Preoperative Serum Levels of Matrix Metalloproteinase-2 (MMP-2) and Survival of Breast Cancer among Korean Women

Breast cancer is the most common cancer in women worldwide. For the past 3 decades, 5-year relative survival rate of the disease has increased from 70% to 90% on account of early detection and improved treatment (1). However, patients with metastasis or recurrence exhibited lower survival than those with primary cancer (2). Therefore, there is a need for prognostic factors to predict postoperative recurrence after curative resection of the tumor.

Many established prognostic factors for breast cancer exist including tumor size, nodal status, micrometastasis, sentinel lymphadenectomy, histologic grade, histologic type, mitotic figure count, and hormone receptor status (3). As prognostic factors, C-erbB-2 (HER2-neu), p53, and lymphatic or vascular channel invasion were validated extensively. However, traditional prognostic factors are limited to predict the outcome because of heterogeneity of breast cancer types and various kinds of manifestation and prognosis as well. Thus, additional prognostic markers would be needed to accurately predict the outcome of the disease such as metastasis, recurrence, or survival.

Because serum biomarkers are various and simple to measure, they can be useful to predict the outcome of breast cancer. A number of potential biomarkers have been discovered with quantitative analysis for circulating proteins, nucleic acids, metabolites, and tumor cells. Matrix metalloproteinases (MMPs), as circulating proteins, regulate tumor initiation and growth in breast cancer contributing to the stimulation of angiogenesis, activation of growth factors, and degradation of inhibitory growth factors (4). MMP-2, known as a gelatinase, a subgroup of MMPs, degrades the components of the basement membranes such as gelatin, elastin, fibronectin, and collagen (5, 6). The collagen-rich extracellular matrix (ECM) is degraded by MMP-2, and it has been correlated with invasiveness, metastasis, and recurrence in tumor cells (7). Therefore, MMP-2 has been thought of as a predictor of recurrence or metastasis risk or prognostic markers in cancer.

There were a number of studies showing that elevated serum levels of MMP-2 were associated with the severity and poorer prognosis in several types of cancers (8–18). In breast cancer, several studies have found that MMP-2 expression in tumor cells was associated with shortened disease-free survival (DFS), recurrence-free survival (RFS), or overall survival (OS; refs. 19–23), whereas few others have suggested inconsistent results (24, 25). Although there were several studies on the association between the levels of serum MMP-2 and clinicopathologic factors, the results were inconsistent across the studies (26–35). So far, only a few studies have investigated the association between MMP-2 levels and breast cancer prognosis (36, 37).

The objective of this study was to investigate the association between the preoperative serum levels of MMP-2 and the correlation with the clinicopathologic factors and breast cancer survival as well.

The patients with histologically confirmed incident breast cancer were recruited between 2004 and 2007 at Seoul National University Hospital and ASAN Medical Center (both in Seoul, Korea). After obtaining written informed consent, the patients completed a questionnaire administered by trained interviewers for information on demographic factors and known breast cancer risk factors (i.e., education level, family history of breast cancer, reproductive history, and lifestyle). Clinical information on the patients with respect to hormone receptor status, lymph node status, histologic grade, nuclear grade, tumor–node–metastasis (TNM) stage, tumor size, C-erbB-2 status, and p53 status was abstracted from the patients' medical record. In addition, peripheral blood samples were taken before operation.

For the 925 of 1,899 invasive breast cancer cases with the biospecimens, serum was available for the measurement. After the patients with insufficient serum samples for the assay, incomplete clinicopathologic data and inaccurate follow-up information were excluded, roughly 40% of the subjects (370 patients) were randomly selected as the eligible population for the analyses of serum biomarkers. After excluding patients who had prior history of any type of cancer, stage 0 or IV breast cancer (multi-cancer or metastatic disease at diagnosis, distant organ metastasis, and in situ breast cancer), 303 patients with breast cancer were included in the final analysis. The median follow-up period for all patients was 4.24 years (range, 0.18–5.28 years).

This study has been approved by the Committee on Human Research of Seoul National University Hospital (IRB No. H-0503-144-004).

