High Serum Levels of Matrix Metalloproteinase-9 and Matrix

Imaging, Diagnosis, Prognosis

High Serum Levels of Matrix Metalloproteinase-9 and Matrix Metalloproteinase-1Are Associated with Rapid Progression in Patients with Metastatic Melanoma Johanna Nikkola,1, 3 PiaVihinen,1, 3 Meri-SiskoVuoristo,4 Pirkko Kellokumpu-Lehtinen,4 Veli-Matti Ka« ha« ri,2 and Seppo Pyrho« nen1, 3

Abstract Purpose: Matrixmetalloproteinases (MMP) are proteolytic enzymes that play an important role in various aspects of cancer progression. In the present work, we have studied the prognostic significance of serum levels of gelatinase B (MMP-9), collagenase-1 (MMP-1), and collagenase- 3 (MMP-13) in patients with advanced melanoma. Experimental Design:Total pretreatment serum levels of MMP-9 in 71patients and MMP-1and MMP-13 in 48 patients were determined by an assay system based on ELISA. Total MMP levels were also assessed in eight healthy controls. The active and latent forms of MMPs were defined by usingWestern blot analysis and gelatin zymography. Results: Patients with high serum levels of MMP-9 (z376.6 ng/mL; n = 19) had significantly poorer overall survival (OS) than patients with lower serum MMP-9 levels (n =52;medianOS, 29.1versus 45.2 months; P = 0.033). High MMP-9 levels were also associated with visceral or bone metastasis (P = 0.027), elevated serum alkaline phosphatase level (P = 0.0009), and presence of liver metastases (P =0.032).SerumlevelsofMMP-1andMMP-13didnotcorrelate with OS. MMP-1and MMP-9 were found mainly in latent forms in serum, whereas the majority of MMP-13 in serum was active (48 kDa) form. MMP-13 was found more often in active form in patients (mean, 99% of the total MMP-13 level) than in controls (mean, 84% of the total MMP-13 level; P < 0.0001). After initiating the therapy, patients with elevated levels of MMP-1 (z29.8 ng/mL, n = 10) progressed more rapidly than patients with lower levels (median, 1.9 versus 3.5 months; P = 0.023). Serum levels of MMP-9 and MMP-13 did not correlate with the time to progression (TTP). In multivariate analysis with age and gender, MMP-9 or MMP-1turned out to be independent prognostic factors for OS [P = 0.039; hazard ratio (HR), 1.8; 95% confidence interval (95% CI), 1.03-3.3] or TTP (P = 0.023; HR, 2.7; 95% CI,1.15-6.4), respectively. Conclusions: Our findings provide evidence that MMP-1, MMP-9, and MMP-13 play important roles at different phases of metastatic melanoma spread and that serum MMP-9, in particular, could have clinical value in identifying patients at high risk for melanoma progression.

Matrix metalloproteinases (MMP) are a family of zinc- MMPs are known to degrade many other cell membrane dependent neutral endopeptidases characterized by their ability and pericellular proteins, including cell membrane precursor to degrade extracellular matrix components. In addition to the forms of growth factors, growth factor binding proteins, various structural components of the extracellular matrix, growth factor receptors, cell adhesion molecules, clotting factors, and proteinase inhibitors as well as their own inactive zymogen forms (1–3). The human MMP family currently Authors’ Affiliations: 1Department of Oncology and Radiotherapy, Turku consists of 23 members, which can be divided into eight University Hospital; 2Departments of Dermatology and Medical Biochemistry and structural classes or, based on their substrate specificity and 3 Molecular Biology and MediCity Research Laboratory, University of Turku,Turku, primary structure, into the more familiar subgroups of Finland and 4Department of Oncology, Tampere University Hospital, Tampere, Finland collagenases, gelatinases, stromelysins, MT-MMPs, and novel Received 12/15/04; accepted 12/15/04. MMPs. MMPs are synthesized as inactive zymogens, which are Grant support: Academy of Finland, the Finnish Cancer Foundation, Finnish activated predominantly pericellularly by other MMPs or Cultural Foundation, Finnish Life and Pension Insurance Companies, the Paulo serine proteinases. The activity of MMPs is specifically Foundation, Sigrid Juselius Foundation, Turku Finnish University Society, Turku inhibited by tissue inhibitors of metalloproteinase and by Graduate School of Clinical Science, Turku University Central Hospital, and Turku University Foundation. nonspecific proteinase inhibitors (e.g., a2-macroglobulin). The The costs of publication of this article were defrayed in part by the payment of page balance between MMPs and their inhibitors is essential in charges. This article must therefore be hereby marked advertisement in accordance many physiologic conditions where rapid remodeling of with 18 U.S.C. Section 1734 solely to indicate this fact. extracellular matrix is required and it is disturbed in Requests for reprints: Pia Vihinen, Department of Oncology and Radiotherapy, pathologic conditions such as cancer. Turku University Hospital, Kiinamyllynkatu 4-8, FIN-20520 Turku, Finland. Phone: 358-2-313-0000; Fax: 358-2-313-2809; E-mail: [email protected]. Collagenases (MMP-1, MMP-8, and MMP-13) are the F 2005 American Association for Cancer Research. principal extracellular proteinases capable of degrading native

