Int. J. Cancer (Pred. Oncol.): 74, 220–223 (1997) r 1997 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer MATRIX METALLOPROTEINASES 1 AND 3, TISSUE INHIBITOR OF METALLOPROTEINASE-1 AND THE COMPLEX OF METALLOPROTEINASE-1/TISSUE INHIBITOR IN PLASMA OF PATIENTS WITH PROSTATE CANCER Klaus JUNG1*, Lars NOWAK1, Michael LEIN1, Friedrich PRIEM2, Dietmar SCHNORR1 and Stefan A. LOENING1 of Urology, University Hospital Charité, Humboldt University Berlin, Berlin, Germany 2Department of Clinical Chemistry, University Hospital Charité, Humboldt University Berlin, Berlin, Germany 1Department We analyzed blood plasma concentrations of matrix metalloproteinase-1 and -3 (MMP-1; MMP-3), the tissue inhibitor of metalloproteinase-1 (TIMP-1) and the complex MMP-1/ TIMP-1, and looked for any correlation with prostate cancer stage. These components were measured by ELISA tests specific for these proteins in healthy male controls (n 5 35), and in patients with benign prostatic hyperplasia (BPH; n 5 29), with prostate cancer (PCa) without metastasis (T2,3pN0M0; n 5 29) and with PCa with metastatic disease (T2,3,4pN1,2M1; n 5 18). Mean values of MMP-1 and of the complex MMP-1/TIMP-1 were not different among the 4 groups studied. The mean MMP-3 and especially TIMP-1 concentrations were significantly higher in PCa patients with metastases compared with controls, BPH and PCa patients without metastases. Ten of these 18 patients had TIMP-1 concentrations higher than the upper reference limit. TIMP-1 concentrations were correlated with staging but not with grading. Our results point towards plasma TIMP-1 concentration as a potential marker of malignant progression of PCa. Int. J. Cancer 74:220–223, 1997. r 1997 Wiley-Liss, Inc. Matrix metalloproteinases (MMP) are a family of enzymes with the common ability to degrade various components of the extracellular matrix such as collagen, elastin and gelatin (Liotta and Stetler-Stevenson, 1991). These enzymes are involved in all physiological processes occurring during tissue remodelling and repair and play a crucial role in pathological conditions such as rheumatoid arthritis, tumor invasion and metastasis (Crawford and Matrisian, 1995). Both in vitro and in vivo investigations have shown that increased concentrations of MMPs are associated with the invasive and metastatic potential in several human malignant tumors, e.g., in breast, colon, gastric and lung cancers (Anderson et al., 1995; Zucker et al., 1995; Murray et al., 1996). It is assumed that these proteases destroy the integrity of the basement membrane and enable cancer cells to invade normal tissue and facilitate tumor cell dissemination via blood vessels and lymphatics. The catalytic activities of MMPs are controlled by various mechanisms including enzyme synthesis, secretion as zymogens which undergo extracellular activation and inhibition of the activated enzymes by specific inhibitors, the tissue inhibitors of metalloproteinases (TIMPs). Thus, the balance between MMPs and TIMPs as both positive and negative modulators of the invasive and metastatic processes appears to be decisive (Crawford and Matrisian, 1995). Since these changes in cellular concentration may be reflected in body fluids, determinations of MMPs, TIMPs and of their complexes in the blood have been recommended as simple, noninvasive tools in cancer diagnostics and monitoring (Zucker et al., 1995). Increased plasma concentrations of the complex gelatinase B:tissue inhibitor have been observed in patients with gastrointestinal tract cancer (Zucker et al., 1995). There has been only one report on MMPs in the blood of prostate cancer (PCa) patients: Baker et al. (1994) noted changes in the concentration of MMP-1, MMP-3, TIMP-1 and TIMP-2 in the blood of such patients using in-house tests. Although these results