Tumours arise from multiple genetic alterations, which affect primarily cell proliferation. A crucial difference between benign and malignant tumours is that the first remain encapsulated for undefined periods of time and do not form metastases, while the second acquire the ability to invade the underlying basal lamina and its adjacent stroma, which in normal tissues do not show passageways for cells of epithelial origin. A critical event, during progression of malignant carcinomas, is the invasive growth of neoplastic cells into the host tissues: this involves the onset of a number of complex interactions occurring at the tumour–host interface, including an extensive remodelling of the extracellular matrix (ECM). Degradation of the ECM requires the concerted action of a number of extracellular enzymes. Several enzyme families are known to be involved in extracellular proteolysis. These include serine proteases (e.g. Blasi and Carmeliet, 2002; Diamandis and Yousef, 2002), matrix metalloproteinases (MMPs; reviewed by Somerville et al, 2003), and disintegrin-metalloproteinase (ADAMs: A Disintegrin And Metalloproteinase Domain; reviewed by O'Shea et al, 2003). The MMP family is one of the most studied, since many members of the group have been recognised as involved in cancer invasion and metastasis. Recently, MMPs have been the object of renewed interest due to the additional roles attributed to their functions, for example, growth factor mobilisation and the processing of surface molecules (cf. Somerville et al, 2003). At present, there are 23 human MMP genes, which are structurally similar to each other, indicating that they evolved by duplication of a common ancestral gene followed by divergent evolution (Moura-da-Silva et al, 1996). Based on substrate specificity and domain organisation, the MMPs have been tentatively divided into four main groups: interstitial collagenases, gelatinases, stromelysins and membrane-type MMPs. Most MMPs are secreted as proenzymes and are organised into distinct structural domains, with some differences in domain composition and number (Nagase and Woessner, 1999). Gelatinase A (MMP-2) and gelatinase B (MMP-9) differ from other MMPs in that they have three tandem fibronectin type II repeats within the amino terminus of the catalytic module that mediates gelatin binding (Murphy et al, 1994). Moreover, these two enzymes interact with physiological inhibitors (respectively, TIMP-2 and TIMP-1) even if they are in the proenzymatic form (Overall et al, 1999). Traditionally, MMP-2 and MMP-9 have been correlated with the invasive stage of carcinomas, because of their ability to degrade type IV collagen, a major constituent of basement membranes (Stetler-Stevenson, 1990). More recent evidence suggests that MMP-2 and MMP-9 may also be involved in breast cancer initiation and growth through complex interactions with the main oncogenes and tumour-suppressor genes involved in the early stage of tumorigenesis (Duffy et al, 2000; Leeman et al, 2003, for reviews). For example, transfection of MCF-10A breast cancer cells with either c-erbB-2 or c-ras resulted in increased expression of MMP-2 (Giunciuglio et al, 1995), whereas transfection of MCF-7 cells with the ets gene PEA-3 led to increased production of MMP-9 (Kaya et al, 1996). Due to the key role of MMPs in tumorigenesis, several authors have proposed MMP-2 and or MMP-9 as useful prognostic markers (i.e. Duffy, 1996). Recent work on breast cancer patients has suggested that MMP-2 negativity may be linked with a favourable prognosis in node-negative breast carcinoma (Hirvonen et al, 2003) and that high activity levels of plasma MMP-9 in breast cancer patients are associated with a worst overall survival rate (Ranuncolo et al, 2003). Thus, the present research was conducted with the following objectives: (1) to support the diagnostic value of MMP-2 and MMP-9 in breast cancer using a highly sensitive zymographic method; (2) to determine the possible association of activity levels of serum forms of MMP-2 and MMP-9 with the current clinical parameters; and (3) to perform the first univariate analysis including c-erbB-2, a candidate marker of tumour aggressiveness, at present in course of validation. In the present study, we examined 80 patients with breast cancer and no detectable metastases. In all, 22 sera samples from healthy donors were used as control specimens. To assess possible correlations between the serum levels of MMP-2 and MMP-9 and the stage of the disease, the following clinical parameters were utilised: tumour size, lymph node involvement, tumour stage, histological grading, oestrogen receptor (ER) levels (76 out of 80 cases), progesterone receptor (PR) levels (76 cases), c-erbB-2 levels (46 cases). The statistical analyses have shown a significant increase of activity levels of both MMP-2 and MMP-9 in the sera of breast cancer patients compared with control sera. The correlation analyses indicated a significant association between high activity levels of both enzymes and c-erbB-2 overexpression. MATERIALS AND METHODS Sample collection A total of 80 patients (of which one Tis, 52 T1, 25 T2, two T3) diagnosed with breast carcinoma but without clinically apparent metastases were involved in this study. Sera were obtained prior to surgery, according to the ethical standards, with informed consent of patients at the Maddalena Hospital. Native serum was prepared using plastic tubes without coagulation accelerators, to prevent the