Clin Cancer Res 2005;11(14) July 15, 2005 5158 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Serum MMP-1, MMP-9, and MMP-13 in Metastatic Melanoma

fibrillar type I, II, III, and V collagens. In addition, MMP-13 can Table 1. Characteristics of patients degrade a wide variety of other matrix substrates. After initial cleavage by collagenases, fibrillar collagens are denatured into Variable nN* gelatin and further degraded by gelatinases [i.e., gelatinase A Patients 48 23 (MMP-2) and gelatinase B (MMP-9)]. MMP-1 is expressed by Gender various types of cells in culture and in vivo (4). The expression Male 29 16 of MMP-9 is more restricted. It is produced by epithelial cells Female 19 7 such as migrating keratinocytes and is stored in the secretory Age (y) granules of neutrophils and eosinophils (5). MMP-9 also plays Median (range) 62 (25-75) 57 (36-75) an important role in angiogenesis (6). The physiologic Disease-free survival (mo) expression of MMP-13 is limited to situations where rapid Median (range) 11 (0-224) 17 (0-58) and effective remodeling of collagenous matrix is required (i.e., Overall survival (mo) fetal bone development and postnatal bone remodeling; ref. 7). Median (range) 44 (5-265) 40 (16-114) Malignant melanoma, the most aggressive of all skin tumors, Survival after the arises from melanocytes, which are normally located at the initiation of therapy (mo) basal cell layer of the epidermis. Patients with localized primary Median (range) 9 (0.3-73) 10 (2.4-64) melanoma have a favorable prognosis with cure rates up to TTP (mo) 90%. However, the prognosis in locoregional and metastatic Median (range) 2 (0.2-70) 4 (0.9-51) systemic disease is poor with a median survival of 24 and 6 c Previous treatment months, respectively (8). Because current treatment results in No treatment 16 7 advanced melanoma are not satisfactory, novel approaches are Radiotherapy 11 3 needed for therapy as well as for classification of patients into Surgery 32 15 subgroups in which certain types of therapeutic regimens IFN-a 27 would be optimal. Treatment arms The expression and significance of MMPs and their inhibitors Dacarbazine + 12 6 in melanoma has been studied extensively using melanoma cell natural IFN-a lines, mouse models, and tissue samples of patients with Dacarbazine + 11 8 primary or metastatic melanoma. The expression of MMP-1 and recombinant IFN-a MMP-13 (9), and the expression of gelatinases, in particular DOBCb +naturalIFN-a 11 5 (10–16), has been associated with melanoma progression. In DOBC + recombinant IFN-a 14 4 our previous study, we found that high expression levels of Tumor burden MMP-1 and MMP-3 in melanoma metastases correlated with Soft tissue metastases 83 shorter disease-free survival (17). We also observed that a high (skin, s.c., and lymph expression of MMP-1 was associated with favorable treatment node metastases) response (18). Lung metastases 15 2 F soft tissue metastases Serum levels of MMPs are known to be altered in many Other metastases 25 18 human pathologic conditions. Elevated serum levels of MMPs Number of metastatic sites have been reported in, for example, polycystic kidney disease One site 12 3 (19), systemic sclerosis (20), rheumatoid arthritis (21), and Two or more sites 36 20 cancer. In some diseases, such as acute viral hepatitis, however, Metastatic sites serum levels of MMPs are reduced (22). In comparison with Bone 13 6 healthy controls, the serum levels of MMP-9 are elevated in Liver 20 13 head and neck squamous cell carcinoma (23, 24) and in colon Lung 23 13 (25), gastric (26), bladder, breast, and prostate cancer (27). Skin/subcutis 16 7 Furthermore, elevated MMP-9 serum levels have been shown to Other sites 34 18 correlate with poor survival in stage I to II non–small cell lung Treatment response cancer patients (28). Complete response 60 There are few studies analyzing the serum levels of MMPs in Partial response 32 patients with metastatic melanoma and their results have been Stable disease 14 11 partly controversial. Vuoristo et al. (29) studied the serum Progressive disease 21 9 levels of MMP-2 in 50 patients with advanced melanoma and Not evaluated 41found that serum MMP-2 had not a role as a prognostic marker. On the other hand, in a study on a larger patient material, Wollina et al. (30) found that the serum levels of MMP-2 were *An additional group of 23 patients was analyzed for total MMP-9 serum levels. significantly higher in patients with metastatic melanoma than cTen patients had both radiotherapy and surgery; one patient had surgery in patients with localized melanoma. In addition, somewhat combined with IFN-a and one patient had both radiotherapy and surgery com- higher MMP-2 levels were noted in patients with more bined with IFN-a. In the additional group of 23 patients, five patients had sur- advanced disease when MMP-2 levels were compared with gery combined with IFN-a and two patients had both radiotherapy and different tumor stages. In the present work, we have determined surgery combined with IFN-a. bDacarbazine, vincristine, bleomycin, and lomustine. the serum levels of MMP-1, MMP-9, and MMP-13 in 71 patients with metastatic melanoma with a special interest to