Preoperative peripheral blood was collected in 10 mL serum storage tubes and then centrifuged at 3,000 × g for 10 minutes at room temperature. The serum was stocked in 0.3-mL aliquots in cryovials and stored at −80°C. ELISA was used to measure the serum levels of free active MMP-2 protein using a human MMP-2 ELISA kit (R&D Systems) according to the manufacturer's instructions. The minimum detectable dose (MDD), which was distinguished from the blank value, was evaluated by adding 2 SDs to the mean optical density value of 20 standard replicates and calculating the corresponding concentration. The MDD of MMP-2 ranged from 0.016 to 0.289 ng/mL, and the mean of the MDD was 0.047 ng/mL. The coefficient of variation (CV) was <10% in intra-assay precision. Optical densities were determined using a CODA Automated EIA analyzer (Bio-Rad Laboratories) at 450 nm. Serum MMP-2 was measured per sample, and the concentrations were reported as ng/mL.

Because the serum levels of MMP-2 were normally distributed (P = 0.09), parametric analyses were done. The Mantel–Haenszel χ² test was used to investigate the association between the serum levels of MMP-2 and the categorical variables for the clinicopathologic factors. The association between MMP-2 levels and continuous variables such as age at a diagnosis and body mass index (BMI) was tested by Pearson correlation test.

DFS was defined as the time between curative resection surgery and the date of first local recurrence, distant metastasis, second primary cancer of another organ, or death from any cause whichever appeared first during the follow-up period. Patients known to be alive with no evidence of disease were censored at the last follow-up date. The survival rates were estimated by Kaplan–Meier method, and the differences in DFS between the subgroups were compared by log-rank test.

Cox proportional hazard regression models were used to estimate the hazard ratios and 95% confidence intervals (CI). The multivariate analysis was tested considering the TNM stage (I, IIA-IIB, and IIIA-IIIC), estrogen receptor (ER) status (negative and positive) but not the tumor size and hormone therapy to reduce the impact of multicollinearity (P_{TNM stage and tumor size} < 0.001 and P_{ER status and hormone therapy} < 0.001 by χ² test), as well as the other factors did not affect adjusted HRs significantly after adjusting for other covariates and were not included in the final adjusted model. In addition, stratified analyses were done according to age at diagnosis, ER status, histologic grade, and nuclear grade and the interaction between MMP-2 levels, and the stratification variables were assessed by likelihood ratio test.

All P values less than 0.05 were considered as statistically significant with a 2-tailed test. The false discovery rate (FDR) test was done to evaluate the type 1 error. All the statistical analyses were done with SAS version 9.2 (SAS Institute Inc.) and STATA version 12.0 (Stata Corporation).

Of 303 subjects in the study population, 87 patients had recurrence (28.7%). Table 1 presents selected characteristics of the 303 patients with breast cancer. In univariate analysis, the TNM stage, tumor size, lymph node status, histologic grade, nuclear grade, ER status, progesterone receptor (PR) status, adjuvant chemotherapy, and hormone therapy had statistically significant influence on the DFS (P < 0.05). The age at diagnosis, BMI, smoking, education level, menopausal status, p53 status, C-erbB-2 status, and radiotherapy had no significant influence on the DFS (P > 0.05). In multivariate analysis, the TNM stage, tumor size, ER status, and hormone therapy maintained their statistically significant influence on the DFS of breast cancer (P < 0.05).

The median serum level for MMP-2 among the 303 patients was 198.5 ng/mL and the mean level was 202.4 ng/mL (range, 38.7–505.4 ng/mL). Table 2 shows the MMP-2 levels by selected characteristics. When MMP-2 levels were divided into 3 groups according to the concentration, the MMP-2 levels were significantly associated with the age at a diagnosis (P = 0.03), histologic grade (P = 0.02), nuclear grade (P = 0.03), and ER status (P = 0.03) inversely. For the other factors, there were no significant associations (P > 0.05).