release of gelatinases during platelet activation (Fernandez-Patron et al, 1999). Tubes were centrifuged at 1600 g for 10 min, 30 min after blood collection. For each sample, determination of protein concentration was performed using the method of Bradford (1976). Sera were aliquoted and stored at −80°C until used. Each aliquot was used only once in order to prevent enzyme activation due to freeze–thawing processes. For all patients, the histological diagnosis and the stage of cancer were established by assessment on paraffin sections at diagnostic laboratory of the Maddalena Hospital. Evaluation of oestrogen receptor (ER) and progesterone receptor (PR) was performed by immunohistochemistry using monoclonal antibodies from Dako Corporation (USA). c-erbB-2 expression levels were evaluated using the HercepTest assay (semiquantitative immunocytochemical assay), with scoring of 0 or 1+ for normal levels of expression and 2+ or 3+ for overexpression (DakoCytomation, DK). Control sera (n=22) were taken from healthy volunteers. Gelatin zymography Gelatin zymography was performed for both healthy control and cancer patients sera as follows: gels (SDS–PAGE, 7.5%) were copolymerised with gelatin (0.1%) (Sigma-Aldrich). For each sample, 28 μg of total serum protein was loaded. Electrophoresis was carried out using the minigel slab apparatus Mini Protean 3 (Biorad) at a constant voltage of 150 V, until the dye reached the bottom of the gel. Following electrophoresis, gels were washed in renaturation buffer (2.5% Triton X-100 in 50 mM Tris–HCl (pH 7.5)) for 1 h in an orbital shaker. Then the zymograms were incubated for 18 h at 37°C in incubation buffer (0.15 M NaCl, 10 mM CaCl2, 0.02% NaN3 in 50 mM Tris–HCl (pH 7.5)). Gels were then stained with Coomassie blue and destained with 7% methanol and 5% acetic acid. Areas of enzymatic activity appeared as clear bands over the dark background. Quantification of enzymatic activity Following zymography, the degree of gelatin digestion was quantified using a Sharp JX-330 scanner equipped with a transparency option interfaced to an IBM PC. Gels were scanned using Image-master software, version 1.2 for DOS (Pharmacia Biotech), in a grey scale mode at 169 μm pixel size and 1250–1650 (X–Y) pixel count, using the autodensity feature on a scale ranging from 0 (clear) to 255 (opaque). The image was digitally inverted, so that the integration of bands was reported as positive values. The pixel density was determined after background subtraction and used to calculate the integrated density of a selected band. Values of integrated density were reported in volume units of pixel intensity per mm2. The integrated density of each band is reported as the mean of three different measurements of the same gel for each sample run in triplicate. Statistical analysis For each group of subjects, data derived from zymographic quantification of activity of MMP-2 and MMP-9 were plotted using MS Excel software. Statistical analyses were performed using both MS Excel and Graph Pad Prism 4 (demo) software (for correlation analysis). Fitness of data to normal distribution was assessed using the method of Kolmogorov and Smirnov. This normality test quantifies the discrepancy between the examined distributions of data and an ideal Gaussian distribution; the test returns a P-value which is considered acceptable for values greater than 0.05. For all the distributions examined (i.e. activity levels of MMP-2 and MMP-9 in control and affected subjects), the test confirmed their fitness with a normal distribution. In order to estimate the significance of differences between cancer patients and control subjects, unpaired Student's T-test (with Welch's correction) was applied. Correlation of MMP-2 and MMP-9 activity levels with clinicopathological variables for breast cancer patients was performed using the Pearson correlation test. In all cases, data were considered significant for values of P 10%) 36 47.4 Progesterone receptor level (76 cases) Negative ( 10%) 35 46 cErbB-2 levels (46 cases) 0 33 71.7 1+ 4 8.7 2+ 2 4.3 3+ 7 15.3 , the cancer patients displayed a variable clinical profile in good agreement with current literature: in particular, node involvement was demonstrated for about 40% of studied cases, and c-erbB-2 overexpression was limited to approx. 20% patients. As shown in Figures 3 Figure 3 Correlation analysis between clinicopathological variables and progelatinase activity levels: For all variables, graphs marked with (A) indicate correlation with pro-MMP-2, while (B) indicates pro-MMP-9. (1) Correlation with lymph node involvement; (2) correlation with tumour size; (3) correlation with tumour staging; (4) correlation with tumour grading. and 4 Figure 4 Correlation analysis between the receptorial state of primary tumour and serum progelatinase activity levels: (A) indicates pro-MMP-2, while (B) indicates pro-MMP-9. (1) Correlation with progesterone receptor positivity; (2) correlation with oestrogen receptor positivity; (3) correlation with c-erbB-2 expression levels. , the activity of MMP-2 and MMP-9 in the sera of breast cancer patients did not correlate with tumour size, stage, node involvement and progesterone receptor levels. Instead, we found a negative correlation with ER expression (P=0.007 for MMP-2 and 0.027 for MMP-9). By contrast, a positive correlation was observed between gelatinolytic levels and c-erbB-2 overexpression (P=0.027 for MMP-2 and 0.007 for MMP-9). Moreover, a borderline