www.aacrjournals.org 5159 Clin Cancer Res 2005;11(14) July 15, 2005 Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Imaging, Diagnosis, Prognosis determine the levels and prognostic value of their active and concentrations were determined from the corresponding latent forms. These MMPs were chosen because of their known standard curves run for each plate separately. role as prognostic factors in our earlier studies with melanoma tumors. In addition, we wanted to see whether proangiogenic Determination of active and latent forms of MMP-9 serves as a prognostic factor among these patients, matrix metalloproteinases which were all treated with IFN-a, a cytokine with antiangio- Sample preparation and positive controls. The proportions of genic properties. latent zymogen forms of MMPs and activated MMPs were determined by Western blot analysis and gelatin zymography. Patients and Methods Due to the high concentration of proteins, the serum samples were diluted in PBS (1:20). A medium known to contain the Patients studied MMPs was used as a positive control. This medium The patient population consisted of 48 patients with was obtained from normal human fibroblasts infected with advanced melanoma, who were recruited into a Finnish adenovirus to generate MMP-13 expression (31) and then randomized multicenter study between 1995 and 1998. The cultured in DMEM containing 0.5% FCS for 4 days. Another treatment arms were dacarbazine or a four-drug chemo- positive control for active MMP-1 and MMP-13 was prepared therapy composed of dacarbazine, vincristine, bleomycin, and by incubating the control medium with 1 mmol/L APMA j lomustine, both combined with either recombinant or (Sigma, St. Louis, MO) for 30 minutes at 37 C to activate natural IFN-a. There were 29 male and 19 female patients latent MMPs. A mixture of control medium and serum was and their median age was 62 years (range, 25-75). The used in a Western blot analysis to obtain a fractionation patients recruited in this study had a progressive, inoperable, similar to serum samples. The patient samples were analyzed histologically verified metastatic melanoma. Some patients in a series where one or more samples from the previous gel had previous metastases, whereas some patients entered the were run on the following gel together with some new study after the appearance of their first metastasis. Before samples. therapy, a full medical examination, including a chest X-ray, Western blot analysis. Equal aliquots of the diluted serum abdominal ultrasound and computerized tomography, a samples were fractionated on 8.5% SDS-PAGE and transferred to bone scan, and blood biochemistry, was carried out. The Hybond ECL nitrocellulose membrane (Amersham Biosciences). evaluation of metastatic sites and tumor burden is based on The amounts of MMP-1 and MMP-13 were determined using a these studies. Because of interesting results of analysis polyclonal anti-MMP-1 antibody (AB806; Chemicon Interna- concerning MMP-9, we expanded the patient population tional, Inc., Temecula, CA) and a monoclonal anti-MMP-13 analyzed for total MMP-9 with an additional group of 23 antibody (IM44L; Oncogene Research Products, Cambridge, patients. These patients had been attending the ongoing MA) in dilutions of 1:5,000 and 1:500, respectively. The multicenter study during our first-line studies. Patient specifically bound primary antibodies were detected with characteristics are summarized in Table 1. The serum samples peroxidase-conjugated secondary antibodies and visualized by were collected before initiating the chemoimmunotherapy. enhanced chemiluminescense (Amersham Biosciences). The control samples were collected among three healthy male Gelatin zymography. Aliquots of the diluted serum samples and five healthy female volunteers ages from 26 to 72 years were fractionated on 10% SDS-PAGE containing 1 mg/mL (median, 41) ranging between the minimum and maximum gelatin (G-9382; Sigma) and 0.5 mg/mL 2-methoxy-2.4- ages of those in the patient group. diphenyl-3(2H)-furanone (64958; Fluka, Buchs SG, Switzer- land). The gels were incubated for 30 minutes in 50 mmol/L Determination of serum matrix metalloproteinase-1, Tris, 0.02% NaN3, and 2.5% Triton X-100 (pH 7.5) and for 30 matrix metalloproteinase-9, and matrix minutes in the same buffer supplement with 5 mmol/L CaCl2 metalloproteinase-13 levels by ELISA and 1 Amol/L ZnCl2 (pH 7.5). Then the gels were incubated at Quantitative analysis of serum MMP-1, MMP-9, and MMp-13 37jC in 50 mmol/L Tris, 0.02% NaN3, 5 mmol/L CaCl2, and 1 was done by using a commercial Human Biotrak ELISA System Amol/L ZnCl2 for until the gelatinase bands were developed. according to the manufacturer’s instructions (Amersham After that, the gels were fixed in 50% methanol/7% acetic acid, Biosciences, Buckinghamshire, United Kingdom). The MMP-1 stained with 0.25% Coomassie brilliant blue G250, and assay recognizes total MMP-1 (i.e., proMMP-1, active MMP-1, photographed. and MMP-1/tissue inhibitor of metalloproteinase-1 complex). The MMP-9 assay is specific for proMMP-9 and MMP-9/tissue Evaluation of the active and inactive components inhibitor of metalloproteinase-1 complex, whereas the MMP-13 of matrix metalloproteinase-13 assay recognizes both proMMP-13 and active MMP-13. Briefly, The area of the gel lanes representing proMMP-13 and active all the kits are based on a two-site ELISA sandwich format. MMP-13 was measured in pixels by using a MCID-M5+ Standards and serum samples were incubated in a microwell computer program, v 4.0 (Imaging Research, Inc., St. Cathar- plate precoated with anti-MMP-1 or anti-MMP-9 or anti-MMP- ines, Ontario, Canada). A computer-driven illuminator and 13 antibody. Any MMP-1, MMP-9, or MMP-13 present in the camera transferred data from the gels. The settings were defined samples was bound to the wells, and the excess was removed by so that the same samples on different gels produced similar extensive washing. These MMPs were then detected by a results. These values in pixels were used in analysis inside the peroxidase labeled antibody, and the amount of peroxidase patient group. For each sample, the percentage of the active was determined by addition of TMB substrate. Reactions were MMP-13 from the total MMP-13 was calculated, and this stopped by adding acid solution and the absorbance was read percentage was used in comparison between the patient group at 450 nm in a microtiter plate spectrophotometer. Serum and the control group.

Clin Cancer Res 2005;11(14) July 15, 2005 5160 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Serum MMP-1, MMP-9, and MMP-13 in Metastatic Melanoma

Table 2. Serum levels of MMP-1,MMP-9, and MMP-13 in patient and control groups

Group Minimum (ng/mL) Maximum (ng/mL) Median (ng/mL) Mean (ng/mL) P* MMP-1 Patient 0 268.2 16.5 28.8 <0.0001 Control 31.5 339.5 104.2 126.5 MMP-9 Patient 48.4 1074.7 260.1 295.4 0.0024 Control 45.7 233.6 132.1 132.2 MMP-13 Patient 0.079 2.850 0.410 0.763 NS Control 0 2.178 0.417 0.646

Abbreviation: NS, not significant. *MannWhitney U test.