are of great interest, no confirmatory data have been presented so far. Serum samples as used by these authors not being suitable for reliable measurements of these components (Jung et al., 1996), and since commercial assays for determining MMP-1, MMP-3, TIMP-1 and the complex of MMP1/TIMP-1 have since been introduced, our objective was to evaluate their potential usefulness in patients with PCa. MATERIAL AND METHODS Patients and samples A total of 111 men was studied, including a control group of 35 healthy individuals (median age 48, range 30–68 years), 29 patients with benign prostate hyperplasia (BPH) (median age 64, range 51–82 years) and 47 with PCa (median age 65, range 46–79 years). The diagnosis of BPH was based on histological analysis of tissue obtained by transurethral resection of the prostate. The diagnosis of PCa was established histopathologically by microscopic examination of prostatic specimens after biopsy or prostatectomy. Clinical staging was made according to the TNM system by digital rectal examination, transrectal ultrasonography, determination of prostatespecific antigen (PSA) and radionuclide bone scan including surgical assessment of lymph node status. Computed tomography and/or magnetic resonance imaging was performed in some individuals. Of the 47 patients with PCa, 29 (median age 65, range 46–79 years) had PCa without metastasis (T2,3pN0M0) and 18 (median age 67, range 52–77 years) had PCa with metastasis (T2,3,4pN1,2M1). Blood was collected before examination in plastic tubes for preparation of serum samples (for PSA and other conventional clinical chemistry parameters) and in lithium-heparin-coated plastic tubes for preparation of plasma samples (for MMP and TIMP) (Monovette Systems 03.1528, 03.1589; Sarstedt, Nümbrecht, Germany). The tubes were centrifuged within 30 min after venipuncture at 1,600g for 15 min at 4°C. The supernatants were carefully removed, divided into aliquots and stored at 280°C until analysis. ELISA tests for MMP and TIMP We used BIOTRAK ELISA assays for MMP-1, MMP-3, TIMP-1 and the MMP-1/TIMP-1 complex (RPN 2610–2612; Amersham, Little Chalfont, UK). The assays are based on the 2-side sandwich principle. Standards, controls and samples are incubated in microtiter wells precoated with respective antibodies. The components MMP-1, TIMP-1 and MMP-1/TIMP-1 complex, respectively, bind to the wells, and other components of the samples are removed by 4 washing steps. A second antibody specific to the respective component is added to the wells and incubated. After washing, the second antibody bound to MMP-1, MMP-3, TIMP-1 or complex is detected using horseradish-labelled antibodies and tetramethyl- Contract grant sponsor: Fund of the German Chemical Industry, contract grant number 400770; Contract grant sponsor: Family-Klee-Foundation. *Correspondence to: Department of Urology, University Hospital Charité, Humboldt University Berlin, Schumannstrasse 20/21, D-10098 Berlin, Germany. Fax: (1149-30)2802-1402. E-mail: email@example.com Received 20 November 1996; revised 24 December 1996 BLOOD METALLOPROTEINASES IN PROSTATE CANCER benzidine as substrate. The reaction is terminated by addition of sulphuric acid and the absorbance is measured on a microplate (HTIII, Anthos Labtec, Salzburg, Austria) at 450 nm using the cubic-spline method for calculation of concentrations (EIA/KINStar software, version 7.0, WEPAH-MED, Berlin, Germany). According to the data provided by the manufacturer, the respective assays specifically recognize the corresponding proteins and there is no crossreaction with other metalloproteinases. All tests were performed in duplicate. The limits of detection for MMP-1, MMP-3, TIMP-1 and the MMP-1/TIMP-1 complex, respectively, defined as the corresponding concentrations located 3 standard