significant negative correlation was found with nuclear grade (P=0.0511 for MMP-2 and 0.0794 for MMP-9). DISCUSSION At present, the strongest predictors of breast cancer metastasis are lymph node involvement and histological grading. These parameters are not enough selective to discriminate the putative subgroup of patients within the same clinical category. In fact, it is well documented that breast cancer patients with the same stage of disease can have markedly different outcome and therapy responses (van't Veer et al, 2002). Therefore, searching for new molecular markers is an open area of interest. In particular, the matrix metalloproteases have since long attracted the interest of investigators, due to their possible use as molecular markers and therapeutic targets (cf. e.g. King et al, 2003). The majority of MMPs are secreted as latent proenzyme forms and are subjected to regulated activation at the cell–matrix boundary (Nagase and Woessner, 1999). Their proenzyme forms are also secreted in body fluids where they can be easily detected. Several authors, including our group (Pucci-Minafra et al, 2001), have measured the serum or plasma levels of MMP-2 and MMP-9 in oncologic patients (e.g. Garbisa et al, 1992; Zucker et al, 1993; Hayasaka et al, 1996; Garbett et al, 1999; Oberg et al, 1999), in some cases reaching different conclusions. One reason could be the different pre-analytical sampling process (reviewed by Jung et al, 2001, 2002). Indeed, plasma preparation requires the addition of anticoagulants such as EDTA or heparin, while serum is often obtained by addition of clot accelerator. All these additives may interfere with measurements of gelatinase activity. In particular, EDTA is a strong inhibitor of MMPs and heparin is known to bind to some MMPs (Keller et al, 1986; Wallon and Overall, 1997). On the other hand, serum preparation with clot accelerators can induce platelets to release MMP-9 (Jung et al, 2001); this can be avoided by preparing native serum in the absence of clot accelerators, since quiescent platelets release negligible amounts of gelatinases (Fernandez-Patron et al, 1999). On this ground, our major aim was to study the gelatinolytic levels of serum forms of MMP-2 and MMP-9 by zymographic assays, and correlate data with current clinicopathological parameters. These zymographic tests have some advantages over immunologic assays, due to lower costs, more rapid time of execution and the possibility of detecting simultaneously multiple forms of the same enzyme. Our qualitative analysis of the zymograms showed that the majority of the 80 sera samples from breast cancer patients displayed sharp bands of lysis corresponding to the proenzyme forms of MMP-2 and MMP-9, respectively, in contrast to the 22 control samples which showed less pronounced lytic bands. In addition, the sera samples from breast cancer patients exhibited two other lytic activities of larger molecular size, similar to those previously identified as pro-MMP-9 dimers (200 kDa) and as a complex formed between pro-MMP-9 and its physiological inhibitor TIMP-1 (116 kDa) in colon cancer sera (Pucci-Minafra et al, 2001). These forms were absent or very faint in the majority of normal sera. Activated forms of both enzymes were never detected in any sera. The densitometric analyses indicated that the average levels of activity of circulating MMP-2 and MMP-9 in breast cancer patients were significantly higher than control sera (P<0.0001), and suggest that serum measures of MMP-2 and MMP-9 activity may have diagnostic value for discriminating subgroups of breast cancer patients. In order to relate our data with disease features, we applied statistical analyses correlating activity levels of both enzymes with clinical parameters. Univariate analysis revealed no correlation between both gelatinases and tumour size, staging, lymph node involvement and PR status, while a borderline correlation was observed with nuclear grading. Conversely, our data showed for the first time an inverse correlation between ER positivity (generally considered as a favourable prognostic factor) and elevated levels of gelatinase activity. More interestingly, high levels of gelatinases are significantly related to c-erbB-2 overexpression. c-erbB-2 overexpression has been described as an independent predictor of survival in breast cancer (Konecny et al, 2001), and its amplification has been reported for 20–25% of breast cancer patients (Dickson and Lippman, 1995) associated with more aggressive clinicopathologic features (Sahin, 2000). In addition, recent clinical investigations (Berney et al, 1998; Allgayer et al, 2000) demonstrated an association between ERBB2 overexpression and tumour-associated proteolysis in gastric and colon cancer, suggesting a direct role for ERBB2 in invasion and metastasis through upregulation of proteolytic enzymes. However, the association of ERBB2 expression with the expression of MMP-2 and MMP-9 has not previously been reported in breast cancer. The present data support the hypothesis that ERBB2 amplification and/or overexpression enhance signalling pathways that may lead to increased production of gelatinases in c-erbB-2-positive breast cancers with higher metastatic potentialities. With regard to biological significance, it appears reasonable to think that the suggested network between c-erbB-2 pathways and MMPs overexpression by in vitro models (Giunciuglio et al, 1995) may be operative also during tumour progression.