Circulating gelatinases and tissue inhibitor of metalloproteinase-1 in colorectal cancer metastatic liver disease

Methods Fifty-seven patients undergoing liver metastasis operation were followed prospectively. ProMMP-2, -9 and TIMP-1 plasma levels were determined by zymography and ELISA, before and after the resection of liver metastases. Data were compared with those of healthy controls ( n = 51) and primary CRC patients ( n = 94). The diagnostic and prognostic potential was investigated with ROC-curves and Kaplan–Meier survival analysis. Results Plasma proMMP-2 levels were lower ( P < 0.001), and TIMP-1 levels higher ( P < 0.001) in CRC metastatic liver disease than in healthy controls. If compared to those in primary CRC patients, no differences were found. In ROC-curves, the area under the curve was 0.48 and 0.61 for proMMP-2 and -9, respectively. Plasma proMMP-2, -9 and TIMP-1 levels were unsuitable to predict survival. In both diagnostic and prognostic examinations, CEA proved to be a better marker. In the postoperative follow-up, protracted low levels of proMMP-2 seemed related to disease recurrence. Conclusion The preoperative plasma proMMP-2, -9 and TIMP-1 levels have no potential value as diagnostic or prognostic markers in CRC liver metastatic disease. Keywords Blood Colorectal cancer Liver Marker Matrix metalloproteinase Introduction Patients with untreated hepatic metastases of colorectal cancer (CRC) are unlikely to survive longer than five years. 1,2 Systemic and regional chemotherapy can prolong the median survival time 3 but resection of the metastasis is ultimately the only treatment option with a substantial chance for cure. Unfortunately, within two years after the resection, two-thirds of the patients develop recurrent disease. 2,4 In order to choose the appropriate treatment for metastatic disease after CRC discriminative tools are needed. A proven impossibility to resect all tumour and extra-hepatic disease constitutes the only firm contra-indication to hepatic resection. 1 Currently, patient prognosis and patient selection for treatment are based largely on factors such as the size and number of liver tumours, the involvement of lymph nodes at the time of primary disease, the disease-free interval between colon and liver disease and the preoperative carcinoembryonic antigen (CEA) level. 1,5 While increased serum CEA is convincingly linked to the presence of liver metastasis, it appears questionable if it correlates with prognosis. 6 Also, some patients with CRC and CRC liver metastases retain normal CEA levels. Therefore, the development of more powerful diagnostic and prognostic markers remains highly necessary. The degradation of the extracellular matrix is intrinsic to the invasion and progression of cancer. 7 The matrix metalloproteinases (MMP) appear to be the terminal effectors of the proteolytic cascade in matrix degradation. 8,9 Both the gelatinases MMP-2 and MMP-9 can degrade collagen IV, the major structural component of basement membranes. The tissue inhibitors of matrix metalloproteinase-1 and -2 (TIMPs) are the natural inhibitors of MMP-9 and MMP-2, respectively, and are part of a family of homologous proteins that potentially may act anti-metastatically. There is considerable evidence for the involvement of MMPs in colorectal cancer. 10 For instance, active and latent MMP-2 tissue levels are higher in tumour than in normal tissues. 11,12 Increased tissue levels of proMMP-9 have been found early in the transition from colon adenoma to adenocarcinoma 13 and enhanced tissue levels of TIMP-1 correlate with poor prognosis in CRC. 10 Consequently, gelatinases and their inhibitors represent logical candidates in the search for diagnostic and prognostic markers in CRC-related disease. In contrast to tissue, little is known about MMPs in the blood of CRC patients. So far, there are no data on circulating gelatinase and TIMP levels measured at the time of resection of metastatic disease. This report aims to examine the possible diagnostic and prognostic value of blood MMP-2, -9 and TIMP-1 levels, as measured before and after potential curative hepatic resection for CRC metastases. Since CEA is currently used as a marker of tumour burden in clinical practice, 1,5,14 it is used as a reference marker. Patients and methods Patients and control