In univariate analysis, no significant association was found between MMP-2 levels and DFS (Fig. 1). In multivariate analysis, however, the HR of the third tertile group was 1.8-fold higher than the first tertile group when adjusted for the TNM stage and ER status (HR, 1.80; 95% CI, 1.04–3.11). As a continuous variable, the adjusted MMP-2 levels also showed a 1.34-fold increased risk of recurrence (HR, 1.34; 95% CI, 1.02–1.75; shown in Table 3).

Table 4 shows the association between DFS and MMP-2 according to age at diagnosis, ER status (Fig. 1A and B), histologic grade (Fig. 1C and D), and nuclear grade (Fig. 1E and F). The associations were found only in patients with poor prognostic factors (younger age, ER-negative, higher histologic grade, and higher nuclear grade). In groups with poor prognostic factors, the HRs for the third tertile levels of MMP-2 were 2.95 (95% CI, 1.42–5.92; P_interaction = 0.03 for histologic grade III) and 2.61 (95% CI, 1.26–5.39; P_interaction = 0.16 for nuclear grade III) compared with the first tertile as reference group. Although P for interaction was not statistically significant, similar stronger association were found among younger age group (HR, 2.13; 95% CI, 1.02–4.46; P_interaction = 0.60) and ER-negative patients (HR, 2.75; 95% CI, 1.32–5.73; P_interaction = 0.12). In contrast, there was no significant association among the group with good prognostic factors.

In the present study, the patients in higher levels of serum MMP-2 had significantly poorer prognosis. When stratified by known prognostic factors, inverse correlation was shown between serum MMP-2 level and prognostic factors, and the associations between MMP-2 and DFS were only significant among patients with poor prognostic factors (early diagnosis, ER-negative, high histologic grade, and high nuclear grade).

In breast cancer, MMP overexpression in tumor tissue is thought to contribute to the progression through proteolysis and cleavage of ECM components and ECM-associated molecules (38–40). Increased levels of MMPs induce epithelial–mesenchymal transition (EMT) of the epithelial cells in the tumor microenvironment. Induced EMT may produce more MMPs that facilitate cancer cell invasion and metastasis. Therefore, we hypothesized that the high serum MMP-2 level may be used in proteolysis and cell migration occurring in the process of metastatic spread. However, we found the relation among patients with bad prognostic factors only.

Consistent with our results, Leppa and colleagues reported that serum levels of MMP-2 were independent significant prognostic factor with similar follow-up duration in 133 patients with node-positive breast cancer (36) who had poor prognostic factors (node-positive). However, our study found that the difference of DFS according to MMP-2 was significant only in patients with ER-negative but not in patients with ER-positive breast cancer, whereas Leppa and colleagues reported the significant association only in patients with ER-positive breast cancer. Although not enough evidence has proved the relation between ER status and MMP-2, the absence of ER-α led to ECM dysregulation and oxidative injury and that induced higher MMP-2 activity in vivo (41). Furthermore, Leppa and colleagues measured the MMP-2 using postoperative serum in which a proteolytic imbalance could occur with breast tissue remodeling and MMP-2 levels were much lower than in our study (42).

In this study, higher levels of MMP-2 were shown among patients with good prognostic markers, and the concentration of MMP-2 was measured as free active MMP-2. Davies and colleagues found that although active MMP-2 levels were higher in breast tumor grade I than in grade III, activity ratio was significantly higher in grade III than in grade I (43). Furthermore, total circulating MMP-2 status is composed of active MMP-2, complex pro-MMP-2 with TIMP-2 and non-complex pro-MMP-2 (30). Although circulating pro-MMP-2 is the major component of total circulating MMP-2, there was no correlation between pro-MMP-2 and active MMP-2 (37), as well as no significant difference of total MMP-2 levels between breast cancer cases and controls, whereas the concentration of active MMP-2 was significantly higher in the healthy controls than in patients with breast cancer (30).