deviations above the measured average blanks (n 5 10), were 0.13 µg/l, 5.0 µg/l, 1.1 µg/l and 4.8 µg/l, respectively. The precision of the data determined with pooled sera were between 3.0–10.7%. PSA and other routine laboratory parameters PSA was measured by the AxSym test (Abbott, Abbott Park, IL). Routine clinical chemistry parameters, including alanine aminotransferase (normal upper limit: 41 U/l), g-glutamyltransferase (49 U/l) and C-reactive protein (5 mg/l), were measured by standard assays on the Hitachi 717 analyzer (Boehringer Mannheim, Germany). Statistical analyses Statistical calculations were performed by the statistical package Statgraphics, version 5.01 (Statistical Graphics, Rockville, MD). The Mann-Whitney U test with independent samples, the correlation coefficient according to Spearman, analysis of variance (ANOVA) and a distribution fitting procedure (KolmogorovSmirnov test) were used. The central 95% reference intervals in the control group were calculated according to the procedure recommended by the IFCC (Solberg, 1987); p , 0.05 was considered statistically significant. RESULTS The control group consisted of 35 healthy men. In the control group, plasma concentrations of MMP-1, MMP-3, TIMP-1 and the complex MMP-1/TIMP-1 did not correlate with each other, and did not correlate with alanine aminotransferase and the C-reactive protein. The preliminary upper reference limits were calculated as means 1 1.96 SD (Solberg, 1987). The corresponding limits were 19.7 µg/l for MMP-1, 19.8 µg/l for MMP-3, 1,740 µg/l for TIMP-1 and 110 µg/l for the complex MMP-1/TIMP-1. Figure 1 shows the values of the MMP-1, MMP-3, TIMP-1 and MMP-1/TIMP-1 complex in plasma of controls and the patients with BPH and PCa. Patients with PCa were subdivided in 2 groups, i.e., patients without metastasis and patients with distant metastasis and/or metastasis in regional lymph nodes as having locally advanced and/or metastatic disease. The mean age of BPH and PCa patients did not differ but it was somewhat higher than in healthy controls (64 and 65 vs. 48 years). However, there was no correlation between the 4 analytes and ages (range, 30–82 years) in the noncarcinoma group of healthy controls and BPH patients (rs 5 20.19, 20.18, 0.09, 0.22; p . 0.05). Thus, no age-effect on the concentrations of these analytes has to be taken into account when comparing concentrations measured in the groups studied. The study groups revealed PSA values typical for each kind of patient. Thus, PSA values in BPH patients, and in PCa patients without metastasis and with metastatic disease were clearly different (median values, 4.2 vs. 6.5 and 23.7 µg/l). However, the mean MMP-1 and MMP-1/TIMP-1 complex concentrations were not increased in the 3 groups of patients compared with those in the control group (Fig. 1a,d; p . 0.05). There was a tendency for rather lower values, and only a few patients were outside the reference limits. In contrast, the mean MMP-3 and especially TIMP-1 concentrations in PCa patients with metastases were significantly higher than those in the control group, BPH patients and PCa patients without metastasis (Fig. 1b,c; p , 0.05). These changes were distinctly more pronounced in the case of TIMP than in that of MMP-3. Ten of the 18 PCa patients with metastases had 221 values above the upper TIMP-1 reference limit, but only 3 were above the MMP-3 limit. Although the mean TIMP-1 value in PCa patients without metastasis was not increased compared with the controls and BPH patients, 7 of these 29 patients showed values higher than the upper reference limit. Ten patients of that group had PSA values ,4 µg/l, generally considered as being the normal upper PSA cutoff limit. It was interesting that 4 patients of that subgroup showed increased TIMP values. There were no correlations of TIMP-1 and MMP-3 values in these patients ( p . 