Statistical analysis melanoma patient and healthy control samples were measured The results from ELISA and Western blot analysis and clinical by ELISA. A statistically significant difference in MMP-9 and data were analyzed with the SAS system for Windows, v 8.2 and MMP-1 levels was noted between the patient and control the Enterprise Guide, v 2.0 (SAS Institute, Inc., Cary, NC). The groups (Table 2). Higher serum MMP-9 levels were found in Mann-Whitney U test (Kruskal-Wallis test) was used in patients than in controls (mean, 295.4 versus 132.2 ng/mL; P = assessing the associations between MMP levels and different 0.0024), whereas serum MMP-1 levels were higher in controls categorized variables such as tumor burden and in assessing the than in patients (mean, 126.5 versus 28.8 ng/mL; P < 0.0001, difference in MMP levels between patient and control groups. Mann-Whitney U test; Fig. 1; Table 2). No such differences were The two-sample t test and ANOVA were used when analyzing found in MMP-13 levels between the patient and control the association between the active MMP-13 and categorized groups (Fig. 1; Table 2). variables. Cumulative survival curves for overall survival and Matrix metalloproteinase-13 is mainly present in active form. time to progression were drawn by the Kaplan-Meier method The qualitative analysis of the serum samples was done by and the difference between the curves was analyzed by the Western blot analysis and gelatin zymography. In patient and Mantel-Cox (log-rank) test. The Cox’s multivariate analysis was control samples, MMP-13 was detected primarily as the active used to further study the effect of MMPs on survival. 48-kDa form, whereas MMP-1 and MMP-9 were primarily in latent form (i.e., as 55-kDa proMMP-1 and 92-kDa proMMP-9; Results Fig. 2A-C). However, in control samples, the latent 60-kDa form of MMP-13 was more abundant than in patient samples Serum levels of matrix metalloproteinase-1, matrix metallopro- (Fig. 2D). To compare the differences in the amounts of active teinase-9, and matrix metalloproteinase-13 in patient and control and latent MMP-13 between patient and control groups, the samples. The serum levels of MMP-1, MMP-13, and MMP-9 in percentage of the active MMP-13 from total MMP-13 was

Fig. 1. Distribution of total serum levels of MMP-1, MMP-9, or MMP-13 in patient and control groups. A, serum MMP-1levels were higher in controls (n =8)thanin patients (n = 48; mean, 126.5 versus 28.8 ng/mL; P < 0.0001). B, serum MMP-9 levels were higher in patients (n = 71) than in controls (n = 8; mean, 295.4 versus 132.2 ng/mL; P = 0.0024). C, there were no statistically significant differences in serum MMP-13 levels between patients (n = 48) and controls (n = 8; mean, 0.763 versus 0.646 ng/mL; P = 0.7967).

www.aacrjournals.org 5161 Clin Cancer Res 2005;11(14) July 15, 2005 Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Imaging, Diagnosis, Prognosis calculated. In the melanoma patient group, proportion of the Serum matrix metalloproteinases and survival. The most active MMP-13 varied from 96% to 100% with a median value of appropriate cutoff points for dividing the serum MMP levels 100% and a mean value of 99%. In the control group, the of the patients into two groups were defined by giving different percentages varied from 65% to 98% with a median and mean values for the cutoff point and determining the corresponding value of 84%. The proportion of the active MMP-13 was P values using the Mantel-Cox (log-rank) test to analyze significantly higher in the patient group than in the control Kaplan-Meier survival curves. This method has been previously group (mean, 99% versus 84%; P < 0.0001, Mann-Whitney U described (32). The cutoff point that corresponded to the test). In addition, the actual levels of the active MMP-13 were lowest P value was used in the analysis (Fig. 3). This cutoff determined from the total MMP-13 levels, measured by ELISA, point was z29.8 ng/mL for MMP-1, z376.6 ng/mL for MMP-9, based on the percentages obtained with the Western blot and z1.508 ng/mL for MMP-13. The pixel values for the active analysis. The levels of active MMP-13 in melanoma patients MMP-13 were divided similarly and this cutoff point was z421 varied from 0.079 to 2.850 ng/mL with a median value of 0.410 pixels. The cutoff points from this study are planned to be used ng/mL and a mean value of 0.756 ng/mL. In controls, the levels in further prospective studies. Corresponding patient numbers of active MMP-13 varied from 0 to 1.742 ng/mL with median in each patient group are shown in Table 3. and mean values of 0.300 and 0.530 ng/mL, respectively. We found that patients with high serum MMP-9 levels (z376.6 ng/mL, n = 19) had a poorer overall survival (OS) with a median of 29.1 months when compared with patients with lower MMP-9 levels (n = 52; median OS, 45.2 months; P = 0.033, log-rank test; Fig. 4A; Table 3). In multivariate analysis including age and gender, MMP-9 turned out to be an independent prognostic factor for OS (P = 0.039; hazard ratio, 1.8; 95% confidence interval, 1.03-3.3). The serum levels of MMP-1 or MMP-13 or the active MMP-13 did not correlate with OS (Table 3). The time to progression (TTP) was calculated from the beginning of the chemoimmunotherapy to the diagnosis of progressive disease. We found that patients with higher levels of serum MMP-1 (z29.8 ng/mL, n = 10) had significantly shorter TTP than patients with lower MMP-1 levels (n = 38; median, 1.9 versus 3.5 months; P = 0.023, log-rank test; Fig. 4B; Table 3). In multivariate analysis with age and gender, MMP-1 was an independent prognostic factor for TTP (P = 0.023; hazard ratio, 2.7; 95% confidence interval, 1.15-6.4). The serum levels of MMP-9, proMMP-13, or the active MMP-13 did not correlate with TTP, and none of the studied MMPs correlated with the survival after the initiation of chemoimmunotherapy (Table 3). Serum matrix metalloproteinases and tumor site. Site of metastasis has a major effect on survival of the metastatic melanoma patients. We used a tumor site classification resembling the M1a-c classification in the new American Joint Committee on Cancer melanoma staging system (8). The patients were divided into three groups according to the sites of their tumors. The first group included patients with soft tissue metastases (i.e., skin, s.c., or lymph node metastases), the second group included patients with lung metastases with or without soft tissue involvement, and the third group included patients with other visceral or bone metastases. The serum MMP-9 levels were found to correlate with tumor site. Fig. 2. Qualitative analysis of serum MMP-1,MMP-9, and MMP-13. A, serum samples were diluted with PBS and analyzed byWestern blot technique. MMP-1was The mean MMP-9 level was 217.6 ng/mL in the first group found exclusively in its 55-kDa proMMP form in patients’ (lanes1-3)andinhealthy (n = 11), 275.4 ng/mL in the second group (n = 17), and persons’ (lanes 4-10) samples. A medium from normal human fibroblasts infected 323.3 ng/mL in the third group (n = 43; P = 0.027, Kruskal- with adenovirus to generate MMP-13expression was used as a positive control.This medium was known to contain MMP-1and MMP-9, as well. Control 1,untreated Wallis test). The levels of MMP-1 (P = 0.63) and MMP-13 positive control; Control 2, positive control activated byAPMA showing also the (P = 0.55; Kruskal-Wallis test) or the active MMP-13 (P = 0.57, active 43-kDa MMP-1. Molecular weight markers in kDa (left). B-C ,MMP-9was ANOVA) did not correlate with tumor site. detected by gelatin zymography and found primarily in its 92-kDa proMMP form in patients’ (top) and in healthy persons’ (bottom) samples.The 72-kDa marker shows In addition, we studied whether serum MMP levels were the position of gelatinaseA (MMP-2). Small quantities of the active 82-kDa MMP-9 elevated in patients with specific metastatic sites (bone, liver, were also detected.The arrows point at proMMP-9 (lane 1), the 82-kDa MMP-9 (lane 2), and proMMP-2 (lane 3). D, MMP-13was detected byWestern blot analysis lung, and skin/subcutis) or whether the levels correlated with and found primarily in its active 48-kDa form. Control 1and Control 2, positive the number of metastases (one or more). The serum levels of controls, of which Control 2 has been activated by APMA (middle). So me of th e MMP-9 were found to be significantly higher in patients with patients’samples (top) contained small quantities of the 60-kDa proMMP-13, but larger quantities of proMMP-13 were detected in the healthy persons’samples liver metastases (n = 33) than in patients without them (n = 38; (bottom). Molecular weight markers in kDa (left). mean, 331.2 versus 264.4 ng/mL; P = 0.032, Mann-Whitney