subjects From September 30, 1997 until October 1, 2000, 57 patients (37 male) with CRC liver metastases were included in this prospective study. In all, the primary tumour had been resected previously. The median age at the time of surgery was 59 years (range 35–78). All patients went into surgery with curative intent and some were also treated with radiofrequency ablation or cryotherapy. Still, at laparotomy the disease in 20 patients proved to be too extensive to proceed with further surgery or other treatment. Thirty-seven patients were thus considered potentially cured by the operation. The median follow-up after operation was 37 months (range 1–68 months). Twenty-nine of the 37 patients developed recurrent disease. Twenty-one patients had recurrences that included the liver, four patients had metastasis to the lung, brain and/or skin and four had abdominal recurrences. The median time to disease relapse was 8 months (range 1–59). The three-year disease-related survival rate was 49% for the total group and 61% for the 37 patients who underwent potentially curative operation. Ninety-four patients (57 male) with primary CRC were included as a reference group. Their median age was 63 years (range 28–86). Data on these patients were obtained from a previous study. 15 Additionally, a control group of 51 healthy blood donors (26 male) was included. Their median age was 55 years (range 45–69). Blood collection and analysis Venous blood samples were obtained with informed consent from all individuals and with permission of the institutional ethics committee. Blood samples of all CRC patients were drawn preoperatively. Postoperative sampling was conducted in the CRC liver metastasis patients in the first month after operation and then approximately every 3 months. If possible, sufficient blood was collected for processing into multiple samples of serum and both heparin- and EDTA-plasma. Both plasma and serum were obtained after centrifugation at 1500 g for 10 min and were stored at −80°C prior to assay. Zymography was performed as described previously. 11,16 Four microliters of heparin plasma was diluted in 16 μl Tris-buffer (50 mM Tris, 0.2 mM NaCl, 5 mM CaCl 2 , pH 7.5) and 20 μl sample buffer (0.125 M Tris–HCl, pH 6.8, 17.4% (w/v) glycerol, 4% sodium dodecylsulphate and 0.01% bromophenol blue). Five microliters of this solution was loaded onto the sodium dodecylsulphate-polyacrylamide gel. Gel electrophoresis was performed using a mini-gel system apparatus (Bio-Rad, USA). Gelatinolytic activity is expressed in arbitrary units (AU) per ml. Plasma proMMP-2, -9 and TIMP-1 immunoreactivity was measured with commercially available, two-step, sandwich BIOTRAK™ ELISA kits (Amersham Pharmacia Biotech, Uppsala, Sweden). The assays were performed in accordance with the manufacturer's instructions. For the MMP-2 ELISA, heparin plasma was diluted 1:50 with assay buffer. For the MMP-9 ELISA, EDTA-plasma was diluted 1:10 with assay buffer. For the TIMP-1 ELISA, heparin plasma was diluted 1:40 with assay buffer. The latter assay recognizes both free TIMP-1 and that complexed with MMPs. The sensitivities of these assays were 0.37 ng/ml for MMP-2, 0.6 ng/ml for MMP-9 and 1.25 ng/ml for TIMP-1. Serum CEA levels were measured by enzyme immunoassay (Axsym; Abbott Laboratories, Chicago, IL, USA). The normal range for CEA in healthy controls is between 0.5 and 5 ng/ml. In order to prevent confusion, data concerning the proMMP blood levels obtained by zymography and ELISA are referred to as (Z) and (E) data, respectively. The specificity and sensitivity of preoperative MMP and CEA for diagnosing liver metastatic disease The ability to distinguish CRC patients with hepatic metastatic disease from those without metastatic disease, based on proMMP(Z) or CEA blood levels, was investigated. Fifty-seven patients with hepatic metastases were compared to 34 patients from the reference group who previously underwent resection of primary CRC and were at least two years disease-free. From the latter group the necessary MMP(E) data were only available for 6 patients, which were insufficient to perform an analysis. The median time of sampling of the preoperative blood samples from the hepatic metastasis group was 9 months after primary CRC operation. The samples from the reference group were collected within