0.05) with those of alanine aminotransferase (rs 5 20.10 and 20.02), g-glutamyltransferase (rs 5 0.28 and 20.13) and C-reactive protein (rs 5 0.07 and 0.07). The correlation matrix between staging, grading, PSA and the metalloproteinase parameters is shown in Table I. MMP-1, MMP-3 and the MMP-1/TIMP-1 complex did not correlate with staging or with grading, whereas TIMP-1 concentrations showed a correlation with staging (N/M state) similar to that of PSA (rs 5 0.383; p , 0.02), but not with grading. Thus, the correlation of TIMP corresponds to the TIMP concentrations found in the 2 PCa groups with and without metastasis. DISCUSSION The MMP family can be divided into 4 subclasses according to substrate specificity and structural similarity (Liotta and StetlerStevenson, 1991). These are the gelatinases (MMP-2 and MMP-9), the collagenases (MMP-1 and MMP-8), the stromelysins (MMP-3 and MMP-10) and a subgroup including matrilysin (MMP-7), stromelysin 3 (MMP-11) and metalloelastase (MMP-12). Furthermore, 3 specific TIMPs are recognized (TIMP-1, TIMP-2 and TIMP-3). It is assumed that these proteins, when increased in cancer tissue, may be elevated in the plasma of patients with more invasive or metastatic cancers (Zucker et al., 1995). Measurements have so far been performed by determining activities using nonstandardized substrate degradations assays and immunological assays using in-house ELISA approaches (Zucker et al., 1995). Commercial tests for MMP-1, MMP-3, TIMP-1 and the MMP-1/ TIMP-1 complex have now been introduced, providing the opportunity for evaluating their potential as tumor markers (Zucker et al., 1995). Our results show that the analytical sensitivity and reliability of these assays are indeed suited to determine these components in human plasma. On human prostate tissue sections, various investigations including immunohistochemical studies and in situ hybridization have demonstrated, similar to findings in other carcinomas, several changes in matrix metalloproteinases in malignant tissue, as compared with adjacent noncarcinoma tissue parts (Pajouh et al., 1991; Boag and Young, 1993; Stearns and Wang, 1993; Hamdy et al., 1994; Knox et al., 1996; Stearns and Stearns, 1996). For example, direct correlations were observed between the intensity of MMP-2 expression and the Gleason score (Stearns and Wang, 1993; Stearns and Stearns, 1996). Changes in the expression of MMP-7 and MMP-9 were also observed in benign prostatic hyperplasia and PCa (Pajouh et al., 1991; Hamdy et al., 1994). Therefore, it appears justified to also look for changes in the concentration of metalloproteinases in the blood of patients with PCa. We found only one report on metalloproteinase in the blood of PCa patients (Baker et al., 1994). Baker et al. (1994), investigating MMP-1, MMP-3, TIMP-1 and TIMP-2 in the serum of PCa patients, found increased TIMP-1 and MMP-1, unchanged MMP-3 and decreased TIMP-2 concentrations. Our measurements of MMP-1, MMP-3, TIMP-1 and the MMP1/TIMP-1 complex show that only TIMP-1 and MMP-3 are elevated in PCa patients with metastases compared with controls, BPH patients and PCa patients without metastases. The increases of MMP-3 were markedly lower than those of TIMP-1. These data are partly in contrast with the results of Baker et al. (1994), who used serum samples. Our previous experiments have shown that the sampling process is a decisive preanalytical condition for reliable JUNG ET AL. 222 FIGURE 1 – Plasma concentrations of MMP-1 (a), MMP-3 (b), TIMP-1 (c) and MMP-1/TIMP-1 complex (d) in controls, and in patients with benign prostate hyperplasia (BPH) and with PCa without (PCa, m2) and with metastasis (PCa, m1). Values are expressed as individual values and as arithmetic means 6 SD. Horizontal