Clin Cancer Res 2005;11(14) July 15, 2005 5162 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Serum MMP-1, MMP-9, and MMP-13 in Metastatic Melanoma

U test). The levels of MMP-1 (P = 0.71), MMP-13 (P = 0.42, Whitney U test). Serum alkaline phosphatase levels did not Mann-Whitney U test) or the active MMP-13 (P = 0.83, correlate with the levels of active (P = 0.7, two-sample t test) ANOVA) did not correlate with the presence of liver metastases. or total MMP-13 (P = 0.2, Mann-Whitney U test). The serum The serum levels of MMPs did not correlate with other levels of alanine aminotransferase did not associate with MMP metastatic sites or with the number of metastases (data not serum levels (data not shown). shown). Serum matrix metalloproteinases and response to therapy or Serum matrix metalloproteinase levels and the level of serum characteristics of the primary tumor. The WHO criteria were alkaline phosphatase. Theroutinemeasurementsofthe used for evaluation of response. There was no association patients’ serum levels of alkaline phosphatase and alanine between the serum MMPs and response to therapy neither in aminotransferase were included in the study. Serum alanine the whole patient group nor in the subgroups of patients with aminotransferase is a quite specific marker for liver failure, different therapy. The serum levels of the MMPs did not whereas serum alkaline phosphatase level is elevated in correlate with prognostic features of the primary tumor disease processes of the bone. In melanoma patients, the including tumor thickness, anatomic site and growth pattern elevated levels of these serum markers can reflect the presence (data not shown). of liver or bone metastases. The patients were divided into two groups with normal or elevated serum alkaline phosphatase Discussion and alanine aminotransferase levels according to the upper limit of the standard laboratory reference value. The patients During the recent years, the important role of MMPs in with elevated S-alp levels (n = 10) had higher serum MMP-9 various stages of cancer progression has been extensively levels than those patients with normal S-alp levels (n = 59; documented. Inhibition of the activity of MMPs by synthetic mean, 469.7 versus 266.0 ng/mL; P = 0.0009, Mann-Whitney inhibitors has been a recent approach in the treatment of U test; S-alp values were missing from two patients). cancer. Small molecule MMP inhibitors (MMPI) BMS-275291, Conversely, patients with elevated S-alp levels (n = 9) had marimastat (BB-2516), neovastat, prinomastat (AG3340), and lower serum levels of MMP-1 than patients with normal S-alp metastat have been studied in advanced phase clinical trials levels (n = 39; mean, 13.5 versus 32.4 ng/mL; P = 0.028, Mann- for the treatment of various types of cancers (1–3). Recently,

Fig. 3. The distribution of P values from Mantel-Cox(log-rank) test for Kaplan-Meier survival curves associated with different cutoff values.The most appropriate cutoff values corresponded with the lowest P values and were (A) z29.8 ng/mL for MMP-1,(B) z376.6 ng/mL for MMP-9, (C) z1.508 ng/mL for total MMP-13, and (D) z421 pixels for the active 48-kDa MMP-13.

www.aacrjournals.org 5163 Clin Cancer Res 2005;11(14) July 15, 2005 Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Table 3. Survival analysis

