A High Serum Matrix Metalloproteinase-2 Level Is Associated with an Adverse Prognosis in Node-Positive Breast Carcinoma

In breast carcinoma, tumor involvement of the axillary lymph nodes is considered to be the most important prognostic factor for the development of a metastatic disease and, therefore, the strongest indicator for delivering adjuvant therapy. However, the clinical outcome of node-positive breast cancer is heterogeneous despite adjuvant therapy. In addition, lymph node involvement does not provide information of the complete biological features of the tumor, nor predict the responsiveness to therapies. Therefore, there is a need for biomarkers that accurately predict long-term outcome in these patients.

Several biomarkers provide prognostic information for patients with breast cancer. For example, overexpression of HER2, p53, vascular endothelial growth factor, and cyclin E proteins in the primary tumor has been shown to predict increased risk for metastatic disease and shorter overall survival (1, 2, 3, 4, 5, 6, 7, 8). Conversely, expression of estrogen receptors (ERs) and progesterone receptors (PgRs) has been associated with favorable outcome (9, 10, 11, 12, 13). To date, however, among many biomarkers studied, only the ER and PgR status are accepted predictors for the responsiveness and used in clinical decision making for adjuvant endocrine treatment. Moreover, a lack exists for soluble biomarkers that could be easily detected from serum both pre- and postoperatively.

Matrix metalloproteinases (MMPs) are a broad family of zinc-dependent endopeptidases that play a key role in extracellular matrix degradation, implicated in tumor cell invasion, metastasis, and angiogenesis (14, 15, 16). Of >25 family members described to date, a gelatinase MMP-2 is most consistently linked to breast carcinoma. The expression and prognostic significance of MMP-2 is not fully clarified, but recent studies implicate that increased MMP-2 expression in the primary tumor is associated with aggressive disease and unfavorable outcome (17, 18, 19). MMP-2 has also been detected in serum of breast cancer patients (20), but the prognostic value of serum MMP-2 levels in breast cancer patients is thus far unclear. Using quantitative enzyme-linked immunoassays we have determined the value of postoperative serum MMP-2 level as a prognostic factor in node-positive breast cancer.

The study population consisted of 133 node-positive breast cancer patients treated in the Department of Oncology, Helsinki University Central Hospital, between 1990 and 1993, and serum sample taken before adjuvant therapies. The characteristics of the patients with respect to tumor histology, differentiation grade, tumor size, nodal status, and ER and PgR status are listed in Table 1. Different treatments are shown in Table 2. All of the patients underwent surgery, and 132 patients received postoperative radiotherapy. Adjuvant chemotherapy was given to premenopausal patients, whereas postmenopausal patients were randomly allocated to receive adjuvant tamoxifen 20 mg/day or toremifen 60 mg/day for 3 years. The follow-up for all of the living patients was 5 years. Patients received all of the treatments within a randomized trial, which assessed to role of clodronate treatment (Bonefos; Schering) in an adjuvant setting (21). Due to interfering effect of clodronate on the prognostic value of serum MMP2, the patients who received clodronate were not included in this report. The prognostic value of serum MMP2 in the clodronate-treated patients will be reported separately.

Peripheral venous blood samples were collected in sterile test tubes postoperatively before any cancer therapy was given, centrifuged at 3000 × g for 10 min and then stored at −20°C.

Serum MMP-2 and MMP-9 concentrations were determined by using human MMP-2 and MMP-9 ELISA (Amersham) according to the manufacturer’s instructions. For each MMP-2 and MMP-9 analyses, 100 μl of serum diluted 1:100 and 1:50 or 1:25 was used, respectively. All of the analyses and calibrations were carried out at least in duplicate. The mean values were used for statistical analyses. The calibrations on each microtiter plate included recombinant standards. Optical densities were determined using a microtiter palate reader (Multiscan RC Type 351; Labsystems, Helsinki, Finland) at 405 nm. The blank was subtracted from the duplicate readings for each standard and sample. Concentrations are reported as ng/ml. When 40 serum samples were measured twice in the same assay, the intra-assay coefficients of variation were 0.01–6.45% for MMP-2 and 0.01–8.69% for MMP-9. When two serum samples were assayed twice in five separate assays, the interassay coefficients of variation for MMP-2 and MMP-9 were 4.74–8.15% and 2.47–8.66%, respectively. Serum MMP-2 and MMP-9 levels were determined without any knowledge of the survival or other clinical data.

