloride (Sigma, St. Louis, MO) treatment. The slides were counterstained with Mayer’s hematoxylin, dehydrated, and mounted with DePex (BDH Ltd, Poole, United Kingdom). A routinely processed breast cancer section without the primary antibody served as a negative control in each staining series. Placenta and known positive breast cancer samples served as positive controls. Control sections stained as expected.
Activator Protein-2 and HER2 Immunohistochemistry.
The immunohistochemical procedures for AP-2 (12) and HER2 (37) have been reported previously. In brief, a rabbit polyclonal antibody for human AP-2 (C-18, specific for AP-2α, AP-2β, and AP-2γ; Santa Cruz Biotechnology, Santa Cruz, CA) was used as a primary antibody at a working dilution of 1:2,000. An AP-2–positive melanoma was used as a positive control. HER2 immunohistochemistry was performed at Kuopio University Hospital according to a routine staining method using a TechMate 500 staining automat (DAKO, Glostrup, Denmark). The primary antibody used was NCL-CB11 (Novocastra Laboratories Ltd., Newcastle-upon-Tyne, United Kingdom), diluted 1:100. Expression was visualized by the labeled streptavidin-biotin system (LSAB, ChemMate Detection Kit; DAKO). A known 3+ stained, amplified breast cancer sample served as a positive control.
Scoring of Immunoreactivity.
The specimens were analyzed by two observers (J. M. P. and K. M. R.), who were unaware of the patients’ clinical outcome. Discrepancies between the observers were found in <10% of the slides examined, and consensus was reached on further review. Expression of both MMP-2 and MMP-9 was cytoplasmic and was recorded separately for tumor and stromal cells. The median value (80% for MMP-2; 85% for MMP-9) was used as a cutoff for MMP expression in the tumor cells. The categories for MMP-2 were reduced (<80%) and high (≥80%), and the categories for MMP-9 were reduced (≤85%) and high (>85%). Stromal expression of MMPs was divided into positive and negative expression groups according a 20% cutoff for positively stained stromal cells. HER2 staining was analyzed according to a scoring proposed by the HercepTest (DAKO), and samples with a score of 2+ or 3+ were considered HER2 positive (37) . For nuclear AP-2 expression, the median value (80%) was used as a cutoff value (12) . If >10% of the cells in the tumor area were stained (12) , cytoplasmic AP-2 expression was considered positive.
The statistical analyses were carried out by using the SPSS for Windows 9.0 program (SPSS Inc., Chicago, IL). The associations between MMP-2 and MMP-9 immunohistochemistry, AP-2 and HER2 expressions, and clinicopathological parameters were tested with contingency tables and a χ2-test. The univariate survival analyses were performed using Kaplan-Meier’s log-rank analysis, and the independent prognostic value of variables was further examined with Cox’s regression analysis. Probability values of ≤0.05 were considered significant in the analyses. In Cox’s multivariate analysis, the Enter method was used with an additional removal limit of P < 0.10. Both BCRS and RFS were examined. The recurrence-free time was defined as the time between diagnosis and the date of first local recurrence or a distant metastasis, whichever appeared first. Patients who remained healthy or died without breast cancer during the follow-up period were censored at the time of the last follow-up visit or death.
Expression of Matrix Metalloproteinase-2 and Matrix Metalloproteinase-9.
The expression patterns of MMP-2 and MMP-9 were cytoplasmic both in tumor cells and in stromal fibroblasts and inflammatory cells. In addition, benign breast epithelium and vascular endothelium stained positively for MMPs. The general expressions of MMP-2 and MMP-9 in the tumor cells were high, with median values of 80% for MMP-2 and 85% for MMP-9 (range, 0–100%). Fifty four percent and 52% of tumors expressed high levels of MMP-2 and MMP-9, respectively. Stromal MMP-2 positivity was observed in 46% of the cases, and MMP-9 positivity was observed in 38% of the cases. High MMP-2 expression in carcinoma cells was positively associated with high stromal MMP-2 expression (χ2 = 6.9; P = 0.009), whereas MMP-9 expression in cancer cells and MMP-9 expression in stromal cells were not associated with each other.
Expression of Activator Protein-2 and HER2.
The expression of nuclear AP-2 (range, 0–100%) was high (≥80%) in 50% of the carcinomas, and cytoplasmic AP-2 expression was seen in 47% of the cases. Pathological membranous HER2 overexpression (i.e., 2–3+) was observed in 13% of the carcinomas.
Relation of Matrix Metalloproteinase-2 and Matrix Metalloproteinase-9 Expressions to Activator Protein-2.
High MMP-2 expression in carcinoma cells was associated with high nuclear AP-2 expression (χ2 = 9.6; P = 0.002) and high cytoplasmic AP-2 expression (χ2 = 22.0; P < 0.001) in the whole patient group, as well as in the nodal subgroups (Table 2)⇓ . Stromal MMP-2 expression was not associated with AP-2 (Table 2)⇓ .