OS (median in mo) Survival after therapy (median in mo) TTP* (median in mo) MMP-1 <29.8 ng/mL (n = 38) 44.3 9.0 3.5 z29.8 ng/mL (n = 10) 46.2 6.8 1.9 P NS NS 0.023c MMP-9 <376.6 ng/mL (n = 52) 45.2 10.0 3.4 z376.6 ng/mL (n =19) 29.1 7.2 2.4 P 0.033 NS NS MMP-13 <1.508 ng/mL (n = 39) 43.9 9.0 3.7 z1.508 ng/mL (n =9) 49.2 6.7 1.9 P NS NS NS 48-kDa MMP-13 <421pixels (n =10) 52.7 10.4 5.9 z421pixels (n = 38) 42.7 8.4 2.0 P NS NS NS

Abbreviation: NS, not significant. *TTP after the initiation of therapy. cDifference between the groups, log-rank test. some encouraging results from trials with synthetic MMPIs In the present work, we have studied the levels of MMP-1, have been reported. In a randomized trial of 369 patients with MMP-9, and MMP-13 and the proportions of their active and advanced gastric cancer, patients who received marimastat inactive forms in sera of patients with metastatic melanoma. gained some survival benefit when compared with patients Elevated serum MMP-9 levels were noted in melanoma patients who received placebo (33). Furthermore, treatment of patients compared with healthy control subjects. In addition, high (n = 44) with recurrent and progressive glioblastoma multi- serum levels of MMP-9 were found to be associated with poor forme with a combination of marimastat and an alkylating overall survival. After standardizing for the age and gender, agent temozolomide resulted in increased progression-free MMP-9 turned out to be an independent prognostic factor for survival at the time point of 6 months (34). In a small group OS in metastatic melanoma. Elevated MMP-9 levels were found of patients with advanced renal cell carcinoma, treatment with to correlate with multiple tumor sites, liver metastases, and a higher dose of neovastat was associated with longer survival elevated levels of serum alkaline phosphatase, all suggesting than treatment with a lower dose (35). that MMP-9 an important role in rapid progression of mela- The early clinical trials that failed to show any benefit from noma. These observations suggest that measurement of serum MMPI treatment have been criticized because patients with MMP-9 of melanoma patients could aid in detection of advanced cancers were studied without information on the melanoma progression. The purpose of this study was also to specific expression profiles of MMPs in these cancer types (1). find out clinically meaningful cutoff points for MMP-9 For example, it has been shown that the expression of MMP-11 expression levels to frame further prospective studies. In and MMP-14 are indicators of poor prognosis in small cell lung addition, these results suggest MMP-9 as a therapeutic target cancer, whereas the expression of MMP-2 is undetectable in this in melanoma patients. tumor type (36). This information could have guided the MMP-13 is characterized by wide substrate specificity and selection of a more appropriate tumor type for testing of the restricted expression. In our previous studies, we have found synthetic MMPI tanomastat (BAY 12-9566), which is specifically that it was expressed tumor cells in human melanoma targeted against MMP-2 and has minimal inhibitory effect on metastases (17, 18). In the present study, we detected MMP- MMP-11. Furthermore, different MMPs are expressed in different 13 in the serum of patients with metastatic melanoma but also stages of cancer progression. MMP-2, MMP-9, and MMP-14 have in the serum of healthy controls. The levels of MMP-13 in been shown to be involved in regulation of angiogenesis by patient serum samples were low when compared with those of different mechanisms (1). Thus, inhibition of their activity to MMP-1 or MMP-9, and they did not differ from those in the prevent the formation of new blood circulation into the tumor control group. However, the MMP-13 in patient sera was should take place before development of an independent blood found almost exclusively in the active 48-kDa form, whereas vessel network. Therefore, more information is required about the sera of control subjects contained inactive MMP-13 as the expression of different MMPs in various tumor types and in well. different stages of the tumors. The recent promising results of In general, in previous studies, higher levels of MMPs have clinical trials with MMPIs suggest that through careful patient been detected in sera of patients than in controls. However, selection and study design, the MMPIs could ultimately be reduced levels have been reported in acute viral hepatitis (22) potential agents in the therapy of cancer. and in postoperative biliary atresia (37). In the present study,

Clin Cancer Res 2005;11(14) July 15, 2005 5164 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Serum MMP-1, MMP-9, and MMP-13 in Metastatic Melanoma

of MMP-1 progressed more rapidly after therapy, and MMP-1 turned out to be an independent prognostic indicator for TTP in multivariate analysis. The role of MMP-1 in melanoma progression seems controversial. It has been shown to be associated with advanced stages of malignant melanoma (9) and to correlate with shorter disease-free survival (17), but on the other hand, to be associated with favorable chemoimmunotherapy response (18) in stage IV melanoma. The results of the current work indicate that MMP-1 could have an important role in the maintenance of normal tissues; but on the other hand, its abundant expression can facilitate rapid disease progression in metastasized melanoma. Several studies have been done to identify different easily assayed quantitative markers that could serve as prognostic indicators for patients with malignant melanoma. One of the most established is the level of the enzyme lactic dehydrogenase assessed as a marker for metastatic disease (38). Serum LHD levels have since been included in the new American Joint Committee on Cancer melanoma staging system, where stage IV melanoma patients with elevated serum lactate dehydrogenase levels are classified into the M1c group with dismal prognosis (8). Another well-documented marker for prognosis is the serum polypeptide S-100h. Patients positive for S-100h have been shown to have significantly shorter disease-free survival than patients negative for S-100h (38). However, there is a need for new easily measurable markers, which could predict disease progression or therapeutic response. In the present study, we have assessed the prognostic value of serum MMP-1, MMP-9, and MMP-13 in stage IV melanoma patients and identified MMP-9 as an independent prognostic factor for OS in metastatic melanoma. Due to the limited number of controls and patients in this study, further studies are obviously required. In addition, studies with serum samples obtained at different phases of melanoma progression could further elucidate the role of MMPs in growth and metastasis of melanoma. In conclusion, ourfindingsprovideevidence,thatMMP-1,MMP-9,and Fig. 4. Serum MMPs and survival. A, OS from diagnosis of primary melanoma in MMP-13 have important roles at different phases of metastatic patients with lower levels (<376.6 ng/mL) of MMP-9 (black line, n =52)andwith high levels (z376.6 ng/mL) of MMP-9 (gray line, n =19;P =0.033).B,TTP spread and that measurement of serum MMP-9, in particular, evaluated from the initiation of therapy to the diagnosis of progressive disease in could be of clinical value when identifying patients at high patients with lower levels (<29.8 ng/mL) of MMP-1 (black line, n =38)andwith risk for progression. higher levels (z29.8 ng/mL) of MMP-1 (gray line, n =10;P = 0.023). we found that the MMP-1 levels were generally lower in Acknowledgments melanoma patients. In addition, lower levels of MMP-1 were We thank Mervi Toriseva for technical guidance and for providing the positive associated with an elevated serum alkaline phosphatase level. control medium, Sari Pitka« nen for collecting the samples of the control group, and On the other hand, the disease of patients with elevated levels TeroVahlberg for help with statistics.