The strength of the associations of MMP-2 with continuous variables was tested with Spearman rank correlation. Survival rates were estimated by the Kaplan-Meier method. Both uni- and multivariate analyses on the prognostic impact of MMP-2 and MMP-9 were performed using Cox proportional hazards model. The most important prognostic factors were included in the Cox regression univariate analyses. The factors, which had statistically significant impact on disease-free survival (DFS) and overall survival (OS), were included in multivariate analyses. The χ² test and Mann-Whitney nonparametric U test was used to assess differences in the frequency of individual prognostic factors. Statistical data processing was carried out with SPSS software for Macintosh (SPSS, Inc., Chicago, IL).

Of the 299 patients randomized to arm or to receive oral clodronate (21), control group consisted of 150 patients. Of these, serum was available from 133 patients, who were included in the study. The median serum levels for MMP-2 and MMP-9 were 5.25 ng/ml (range, 11.67 ng/ml) and 2.76 ng/ml (range, 9.95 ng/ml), respectively. When the median MMP-2 was used as a cutoff value between two groups (low versus high), no differences in tumor histology, grade, size, lymph node status, ER status, or PgR status was observed (Table 1). No correlation was found with serum MMP-2 levels and age, menopausal status, nodal status, receptor status, or adjuvant therapies (Spearman rank-order correlation coefficients, P > 0.11 for all of the analyses). MMP-2 levels tended to be higher in patients with large primary tumor, but statistical significance was not reached. Furthermore, no association between serum MMP-2 and MMP-9 levels was found (r_s = 0.07; P = 0.403). The surgery type among the patients in MMP-2 high group tended to be breast-conserving more often than in the MMP-2 low group (P = 0.055). Otherwise, different treatments were balanced between the groups.

A significant inverse relationship of serum MMP-2 levels with a poor prognosis was observed during a 5-year follow-up (r_s = −0.35; P < 0.0001). The 5-year OS for patients with low MMP-2 levels (<median) was 91% compared with 75% for those with high MMP-2 levels (>median; P = 0.020; Fig. 1,A; Table 3). Conversely, the DFS for women with low and high MMP-2 levels were 82% and 58% (P = 0.005), respectively (Fig. 1,B; Table 3). Whereas local recurrence rates were nonsignificantly increased in MMP-2 high study group in comparison with low group (18% versus 6%; P = 0.16; Fig. 2,A), distant metastases appeared considerably more often for MMP-2 high group patients (37% versus 18%; P = 0.008). Additional analysis showed that the appearance of both bone and visceral metastasis were associated with high MMP-2 levels (24% versus 11%, P = 0.050; and 34% versus 12%, P = 0.018, respectively; Fig. 2, B and C). The dose-response relationship of MMP-2 levels and survival was confirmed by using the highest tertile as cutoff value in survival analyses (DFS, 81% versus 50%, P = 0.0001; OS, 88% versus 75%, P = 0.048). Unlike MMP-2, serum MMP-9 levels did not correlate with outcome (r_s = −0.06; P = 0.51; Table 3).

In a subanalysis of ER-positive patients, OS was 95% in the MMP-2 low group and 78% in the high group (P = 0.052; Fig. 3,A). DFS was also significantly better in patients with low MMP-2 levels than high levels (85% versus 58%; P = 0.014; Fig. 3,B). In contrast, in ER-negative patients, no significant differences in the OS or DFS between MMP-2 low and high groups was observed (OS, 73% versus 69%, P = 0.25; DFS, 73% versus 63%, P = 0.32; Fig. 3, C and D).

The results of Cox uni- and multivariate analyses confirmed the prognostic effect of serum MMP-2. In univariate analyses, significant prognostic factors for DFS nodal status (P < 0.0001), PgR status (P = 0.001), tumor size (P < 0.001), and serum pretreatment MMP-2 level (low versus high level, P = 0.004). Age, ER status, and adjuvant treatment (antiestrogen versus chemotherapy) had no significant influence on DFS (P = 0.39, P = 0.28, P = 0.36, respectively). All of the variables that had a statistically significant impact on DFS were included in the multivariate analyses. Nodal and PgR status, tumor size, and serum MMP-2 level remained statistically significant for DFS also in multivariate analyses (Table 4). According to serum MMP-2 levels, the relative risk of recurrence was 2.21.

In the univariate analyses for OS, significant prognostic factors were nodal status (P < 0.0001), PgR status (P < 0.001), ER status (P = 0.032), tumor size (P = 0.003), and serum pretreatment MMP-2 level (P = 0.020), whereas age and adjuvant treatment had no significant influence on OS (P = 0.21 and P = 0.36, respectively). In the multivariate analyses, all of the other prognostic factors, except ER status, remained statistically significant for OS (Table 4). According to serum MMP-2 levels, the relative risk of death was 3.25.