a similar period. The specificity and sensitivity, expressed in the area under the curve (AUC), determines the potential value of the diagnostic test and is best illustrated by a receiver operating characteristic (ROC) curve. Preoperative MMP and CEA and their relation to survival The correlation of proMMP-2, -9, TIMP-1 and CEA levels with the length of disease-specific survival in patients who were potentially cured after resection was investigated. The survival analysis was performed with the Kaplan–Meier method. Five of the 37 patients died of unknown causes and were excluded from analysis. Patients who were alive and one patient with a non-CRC death were regarded as censored observations at the end of their follow-up. Survival distributions in patients with high and low proMMP or TIMP-1 levels were compared by stratifying patient data into two groups with the median being the cut off value. The cut off value for CEA was set at 5 ng/ml. The effects of liver surgery on MMP and CEA levels The percentage change between pre- and postoperative levels of proMMP(Z) and CEA were determined in order to evaluate the effect of CRC liver metastasis operation on blood levels in those patients who were potentially cured after operation ( n = 37). In the postoperative period investigated here (30–60 days after operation) ELISA data on MMPs and TIMP-1 were incomplete and their numbers insufficient to present a meaningful analysis. ProMMP(Z) and CEA comparisons involved 16 and 33 evaluable cases, respectively. ProMMP(Z) data are based on the comparison of the preoperative levels with those measured between 30 and 60 days after operation, since earlier levels may be affected by the healing response. 17 The CEA results are based on the comparison of preoperative levels with those measured within 1 month after operation. Changes in MMP and CEA prior to disease relapse After liver surgery, the changes in the blood levels of proMMP-2, -9, TIMP-1 and CEA were followed. The reported changes were calculated from the difference between the last two consecutive samples taken before the discovery of disease relapse. The samples used were all obtained within the year preceding the discovery of disease relapse. The first sample was taken at least 1 month after surgery and there was a period of at least 30 days between the second last and the last sample; the time between the last sample and the discovery of disease relapse was never longer than 6 months. On the basis of these requirements 20, 18 and 24 patients yielded data on proMMP(Z), proMMP(E) and CEA, respectively. The percentage change in the relapse group was compared to the changes found in the blood samples of patients who did not relapse within two years after operation ( n = 8). The mean time of sampling after liver surgery was approximately 8 months and was similar in both groups. Statistical analysis The Mann–Whitney U -test was used to test for differences between two groups. For paired testing, the Wilcoxon signed ranks test was used. The one-way analysis of variance and the Kruskal–Wallis test were used to test for differences between more than two groups. In-between comparisons were performed using the Tukey–Kramer multiple-comparisons test, Dunn's multiple-comparisons test or the Mann–Whitney U -test with an adjusted P -value ( P < 0.05/ k , with k being the number of comparisons made). The sensitivity and specificity of proMMP, TIMP-1 and CEA data in diagnosing CRC liver metastatic disease were illustrated in ROC-curves. The 95% confidence intervals (CI) for the sensitivities at 95% specificity were calculated by the Fisher's exact test. Differences between AUCs were estimated with the method of DeLong et al. 18 The Kaplan–Meier method was used in survival analysis. Equality between groups was determined with the log-rank test. All differences were considered statistically significant at the P < 0.05 level. Results Preoperative MMP, TIMP-1 and CEA levels In both patients and controls, zymographic analysis of plasma samples demonstrated the presence of measurable levels of the latent forms of MMP-2 and -9, while the active forms were absent. Thus, the zymographic data reflect levels of free circulating proMMPs plus those previously bound to TIMPs. In preoperative plasma of CRC liver metastasis patients, both zymography and ELISA