References 1. Coussens LM, Fingleton B, Matrisian LM. Matrixmet- phils in squamous cell carcinoma. Am J Pathol 1993; collagenases-1and -3 and their inhibitorsTIMP-1and - alloproteinase inhibitors and cancer: trials and tribula- 142:995^1000. 3 correlates with the level of invasion in malignant mel- tions. Science 2002;295:2387 ^ 92. 6. Johnson C, Sung HJ, Lessner SM, Fini ME, Galis ZS. anomas. BrJ Cancer 1999;80:733 ^ 43. 2. Egeblad M, Werb Z. New functions for the matrix Matrixmetalloproteinase-9 is required for adequate an- 10. Va« isa« nen A, Tuominen H, Kallioinen M, Turpeen- metalloproteinases in cancer progression. Nat Rev giogenic revascularization of ischemic tissues: potential niemi-Hujanen T. Matrixmetalloproteinase-2 (72kD Cancer 2002;2:161 ^ 74. role in capillary branching. Circ Res 2004;94:262^8. type IV collagenase) expression occurs in the early 3. Vihinen P, Ka« ha« ri V-M. Matrixmetalloproteinases in 7. Johansson N, Saarialho-Kere U, Airola K, et al. stage of human melanocytic tumour progression and cancer: prognostic markers and therapeutic targets. Collagenase-3 (MMP-13) is expressed by hypertro- may have prognostic value. JPathol1996;180:283 ^ 9. Int J Cancer 2002;99:157 ^ 66. phic chondrocytes, periosteal cells and osteoblasts 11. Va« isa« nen A, Kallioinen M,Taskinen PJ,Turpeenniemi- 4. Ka« ha« ri V-M, Saarialho-Kere U. Matrixmetalloprotei- during human fetal bone development. Dev Dyn 1997; Hujanen T. Prognostic value of MMP-2 immunoreac- nases and their inhibitors in tumor growth and inva- 208:387^97. tive protein (72kD type IVcollagenase) in primary skin sion. Ann Med1999;31:34^45. 8. Kim CJ, Reintgen DS, Balch CM. The new melanoma melanoma. J Pathol 1998;186:51 ^ 8. 5. Sta« hle-Ba« ckdahl M, ParksWC. 92-kd gelatinase is ac- staging system. Cancer Control 2002;9:9 ^15. 12. Hofmann UB,WestphalJR, Zendman AJ, BeckerJC, tively expressed by eosinophils and stored by neutro- 9. Airola K, KaronenT,Vaalamo M, et al. Expression of Ruiter DJ, van Muijen GN. Expression and activation

www.aacrjournals.org 5165 Clin Cancer Res 2005;11(14) July 15, 2005 Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