Kuvaja and colleagues showed that lower circulating levels of active MMP-2 were associated with bad progression of breast cancer among 71 women (P = 0.06 by log-rank test); however, the patients were mostly diagnosed with early-stage (I–II, >90%), and the association was shown by the univariate analysis without any adjustment of known prognostic markers (37). Kuvaja and colleagues also reported that total MMP-2 concentration had inverse correlation with nodal status, stage of tumor, and nuclear grade. In our study, relatively low levels of MMP-2 were shown in patients with poor prognostic factors (ER-negative, histologic grade III, nuclear grade III). Other studies on the association between MMP-2 levels and clinicopathologic factors were not inconsistent. While several studies reported that higher serum levels of MMP-2 were shown in the patients with more advanced tumor stage (26, 28, 33, 34), the others showed lower MMP-2 levels in patients with many poor prognostic factors including advanced tumor stage or grade (27, 29, 35).

The patients with ER-negative, high histologic grade, or high nuclear grade were thought to have a poor prognosis including metastasis, recurrence, and death. Nevertheless, the clinical outcomes of those prognostic factors were heterogeneous in our study. In the patients with those poor prognostic factors, we found MMP-2 as the indicator for poorer prognosis and the results that MMP-2 was associated with DFS were comparable with patients with good prognostic factors (ER-positive, low histologic grade, or low nuclear grade).

We searched the original articles using PubMed with "breast[All fields] AND (“MMP-2”[All fields] OR MMP-2[MeSH Terms]) AND “serum”[All fields] AND “humans”[MeSH Terms]" as the search terms. And the previous studies on serum MMP-2 in breast cancer are summarized in Supplementary Table S1. In our study, preoperative serum levels of MMP-2 ranged from 38.7 to 505.4 ng/mL in patients with breast cancer (median, 198.5 ng/mL; mean, 202.4 ± 69.9 ng/mL). Other studies, however, showed that MMP-2 levels varied from 5.26 to 1,971 ng/mL (27, 28, 30, 32, 34–37, 42, 44, 45). The wide range of MMP-2 levels can result from the differences in assays, time of sample collection, sample processing, different MMP-2 form, and composition of study population. In our study, the sample collection and process was proceeded by standard operating procedure and the assays were reliable [CV (%) < 10, mean MDD = 0.047 ng/mL]. A relatively small sample size and short follow-up time were the limitations of our study. The estimated power for 2-sample comparison of survivor functions of exponential test with α = 0.05, hazard difference (H₀ = 1.00, H₁ = 1.34), follow-up duration = 5.28 years, and the proportion of controls = 0.33 (first tertile of MMP-2 levels) was 70%, and FDR test showed no statistically significant on account of a small sample size.

In conclusions, elevated serum levels of MMP-2 were associated with a poor prognosis in patients with breast cancer. When stratified by prognostic factors, this relationship was only observed in patients with bad prognostic factors (histologic grade III, nuclear grade III, and ER-negative). Further studies including a larger sample size are needed to validate our results.

No potential conflicts of interest were disclosed.

Conception and design: N. Song, H. Sung, J.-Y. Choi, K.-M. Lee, S.K. Park, K.-Y. Yoo, S.-A. Lee, D. Kang

Development of methodology: N. Song, H. Sung, S.K. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Song, H. Sung, K.-Y. Yoo, D.-Y. Noh, S.-H. Ahn

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Song, H. Sung, J.-Y. Choi, S. Han, S. Jeon, M. Song, Y. Lee, C.-B. Park, S.K. Park

Writing, review, and/or revision of the manuscript: N. Song, J.-Y. Choi, S. Han, S. Jeon, M. Song, S.K. Park

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Jeon, S.-H. Ahn, S.-A. Lee

Study supervision: K.-M. Lee, K.-Y. Yoo, D.-Y. Noh, D. Kang

Discussion and interpretation of the results: N. Song, H. Sung, J.-Y. Choi, S.K. Park, K.-M. Lee, K.-Y. Yoo, S.-A. Lee, D. Kang

Data collection and sample selection: Y. Lee, C.-B. Park

All authors have read and approved the final manuscript. D. Kang, D. Noh, and S.-H. Ahn were principal investigators (PI) for each of the participating cooperative groups of the Seoul Breast Cancer Study (SeBCS).

This work was supported by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology (2011-0027212).

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.