demonstrated significantly ( P < 0.001) lower levels of proMMP-2 and equal levels of proMMP-9 if compared to healthy controls. Plasma TIMP-1 and serum CEA levels were significantly (both P < 0.001) higher than in controls. In comparison to preoperative plasma levels from primary CRC patients, similar levels of proMMP-2, -9 and TIMP-1 were found. The serum CEA levels were significantly ( P < 0.001) higher in patients before liver metastasis operation than before primary CRC operation ( Table 1 ). Preoperative proMMP(Z) and CEA as a diagnostic tool A ROC curve, plotting the specificity against the sensitivity, is shown in Fig. 1 . At 95% specificity (i.e. maximum of 5% false positives), zymographic data yielded very low sensitivities of 11% (95% CI: 4–23%) and 21% (95% CI: 11–34%) for proMMP-2(Z) and proMMP-9(Z), respectively. For CEA, sensitivity at 95% specificity was 61% (95% CI: 46–74%). The AUC, representing the diagnostic ability of a test, was the highest (0.92) for CEA. This was significantly ( P < 0.001) higher than for proMMP-2(Z) and proMMP-9(Z). Their AUCs were 0.48 and 0.61, respectively, and were not statistically different ( P > 0.05) from the AUCs of tests that are positive or negative just by chance. Preoperative MMP, TIMP-1, and CEA levels and clinicopathological parameters Higher preoperative proMMP-2(Z) levels were found in patients with a short (<36 months) disease-free period after liver resection ( P = 0.023; Table 2 ). Significantly ( P = 0.019) higher levels of proMMP-9(Z) were found in patients with abnormal CEA levels (>5 ng/ml). ProMMP-9(E) levels were three times lower ( P = 0.010) in patients who recently (within 0–4 months before operation) received chemotherapy. TIMP-1 levels were related to patient age, with significantly ( P = 0.015) higher levels in patients older than 60 years. No relations were found with gender, stage and location of the original primary tumour, and the number and size of liver metastatic lesions. Preoperative MMP, TIMP-1 and CEA levels and survival The preoperative proMMP-2, -9 and TIMP-1 levels in potentially cured patients did not correlate significantly with disease-related survival as determined by the Kaplan–Meier method. Results on proMMP-2(Z) and CEA are illustrated by Kaplan–Meier curves in Fig. 2 . Patients with high proMMP-2(Z) levels (>135 AU/ml) showed no statistically significant difference ( P = 0.161) in survival when compared to patients with lower levels ( Fig. 2 A). Contrary to the data on proMMP-2, -9 and TIMP-1, patients with preoperative CEA levels above 5 ng/ml showed a significantly worse disease-related survival ( P = 0.030) than patients with low preoperative levels ( Fig. 2 B). The three-year postoperative survival probability was 53% (95% CI: 28–79%) in patients with CEA serum levels above 5 ng/ml and 74% (95% CI: 52–96%) in patients with lower levels. The effects of liver surgery on MMP and CEA levels After surgery, proMMP-2(Z) plasma levels increased by 12% in potentially cured patients. The median preoperative level of 127 AU/ml rose to 141 AU/ml ( P = 0.018). Changes in proMMP-9(Z) levels did not reach statistical significance. Median CEA levels fell from 5.1 ng/ml to 2.1 ng/ml ( P < 0.001). The changes found were not dependent of the type of procedure that was performed (resection, local tumour ablation, or a combination of both). Changes in MMP and CEA prior to disease relapse No significant changes in the blood levels of proMMP-2(Z and E), proMMP-9(Z and E) and TIMP-1 were found before disease relapse ( Table 3 ). In patients who remained disease-free (at least two years) after the resection of liver metastatic disease the proMMP-2(Z) levels increased significantly ( P = 0.040) over time. Still, no significant differences were observed between the two groups for any of these parameters. Median CEA serum levels increased significantly ( P < 0.001) in the patients that relapsed after liver operation and remained low in those which were disease-free after two years. The difference in the percentage change (26% vs 0%) in the two groups was statistically significant ( P = 0.014). Fig. 3 illustrates, in three representative patients, how blood proMMPs(Z), TIMP-1 and CEA levels behave prior to the discovery of disease relapse. Discussion The current study was undertaken primarily to assess the potential for preoperative plasma proMMP-2 and -9 levels to act as diagnostic or prognostic markers in CRC liver metastatic disease. The data presented here, though obtained on a study population of moderate size, yield no indication for such potential at all. Measurements taken after liver surgery indicate that the postoperative course of proMMP-2 should be investigated further since it shows some promise in diagnosing liver metastasis recurrence. The rationale for investigating these parameters is a firm one. The involvement of MMPs and TIMPs in cancer has been shown convincingly. 