Similarly, high MMP-9 expression in malignant cells was associated with high nuclear AP-2 expression (χ2 = 8.5; P = 0.004) in the whole patient group and in the node-negative patient group (χ2 = 5.4; P = 0.020). Stromal MMP-9 expression and AP-2 were not statistically related. Results are summarized in Table 2⇓ .
Relation of Matrix Metalloproteinase-2 and Matrix Metalloproteinase-9 Expressions to HER2.
Stromal MMP-2 positivity was associated with HER2 overexpression in the whole patient group (χ2 = 5.8; P = 0.016) and in the node-negative patient group (χ2 = 7.7; P = 0.005; Table 2⇓ ). Expression of MMP-2 in carcinoma cells was not associated with HER2 expression (Table 2)⇓ .
In the node-negative patient group, high MMP-9 expression in carcinoma cells was associated with HER2 overexpression (χ2 = 4.0; P = 0.046; Table 2⇓ ). This was not seen in the whole patient group or in the node-positive patient subgroup (Table 2)⇓ . Positive stromal MMP-9 expression was related to HER2 overexpression in ER+ tumors (χ2 = 4.4; P = 0.036), but not in other investigated subgroups.
Relation of Matrix Metalloproteinase-2 and Matrix Metalloproteinase-9 Expressions to Clinicopathological Data.
High MMP-2 expression in carcinoma cells associated with stage I disease (χ2 = 6.4; P = 0.040). Stromal MMP-2 positivity, instead, was related to poor differentiation (χ2 = 21.8; P < 0.001), ER negativity (χ2 = 20.0; P < 0.001), and ductal or other carcinoma types (χ2 = 15.7; P < 0.001).
High MMP-9 expression in carcinoma cells was associated with small tumor size (χ2 = 10.0; P = 0.007) and lower recurrence rate (χ2 = 4.4; P = 0.036). Positive stromal MMP-9 expression was associated with poor differentiation (χ2 = 8.8; P = 0.013) and ductal carcinoma type (χ2 = 8.1; P = 0.018).
Univariate Survival Analysis.
In the univariate analysis, MMP-2 expression, in either carcinoma cells or stromal cells, possessed no prognostic value for RFS or BCRS in the whole patient group (Table 3)⇓ or in the nodal subgroups. The 5-year RFS rate for patients with high MMP-9 expression in carcinoma cells was 83%, compared with 74% for patients with low MMP-9 expression (P = 0.0351; Fig. 1⇓ ; Table 3⇓ ). This survival advantage was not seen in the BCRS analysis (Table 3)⇓ or in the nodal subgroups.
Interestingly, positive MMP-9 expression in stromal cells predicted shorter RFS and BCRS in ER+ disease (P = 0.0389 for RFS, and P = 0.0081 for BCRS; n = 321). A similar result was obtained in a subgroup of T1, ER+ tumors (P = 0.0031 for RFS, and P = 0.0089 for BCRS; n = 171), in which the 5-year RFS for stromal MMP-9–negative patients was 89% compared with 70% for stromal MMP-9–positive patients (Fig. 2)⇓ . However, stromal MMP-9 expression did not have prognostic value in the whole patient group (Table 3)⇓ or in the nodal subgroups.
Shorter RFS and BCRS were also predicted by advanced-stage disease (P < 0.0001 for both), axillary lymph node positivity (P < 0.0001 for both), and HER2 overexpression (P = 0.0056 for BCRS, and P < 0.0001 for RFS). Shorter BCRS was, in addition, predicted by ER and PR negativity (P < 0.0001 for ER, and P = 0.0009 for PR). The results are summarized in Table 3⇓ .
There were 402 patients with complete clinical data available for BCRS and 403 patients with complete clinical data available for RFS analysis. Variables with a prognostic value in the univariate analysis were included in the multivariate analysis (stage, ER and PR status, and HER2 expression in BCRS; stage, MMP-9 expression in carcinoma cells, and HER2 expression in RFS). Shorter BCRS was independently predicted by advanced-stage disease (P < 0.0001; Table 4⇓ ). The independent predictors of shorter RFS were reduced MMP-9 expression in carcinoma cells (P = 0.0248), advanced-stage disease (P = 0.0002), and HER2 overexpression (P = 0.0001; see Table 4⇓ ).