lines indicate upper reference limits calculated from controls (see Results). Significance (at least p , 0.05) in comparison with: (a) controls; (b) BPH patients; (c) PCa without metastasis; (d) PCa with metastasis. TABLE I – CORRELATIONS OF MMP-1, MMP-3, TIMP-1 AND MMP-1/TIMP-1 COMPLEX WITH TUMOR STAGE, GRADE AND PSA1 Parameter rs values MMP-1 MMP-3 TIMP-1 MMP-1/TIMP-1 complex 0.04 20.22 20.08 20.10 0.363 0.26 0.25 0.343 0.28 0.523 0.12 0.363 0.333 0.09 20.05 0.24 Staging2 T N/M Grading PSA 1R values shown are Spearman rank correlation coefficients and s were calculated by combining the 47 patients with PCa.–2Patients were classified according to the TNM system and a grading scale of 1–3. Three situations of lymph node and distant metastasis were considered for the calculation of correlation: N0M0; N1,2M0; N1,2M1 or N3M1.–3Significant correlations: at least p , 0.05. measurements of these components in the blood (Jung et al., 1996). For example, MMP-1 and TIMP-1 measured in the serum were about 5–7 times higher than in lithium heparin plasma because these components are probably released during the clotting process. Therefore, the results described by Baker et al. (1994) must be interpreted with caution. Our most important finding is that of an increase in TIMP-1 values in patients with metastases. Our results showed a stagedependence of TIMP-1 (Table I) and a correlation with PSA, a classical PCa parameter. However, there were 4 cases with increased TIMP values in the group of PCa patients without metastases (n 5 10) and normal PSA values (,4 µg/l). This result suggests the potential usefulness of assaying TIMP-1 in combination with PSA, but not instead of PSA. Follow-up studies will be necessary to validate this assumption. Increased TIMP-1 values have been found in plasma of patients with liver diseases, systemic sclerosis, rheumatoid arthritis and diabetes mellitus (Li et al., 1994; Manicourt et al., 1995). An interesting aspect raised by our results concerns the cellular origin of TIMP. Since there was no correlation of TIMP-1 with C-reactive protein and liver enzymes in the PCa patients with metastases, inflammatory processes or liver dysfunction may probably be excluded. The increased plasma levels may be either a result of a general reaction related to the tumor burden or, more probably, BLOOD METALLOPROTEINASES IN PROSTATE CANCER originate from the cancer tissue and/or metastatic tissue directly. Increased amounts of TIMP-1 have been demonstrated in colon cancer and hepatocellular carcinoma tissue (Zeng et al., 1995; Nakatsukasa et al., 1996). Tissue TIMP concentrations were indeed higher in advanced carcinomas cases. Cytokines, such as transforming growth factor-b (TGF-b), tumor necrosis factor-a and interleukins, are potent inducers of TIMP-1 (Chua and Chua, 1990). There is a close correlation between cytokines and TIMP-1 in tumor tissues. For example, TGF-b1 has been found to be higher in cancerous than in normal tissue of the prostate and was positively correlated with metastasis (Kakehi et al., 1996). However, there is no information on tissue TIMP values in PCa. TIMP-1 exerts 2 apparently opposite effects on tumor progression. It is a potent antagonist of metalloproteinases and inhibits important steps of tumor progression, especially proteolytic degradation of extracellular matrix and invasion. However, TIMP-1 stimulates the growth of normal and malignant cells independently of its inhibitory capacity (Hayakawa et al., 1992). Growthpromoting activities have been observed with TIMP concentrations composed between 10–100 ng/ml, whereas the inhibition of proteolytic degradation of the extracellular matrix needs concentrations above 1 µg/ml. Thus, both increased and decreased TIMP 223 concentrations may be characteristic of cancer