9,19 In CRC, clear correlations have been found between disease progression and tissue levels of various MMPs, among them the gelatinases. 20,21 Still, little is known about the relation between circulating MMP and TIMP levels and the development of cancer. A certain degree of diagnostic and prognostic power has been reported for circulating MMP-9 in lung cancer, 22,23 head and neck squamous cell cancer 24 and in papillary thyroid cancer. 25 Sonnate et al. have suggested that plasma MMP-9(E) may be an independent marker of CRC, potentially predicting invasion and metastasis. 26 In a preceding paper, we have established that preoperative proMMP-2 blood levels correlate with CRC diagnosis, identifying patients with 70% sensitivity at 95% specificity. In a combination with TIMP-1, this diagnostic sensitivity increases even to 84%. In addition, circulating TIMP-1 has a prognostic value similar to CEA in predicting patient survival. 15 While the number of patients within our study is not large enough to draw unequivocal conclusions regarding sub-populations, it is certainly sufficient to confirm (and extend) what is known about CEA and to investigate if MMP-2 and -9 blood levels are potential better markers than CEA. In the present study the main parameter to measure gelatinases in blood is gelatin zymography, which is quite quantitative a tool. 11,12,15,17 If available in sufficient quantity to allow a meaningful analysis, we have also supplied ELISA data ( Tables 1–3 ). It should be emphasized that both techniques measure essentially different properties of the enzymes, bioactivity vs. immunoreactivity, and that correlations between both data sets are no necessity. In the present study, the data from both types of assay point, almost without exception, the same way. While proMMP-2 plasma levels reportedly show potential as a diagnostic marker (AUC = 0.89) in primary CRC, 15 this diagnostic potential was absent in CRC liver metastasis (AUC = 0.48). This apparent contradiction may be the result of the different reference groups used. In primary CRC, preoperative plasma levels have been analysed against those in healthy subjects. Here, we have analysed the preoperative levels in CRC metastatic liver disease against samples of curatively operated CRC patients. The latter patients were sampled 9 months after the primary operation and remained disease-free for at least two years. This group has been chosen as a reference group because it corresponds best to those patients who are actually screened for liver metastases in clinical practice. Surprisingly, in an additional analysis (results not shown) it was established that, if samples from the reference group were taken 18 months, instead of 9, after primary CRC operation, the diagnostic ability of proMMP-2 increased (AUC = 0.67). It thus seems that the plasma proMMP-2 levels increase slowly after curative operation toward those of healthy controls. However, despite this observation, one must conclude that neither proMMP-2 nor for that matter proMMP-9 or TIMP-1, have any convincing diagnostic value during the formation of liver metastases in CRC patients. The present data demonstrate that circulating proMMP-2, -9 and TIMP-1 levels do not discriminate between primary CRC and CRC metastatic disease. This finding indicates that the same enzymes appear to be involved in primary CRC and in metastatic disease in the absence of the primary tumour (which has been resected). CEA behaves differently: circulating CEA levels are higher in the presence of liver metastasis than in primary disease. CEA seems, therefore, to be the marker most related to the actual metastatic process. With respect to the relations with clinicopathological parameters, decreased preoperative proMMP-2 levels appear to be linked with a longer disease-free period. This finding seems contrary to expectations when considering the other findings in this study. However, the “within-the-group”-comparison pertains to a small sample size of curatively operated patients only, which may influence the outcome. Plasma proMMP-9 levels are higher in patients which have received recent chemotherapy. Such an increase has also been found in our previous study on tissue levels in CRC liver metastatic disease. 