Multivariate analysis was also performed in the subgroup of ER+ patients, in which positive stromal MMP-9 expression (P = 0.0023) and advanced-stage disease (P < 0.0001) independently predicted shorter BCRS (n = 306) and HER positivity (P = 0.0210) and advanced-stage disease (P = 0.0012) independently predicted shorter RFS (n = 309). In the subgroup of ER+, T1 tumors (n = 152), shorter BCRS was independently predicted by positive stromal MMP-9 expression (P = 0.0301) and advanced-stage disease (P = 0.0096), and shorter RFS (n = 167) was independently predicted by stromal MMP-9 positivity (P = 0.0131) and HER2 overexpression (P = 0.0050).
In the present study we investigated in a large prospective series of breast cancer patients the expression and prognostic value of MMP-2 and MMP-9, as well as their relation to AP-2, HER2, and known prognostic parameters. High gelatinase expression in carcinoma cells was associated with high AP-2 expression, and positive stromal MMP expression was related to HER2 overexpression, supporting a function for AP-2 and HER2 in MMP regulation. In addition, positive stromal MMP-9 expression predicted poor survival in ER+, small T1 breast cancers, whereas high MMP-9 expression in carcinoma cells favored survival.
In this study, the adjacent benign breast epithelium was positive for both MMPs. In line with our results, Soini et al. (6) detected MMP-9 mRNA as well as MMP-2 and MMP-9 proteins in benign epithelial cells. Positive protein expression of MMP-2 and MMP-9 in benign epithelium was also observed by Lebeau et al. (38) . In addition, Dahlberg et al. (39) detected MMP-2 mRNA in some benign glandular cells. Indeed, in zymography studies, the expression of MMP-2 and MMP-9 has been detected in benign tissue, and enzyme activities have been less than that in malignant tissue (40, 41, 42) . However, benign epithelial cells have also remained negative for protein and/or mRNA (5 , 43) , or the expression has been localized in other structures (3 , 44) .
In the present study, the antibodies used recognized pro–MMP-2 and both the pro- and active forms of MMP-9. The gelatinase expressions were cytoplasmic in both carcinoma and stromal cells, as also described previously (3 , 5 , 6 , 44) . However, in some studies, MMP-2 protein has been expressed mainly in the carcinoma cells (7, 8, 9, 10 , 45) , whereas MMP-9 protein and mRNA have been mostly localized into stromal cells (40 , 46) . In addition, in several in situ hybridization studies, MMP-2 and MMP-9 mRNAs have been identified mainly in stromal cells, although some has been detected in carcinoma cells (6 , 40 , 43 , 46, 47, 48) . Previously, it has been suggested that stromal fibroblasts secrete MMPs, which are stored and activated in carcinoma cells (4) .
In the present series, high nuclear AP-2 expression was often associated with high MMP-2 and MMP-9 expression in carcinoma cells. This is noteworthy because MMP transcription has also been shown to occur in breast carcinoma cells (6) , a situation that may be regulated in part by AP-2 in breast carcinomas. Further evidence for this possible phenomenon has been obtained from in vitro studies, in which AP-2 has been an important factor in induction of MMP-2 transcription (16 , 21) . In addition, Sivak et al. (24) have shown that AP-2α is required for MMP-9 transcription in corneal epithelium. On the other hand, in melanoma cells, wild-type AP-2 could down-regulate MMP-2 transcription, and inactivation of AP-2 could lead to increased invasiveness via an increase in MMP-2 (14) . Based on our results and those in the literature (12 , 49) , we hypothesize that the role of AP-2 in MMP regulation in breast cancer may be tumor suppressive, by unknown mechanisms. Possibly this happens via tissue inhibitor of metalloproteinase (TIMP)-1 and TIMP-2 regulation (50 , 51) , both of which can inhibit activated MMPs, including MMP-2 and MMP-9 (52, 53, 54) . The complex regulation between AP-2 and MMPs and the separate roles of different AP-2 family members in breast cancer still need further investigation (55, 56, 57, 58) .
In the present study, positive stromal MMP-2 and MMP-9 expressions were related to HER2 overexpression. Previously, high MMP-2 mRNA expression in the stromal cells of breast carcinoma has been associated with c-erbB2 positivity, too (48) . Interestingly, association was seen in early carcinogenesis (in N−, in stage I disease, and in T1 tumors for MMP-2; data not shown). Positive stromal MMP-9 was associated with HER2 overexpression in ER+ disease, in which MMP-9 predicted shortened survival. This may indicate that HER2 induces stromal MMP expression and/or activation in an early phase of breast cancer, which further increases disease progression and metastasis, as suggested in the literature (27 , 28 , 30 , 59 , 60) . A possible explanation may be increased stimulus mediated by the extracellular matrix metalloproteinase inducer, EMMPRIN (61 , 62) . Differences between MMP-2 and MMP-9 expression in relation to HER2 in the present study may be due to different antibodies as well as different transcriptional regulation of the gelatinases (19 , 63) .