of matrixmetalloproteinase-2 (MMP-2) and its co- sclerosis, primary Raynaud’s phenomenon, and in nor- 30. Wollina U, Hipler UC, Knoll B, Graefe T, Kaatz M, localization with membrane-type 1 matrixmetallo- mal controls. Ann Rheum Dis 2001;60:846 ^ 51. Kirsch K. Serum matrixmetalloproteinase-2 in patients proteinase (MT1-MMP) correlate with melanoma 21. Klimiuk PA, Sierakowski S, Latosiewicz R, Cylwik B, with malignant melanoma. J Cancer Res Clin Oncol progression. JPathol 2000;191:245 ^ 56. Skowronski J, Chwiecko J. Serum matrixmetallo- 2001;127:631 ^ 5. 13. Kurschat P,Wickenhauser C, GrothW, KriegT,Mauch proteinases and tissue inhibitors of metalloprotei- 31. Ala-aho R, Johansson N, Baker AH, Ka« ha« riV-M. Ex- C. Identification of activated matrixmetalloproteinases in different histological variants of rheumatoid pression of collagenase-3 (MMP-13) enhances invasion nase-2 (MMP-2) as the main gelatinolytic enzyme in synovitis. Rheumatology 2002;41:78 ^ 87. of human fibrosarcoma HT-1080 cells. Int J Cancer malignant melanoma by in situ zymography. J Pathol 22. Koulentaki M,Valatas V, Xidakis K, et al. Matrixmet- 2002;97:283 ^ 9. 2002;197:179 ^ 87. alloproteinases and their inhibitors in acute viral hepa- 32. Kronqvist P, KuopioT, CollanY.Morphometric grad- 14. Itoh T,Tanioka M, Yoshida H, Yoshioka T, Nishimoto titis. J Viral Hepat 2002;9:189 ^ 93. ing of invasive ductal breast cancer. I. Thresholds for H, Itohara S. Reduced angiogenesis and tumor pro- 23. Riedel F, Gotte K, Schwalb J, Hormann K. Serum nuclear grade. BrJ Cancer1998;78:800 ^ 5. gression in gelatinase A-deficient mice. Cancer Res levels of matrixmetalloproteinase-2 and -9 in patients 33. Bramhall SR, Hallissey MT,Whiting J, et al. Marima- 1998;58:1048^51. with head and neck squamous cell carcinoma. Anti- stat as maintenance therapy for patients with ad- 15. van den Oord JJ, Paemen L, Opdenakker G, de cancer Res 2000;20:3045 ^ 9. vanced gastric cancer: a randomized trial. Br J Cancer Wolf-Peeters C. Expression of gelatinase B and the 24. KuropkatC,PlehnS,HerzU,DunneAA,RenzH, 2002;86:1864 ^ 70. extracellular matrix metalloproteinase inducer EMM- Werner JA. Tumor marker potential of serum matrix 34. Groves MD, Puduvalli VK, Hess KR, et al. Phase II PRIN in benign and malignant pigment cell lesions of metalloproteinases in patients with head and neck trial of temozolomide plus the matrixmetalloprotei- the skin. Am J Pathol 1997;151:665^70. cancer. Anticancer Res 2002;22:2221 ^ 7. nase inhibitor, marimastat, in recurrent and progres- 16. Itoh T,Tanioka M, Matsuda H, et al. Experimental 25. Pucci-Minafra I, Minafra S, La Rocca G, et al. Zymo- sive glioblastoma multiforme. J Clin Oncol 2002;20: metastasis is suppressed in MMP-9-deficient mice. graphic analysis of circulating and tissue forms of co- 1383^8. Clin Exp Metastasis 1999;17:177 ^ 81. lon carcinoma gelatinase A (MMP-2) and B (MMP-9) 35. Batist G, Patenaude F, Champagne P, et al. Neova- 17. Nikkola J,Vihinen P,Vlaykova T, Hahka-Kemppinen separated by mono- and two-dimensional electropho- stat (AE-941) in refractory renal cell carcinoma M, Ka« ha« ri V-M, Pyrho« nen S. High expression levels of resis. MatrixBiol 2001;20:419^ 27. patients: report of a phase II trial with two dose levels. collagenase-1and stromelysin-1correlate with shorter 26. Endo K, MaeharaY, Baba H, et al. Elevated levels of Ann Oncol 2002;13:1259 ^ 63. disease-free survival in human metastatic melanoma. serum and plasma metalloproteinases in patients with 36. Michael M, Babic B, Khokha R, et al. Expression and Int J Cancer 2002;97:432^ 8. gastric cancer. Anticancer Res 1997;17:2253 ^ 8. prognostic significance of metalloproteinases and 18. Nikkola J,Vihinen P,VlaykovaT, Hahka-Kemppinen 27. Zucker S, Hymowitz M, Conner C, et al. Measure- their tissue inhibitors in patients with small-cell lung M, Ka« ha« ri V-M, Pyrho« nen S. High collagenase-1 ex- ment of matrixmetalloproteinases and tissue inhibitors cancer. J Clin Oncol 1999;17:1802^ 8. pression correlates with a favourable chemoimmuno- of metalloproteinases in blood and tissues. Clinical 37. KobayashiH,LiZX,YamatakaA,LaneGJ,Miyano therapy response in human metastatic melanoma. and experimental applications. Ann N Y Acad Sci T. Clinical evaluation of serum levels of matrixmetal- Melanoma Res 2001;11:157 ^ 66. 1999;878:212^27. loproteinases and tissue inhibitors of metallopro- 19. Nakamura T, Ushiyama C, Suzuki S, Ebihara I, 28. Laack E, Kohler A, Kugler C, et al. Pretreatment se- teinases as predictors of progressive fibrosis in Shimada N, Koide H. Elevation of serum levels of met- rum levels of matrixmetalloproteinase-9 and vascular postoperative biliary atresia patients. J Pediatr Surg alloproteinase-1, tissue inhibitor of metalloproteinase- endothelial growth factor in non-small-cell lung can- 2002;37:1030 ^ 3. 1and type IV collagen, and plasma levels of metallo- cer. Ann Oncol 2002;13:1550 ^ 7. 38. Lotze MT, Dallal RM, Kirkwood JM, Flickinger JC. proteinase-9 in polycystic kidney disease. AmJ Neph- 29.Vuoristo MS, Kellokumpu-Lehtinen P, Parvinen LM, Cutaneous melanoma. In: DeVita VT, Hellman S, rol 2000;20:32 ^ 6. et al. Serum matrixmetalloproteinase-2 as a prognos- Rosenberg SA, editors. Cancer: principles and practice 20. Young-Min SA, Beeton C, Laughton R, et al. Serum tic marker in advanced cutaneous melanoma. Acta of oncology, 6th ed. Vol. 2. Philadelphia: Lippincott TIMP-1,TIMP-2 and MMP-1in patients with systemic Oncol 2000;39:877 ^ 9. Williams & Wilkins; 2001.p. 2012^ 69.

Clin Cancer Res 2005;11(14) July 15, 2005 5166 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2005 American Association for Cancer Research. High Serum Levels of Matrix Metalloproteinase-9 and Matrix Metalloproteinase-1 Are Associated with Rapid Progression in Patients with Metastatic Melanoma

Johanna Nikkola, Pia Vihinen, Meri-Sisko Vuoristo, et al.

Clin Cancer Res 2005;11:5158-5166.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/14/5158

Cited articles This article cites 37 articles, 6 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/14/5158.full#ref-list-1

Citing articles This article has been cited by 17 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/14/5158.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/11/14/5158. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.