In the present study, we demonstrated that postoperative serum MMP-2 level is a significant independent prognostic factor in node-positive breast cancer patients. Measurement of serum MMP-2 level allowed us to identify a subgroup of patients with favorable prognosis.

To date, nodal status is the strongest prognostic factor in primary breast cancer. Node-positive breast carcinoma is thought to have a poor prognosis, as ∼40% of patients develop distant metastases and ultimately die for the disease even if they have undergone curative surgery and adjuvant therapy (22). Nevertheless, the clinical outcome of node-positive patients is heterogeneous and includes patients whose outcome is comparable with node-negative patients. Those are mainly cases with ER- and PgR-positive tumors, and low proliferative activity. Interestingly, in the present study we found a new indicator for favorable prognosis. Patients with low postoperative serum MMP-2 level had a significantly better prognosis than patients with high MMP-2 level. Only 18% of node-positive breast cancer patients with low serum MMP-2 levels relapsed during a 5-year follow up in comparison with 42% of the patients in a high MMP-2 group. This was especially seen in ER-positive patients, where the difference was even larger (27% in favor of MMP-2 low group). The 5-year DFS and OS for patients with low postoperative MMP-2 level and ER-positive tumor were 87% and 95%, respectively. No significant difference was observed in ER-negative patients.

MMP-2 may play a role in the development of distant metastases in various cancers (23). In breast carcinoma, MMP-2 expression in the tumor tissue has been associated with unfavorable prognosis both in node-negative and -positive cases (17, 18, 19, 24). Our finding that elevated postoperative MMP-2 levels in the circulation of breast cancer patients are associated with poor prognosis is consistent with these studies. Although the prognostic influence of soluble MMP-2 is novel in breast cancer, high serum MMP-2 levels have shown previously to predict the recurrence of urothelial carcinoma, colorectal carcinoma, ovarian cystadenocarcinoma, and thymoma (25, 26, 27, 28). Furthermore, in several different cancer types, such as melanoma, thymoma, lung, colon, prostate, and breast carcinomas, high MMP-2 levels are associated with more advanced disease (20, 28, 29, 30, 31, 32).

The significance of circulating MMP-2 is unclear. Approximately 40% of patients with apparently localized breast cancer probably have a micrometastatic disease, which is clinically undetectable at the time of diagnosis, but after several years of delay accounts for recurrences and disease-related deaths (33, 34, 35). Soluble biomarkers, which identified micrometastatic disease, would be valuable in defining the risk of recurrence and guiding clinicians in treatment-related decisions for adjuvant therapies. To date, however, the prognostic impact of soluble biomarkers in breast cancer is lacking. Together with a previous report demonstrating that high preoperative serum MMP-2 levels reflect more advanced Tumor-Node-Metastasis staging (20), the data in this study show that serum MMP-2 is a strong and independent indicator of biologically aggressive disease. Although the origin of serum MMP-2 cannot be determined, it is possible that circulating MMP-2 is derived from the areas of tumor tissue undergoing active matrix dissolution. The hypothesis is supported by immunohistochemical stainings of the primary tumors, in which we detected the most prominent MMP-2 reactivity on the surfaces of cancer cells in the invasive fronts of the tumor tissues (data not shown). Because it is well established that MMP-2 in carcinoma tissue is primarily synthesized by peritumor stromal cells, and only to a lesser extend by cancer cells (36, 37, 38), the cellular source of MMP-2 may also be stromal fibroblast. On the basis of current data in the literature (15, 16, 39), we speculate that before MMP-2 becomes soluble, pro-MMP-2 is secreted into the pericellular matrix and activated by membrane type MT1-MMPs, which are localized on and produced by carcinoma cells. Subsequent to protease activation, matrix undergoes dissolution, and MMP-2 can be released into circulation.

In conclusion, measurement of serum MMP-2 levels allowed us to distinguish node-positive breast cancer patients with different outcomes after treatments. Our data identifying low MMP-2 levels in the serum of node-positive patients with favorable prognosis provides, to our knowledge, one of the first analyses of a soluble biomarker as a prognostic factor in breast carcinoma. It appears that the outcome of patients with low serum MMP-2 levels after conventional adjuvant therapies is reasonable. In contrast, association of high serum MMP-2 with poor outcome defines a group of patients who requires new treatments. Prospective studies in a larger population including also node-negative patients should be carried out to confirm our findings. However, it is tempting to speculate that serum MMP-2 may be useful in the treatment of breast cancer as a decision-making biomarker for adjuvant therapies.

We thank Anitra Ahonen for excellent technical assistance.