12 Possibly, chemotherapy triggers the patient's immune system in such a way that macrophages and neutrophils, both sources for MMP-9, secrete their proteases. If so, this must be taken into account when MMP-9 would become a therapeutic target in any type of cancer treatment that involves chemotherapy or immunotherapy. The relation between TIMP-1 and age in primary CRC has been described previously, 27 but is also present in healthy controls 28 and thus probably not specifically related to cancer. The preoperative plasma levels of proMMP-2, -9 and TIMP-1 do not predict patient survival and therefore CEA remains the best circulating marker now available for predicting survival in metastatic CRC. Perhaps the finding, as it relates to TIMP-1, is somewhat unexpected since TIMP-1 reportedly has shown predictive power in this respect in primary CRC. 15,27 As expected for a marker of tumour burden, CEA levels go down after potentially curative resection of the liver metastases. ProMMP-2 levels increase somewhat toward the levels that are more representative of healthy controls. An interesting new finding is that, prior to a documented disease relapse, CEA levels rise again in most cases while previously increasing levels of proMMP-2 apparently become stable. On the contrast, the proMMP-2 levels in non-relapsing continue to increase ( Table 3 ). One could speculate that prolonged low levels of plasma proMMP-2 after curative liver metastases resection are indicative of incomplete resection or disease recurrence. In those patients where a repeat resection of liver metastases is clinically still possible, this finding could be of therapeutic value, especially for those patients who have tumours without increased CEA levels. In order to fully establish the value of this finding, further data are needed. In conclusion, the importance of optimal diagnostic and prognostic markers, essential in the selection of specific patients for specific treatment protocols, should again be emphasized. Because of their undisputed involvement in cancer growth and metastasis, circulating gelatinases should be fully analysed for this purpose. However, neither preoperative plasma proMMP-2 or proMMP-9 nor TIMP-1 exhibits diagnostic or prognostic value in CRC liver metastatic disease. Their concentrations are largely similar to those found in primary CRC and appear to be unrelated to patient survival. Still, the apparent prolonged low level of proMMP-2 after liver resection in relation to disease recurrence is certainly of interest and warrants further investigation as to its diagnostic potential. Acknowledgements This work was funded by the Dutch Cancer Society, grant KUN 97-1380. The authors thank B.Wiering for carefully reviewing the manuscript. References 1 Y. Fong J. Salo Surgical therapy of hepatic colorectal metastasis Semin Oncol 26 5 1999 514 523 2 P. Hohenberger P.M. Schlag T. Gerneth C. Herfarth Pre- and postoperative carcinoembryonic antigen determinations in hepatic resection for colorectal metastases. Predictive value and implications for adjuvant treatment based on multivariate analysis Ann Surg 219 2 1994 135 143 3 K. Brand A.H. Baker A. Perez-Canto A. Possling M. Sacharjat M. Geheeb Treatment of colorectal liver metastases by adenoviral transfer of tissue inhibitor of metalloproteinases-2 into the liver tissue Cancer Res 60 20 2000 5723 5730 4 B. Nordlinger J.C. Vaillant M. Guiguet P. Balladur F. Paris P. Bachellier Survival Benefit of repeat liver resections for recurrent colorectal metastases: 143 cases. Association Francaise de Chirurgie J Clin Oncol 12 7 1994 1491 1496 5 B. Nordlinger M. Guiguet J.C. Vaillant P. Balladur K. Boudjema P. Bachellier Surgical resection of colorectal carcinoma metastases to the liver. A prognostic scoring system to improve case selection, based