In this study, high MMP-9 expression in carcinoma cells was associated with small tumors and a lower recurrence rate. Previously, Scorilas et al. (5) reported that reduced MMP-9 expression in breast carcinoma cells was associated with poor prognostic factors such as large tumor size, giving further strength to our results. Jones et al. (3) found high MMP-9 expression in lobular carcinoma cells but failed to demonstrate other associations between MMP-9 in carcinoma or stromal cells and clinicopathological parameters. High MMP-2 expression in carcinoma cells, instead, has been related to only a few inverse prognostic factors (9 , 44) or shown to have no association with clinicopathological parameters in breast cancer (3 , 7) . In this material, high MMP-2 expression in carcinoma cells was related to stage I disease. Positive stromal expression of MMP-2 and MMP-9, instead, was associated with clinicopathological factors related to aggressive disease. Previously, an increase in stromal MMP-2 mRNA in poorly differentiated breast tumors has been reported (43) , which is in accordance with our results. Positive correlations of MMP-2 and MMP-9 with tumor grade have been observed in zymography, too (40 , 45) .
In our material, neither stromal nor carcinoma cell MMP-2 expression possessed prognostic value, in line with previous findings in breast cancer (42 , 48 , 64) . On the other hand, MMP-2 has often been related to poor outcome in breast carcinomas (7, 8, 9, 10 , 65 , 66) . In ovarian adenocarcinoma, high stromal MMP-2 expression has been related to shorter disease-free survival (67) . In early-stage oral squamous cell carcinoma, MMP-9 and TIMP-2 expressions, but not MMP-2, have shown prognostic value (68) . MMP-2 staining results reported here represent pro–MMP-2 protein expression, and thus do not tell about the enzyme activity, which may be better related to prognosis. Several other factors such as membrane-type MMP-1 and TIMP-2 take part in the activation and regulation of MMP-2 and may alter prognosis. In fact, patient outcome may depend on the balance between MMP-2 and TIMP-2 in breast cancer, as suggested by Nakopoulou et al. (44) .
The prognostic value of stromal MMP-9 expression has not been studied previously in a large clinical breast cancer series. Interestingly, positive stromal MMP-9 expression predicted significantly shorter RFS and BCRS for ER+ patients, especially if the tumors were small (i.e., in a group with usually very good prognosis). A possible explanation is that positive stromal MMP-9 expression may represent an increase in the active enzyme form that could lead to metastatic process and increased malignancy, for example, via enhanced angiogenic activity (69, 70, 71) . MMP-9 expression, in addition, may be affected by hormones. It has been shown that activation of ER-α by estrogen has resulted in tumor progression by stimulating cell growth and invasiveness via acceleration of the expression of MMPs including MMP-9 (72) . Other studies have confirmed the relationship between steroid hormones and increased gelatinase activity as well (73, 74, 75) . Evaluating stromal MMP-9 expression may add valuable information on breast cancer prognosis, especially in early carcinogenesis.
High MMP-9 expression in carcinoma cells independently predicted improved RFS in the whole patient group. This is in line with the results of Scorilas et al. (5) , who reported that low MMP-9 expression in breast carcinoma cells predicted poor survival for N− patients (5) . A similar trend was seen in a study with a small number of breast cancer patients (65) . Increased MMP-9 mRNA expression, instead, has been inversely associated with the number of survivors in breast cancer (66) , whereas in other studies, MMP-9 has not been associated with prognosis at all (42 , 64) . Additional studies are required to assess whether the results presented here may be related to a shift from inactive to active MMPs during tumor progression.
We conclude that immunohistochemically detected positive stromal MMP-9 expression predicts shortened survival in hormone-sensitive, small T1 breast cancers, whereas MMP-9 expression in carcinoma cells offers survival advantage. HER2 overexpression may be responsible in part for MMP induction in early breast cancer, resulting in increasing invasiveness. The association between AP-2 and MMP expression suggests a role for AP-2 in gelatinase regulation, but the exact mechanism, however, remains to be investigated. Finally, evaluation of MMP-9 expression may add valuable information regarding breast cancer.
The authors thank Anna-Kaisa Lyytinen, Mervi Malinen, Aija Parkkinen, and Riikka Eskelinen for skillful technical assistance. The statistical advice of Pirjo Halonen is also acknowledged.
Grant support: EVO funding of Kuopio and Tampere University Hospitals, Kuopio University Funding, Research Foundation of Orion Corporation, The Culture Fund of Finland (North Savo Fund), and The Finnish Medical Society Duodecim.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Veli-Matti Kosma, Department of Pathology and Forensic Medicine, University of Kuopio, P. O. Box 1627, FIN-70211 Kuopio, Finland. Phone: 358-17-162-750, Fax: 358-17-162-753; E-mail:
- Received May 31, 2004.
- Revision received August 8, 2004.
- Accepted August 13, 2004.