on 1568 patients. Association Francaise de Chirurgie Cancer 77 7 1996 1254 1262 6 T.H. Schiedeck C. Wellm U.J. Roblick R. Broll H.P. Bruch Diagnosis and monitoring of colorectal cancer by L6 blood serum polymerase chain reaction is superior to carcinoembryonic antigen-enzyme-linked immunosorbent assay Dis Colon Rectum 46 6 2003 818 825 7 P. Vihinen V.M. Kahari Matrix metalloproteinases in cancer: prognostic markers and therapeutic targets Int J Cancer 99 2 2002 157 166 8 W.G. Stetler-Stevenson A.E. Yu Proteases in invasion: matrix metalloproteinases Semin Cancer Biol 11 2 2001 143 152 9 M. Egeblad Z. Werb New functions for the matrix metalloproteinases in cancer progression Nat Rev Cancer 2 3 2002 161 174 10 S. Zucker J. Vacirca Role of matrix metalloproteinases (MMPs) in colorectal cancer Cancer Metastasis Rev 23 1–2 2004 101 117 11 E.T. Waas R.-M.L.M. Lomme J. DeGroot T. Wobbes T. Hendriks Tissue levels of active matrix metalloproteinase-2 and -9 in colorectal cancer Br J Cancer 86 12 2002 1876 1883 12 E.T. Waas T. Wobbes R.M. Lomme J. DeGroot T. Ruers T. Hendriks Matrix metalloproteinase 2 and 9 activity in patients with colorectal cancer liver metastasis Br J Surg 90 12 2003 1556 1564 13 S.L. Parsons S.A. Watson H.M. Collins N.R. Griffin P.A. Clarke R.J. Steele Gelatinase (MMP-2 and -9) expression in gastrointestinal malignancy Br J Cancer 78 11 1998 1495 1502 14 J. Watine M. Miedouge B. Friedberg Carcinoembryonic antigen as an independent prognostic factor of recurrence and survival in patients resected for colorectal liver metastases: a systematic review Dis Colon Rectum 44 12 2001 1791 1799 15 E.T. Waas T. Hendriks R.-M.L.M. Lomme T. Wobbes Plasma levels of matrix metalloproteinase-2 and tissue inhibitor of metalloproteinase-1 correlate with disease stage and survival in colorectal cancer patients Dis Colon Rectum 48 4 2005 700 710 16 D.E. Kleiner W.G. Stetler Stevenson Quantitative zymography: detection of picogram quantities of gelatinases Anal Biochem 218 2 1994 325 329 17 I.H. de Hingh E.T. Waas R.M. Lomme T. Wobbes T. Hendriks Circulating matrix metalloproteinase-9 is transiently elevated after colorectal surgery Int J Colorectal Dis 19 5 2004 446 450 18 E.R. DeLong D.M. DeLong D.L. Clarke-Pearson Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach Biometrics 44 3 1988 837 845 19 L.M. Coussens B. Fingleton L.M. Matrisian Matrix metalloproteinase inhibitors and cancer: trials and tribulations Science 295 5564 2002 2387 2392 20 O.R. Mook W.M. Frederiks C.J. Van Noorden The role of gelatinases in colorectal cancer progression and metastasis Biochim Biophys Acta 1705 2 2004 69 89 21 M.F. Leeman S. Curran G.I. Murray New insights into the roles of matrix metalloproteinases in colorectal cancer development and progression J Pathol 201 4 2003 528 534 22 T. Iizasa T. Fujisawa M. Suzuki S. Motohashi K. Yasufuku T. Yasukawa Elevated levels of circulating plasma matrix metalloproteinase 9 in non-small cell lung cancer patients Clin Cancer Res 5 1 1999 149 153 23 E. Hrabec M. Strek D. Nowak Z. Hrabec Elevated level of circulating matrix metalloproteinase-9 in patients with lung cancer Respir Med 95 1 2001 1 4 24 S.M. Ranuncolo E. Matos D. Loria M. Vilensky R. Rojo E. Bal-de-Kier-Joffe Circulating 92-kilodalton matrix metalloproteinase (mmp-9) activity is enhanced in the euglobulin plasma fraction of head and neck squamous cell carcinoma Cancer 94 5 2002 1483 1491 25 S.Y. Lin Y.Y. Wang W.H. Sheu Preoperative plasma concentrations of vascular endothelial growth factor and matrix metalloproteinase 9 are associated with stage progression in papillary thyroid cancer Clin Endocrinol (Oxf) 58 4 2003 513 518 26 A.M. Sonnante M. Correale M. Linsalata A. Di Leo V. Guerra Circulating levels of matrix metalloproteinase-9 in patients with colorectal cancer Scand J Gastroenterol 35 6 2000 671 672 27 M.N. Holten-Andersen R.W. Stephens H.J. Nielsen G. Murphy I.J. Christensen W. Stetler-Stevenson High preoperative plasma tissue inhibitor of metalloproteinase-1 levels are associated with short survival of patients with colorectal cancer Clin Cancer Res 6 11 2000 4292 4299 28 M.N. Holten Andersen G. Murphy H.J. Nielsen A.N. Pedersen I.J. Christensen G. Hoyer Hansen Quantitation of TIMP-1 in plasma of healthy blood donors and patients with advanced cancer Br J Cancer 80 3–4 1999 495 503