tissues. These are tumor-specific (Zeng et al., 1995; Bramhall et al., 1996). For example, elevated mRNA TIMP-1 values observed in colon tumors are strongly correlated with lymph node and distant metastases (Zeng et al., 1995), whereas decreased levels have been noted in pancreatic cancer (Bramhall et al., 1996). We assume that the increased plasma TIMP-1 amounts observed in PCa patients reflect the situation in cancer tissues. In conclusion, we found significantly increased plasma MMP-3 and TIMP-1 concentrations in patients with advanced PCa compared with controls, in PCa patients without metastasis and in patients with benign prostatic hyperplasia, whereas MMP-1 and MMP-1/TIMP-1 were not elevated. Our results point towards plasma TIMP-1 as a potential marker of malignant progression in PCa, and these results warrant further studies. ACKNOWLEDGEMENTS This work was supported by grants from the Fund of the German Chemical Industry (to K.J.; 400770) and the Family-KleeFoundation (to M.L.). This study includes parts of the doctoral thesis of L.N. REFERENCES ANDERSON, I.C., SUGARBAKER, D.J., GANJU, R.K., TSARWHAS, D.G., RICHARDS, W.G., SUNDAY, M., KOBZIK, L. and SHIPP, M.A., Stromelysin-3 is overexpressed by stromal elements in primary non-small cell lung cancers and regulated by retinoic acid in pulmonary fibroblasts. Cancer Res., 55, 4120–4126 (1995). BAKER, T., TICKLE, S., WASAN, H., DOCHERTY, A., ISENBERG, D. and WAXMAN, J., Serum metalloproteinases and their inhibitors: markers for malignant potential. Brit. J. Cancer, 70, 506–512 (1994). BOAG, A.H. and YOUNG, I.D., Immunohistochemical analysis of type IV collagenase expression in prostatic hyperplasia and adenocarcinoma. Mod. Pathol., 6, 65–68 (1993). BRAMHALL, S.R., STAMP, G.W., DUNN, J., LEMOINE, N.R. and NEOPTOLEMOS, J.P., Expression of collagenase (MMP2), stromelysin (MMP3) and tissue inhibitor of the metalloproteinases (TIMP1) in pancreatic and ampullary disease. Brit. J. Cancer, 73, 972–978 (1996). CHUA, C.C. and CHUA, B.H., Tumor necrosis factor-alpha induces mRNA for collagenase and TIMP in human skin fibroblasts. Connect. Tissue Res., 25, 161–170 (1990). CRAWFORD, H.C. and MATRISIAN, L.M., Tumor and stromal expression of matrix metalloproteinases and their role in tumor progression. Invasion Metastasis, 14, 234–245 (1995). HAMDY, F.C., FADLON, E.J., COTTAM, D., LAWRY, J., THURRELL, W., SILCOCKS, P.B., ANDERSON, J.B., WILLIAMS, J.L. and REES, R.C., Matrix metalloproteinase 9 expression in primary human prostatic adenocarcinoma and benign prostatic hyperplasia. Brit. J. Cancer, 69, 177–182 (1994). HAYAKAWA, T., YAMASHITA, K., TANZAWA, K., UCHIJIMA, E. and IWATA, K., Growth-promoting activity of tissue inhibitor of metalloproteinases-1 (TIMP-1) for a wide range of cells. A possible new growth factor in serum. FEBS Lett., 298, 29–32 (1992). JUNG, K., NOWAK, L., LEIN, M., HENKE, W., SCHNORR, D. and LOENING, S.A., Role of specimen collection in preanalytical variation of metalloproteinases and their inhibitors. Clin. Chem., 42, 2043–2045 (1996). KAKEHI, Y., OKA, H., MITSUMORI, K., OGAWA, O. and YOSHIDA, O., Serum TGF-b1 levels in patients with prostatic cancer: positive correlation with metastasis and inverse correlation with age. J. Urol., 155, 419A (1996). KNOX, J.D., WOLF, C., MCDANIEL, K., CLARK, V., LORIOT, M., BOWDEN, G.T. and NAGLE, R.B., Matrilysin expression in human prostate carcinoma. Mol. Carcinogenesis, 15, 57–63 (1996). LI, J., ROSMAN, A.S., LEO, M.A., NAGAI, Y. and LIEBER, C.S., Tissue inhibitor of metalloproteinase is increased in the serum of precirrhotic and cirrhotic alcoholic patients and can serve as a marker of fibrosis. Hepatology, 19, 1418–1423 (1994). LIOTTA, L.A. and STETLER-STEVENSON, W.G., Tumor invasion and metastasis: an imbalance of positive and negative regulation. Cancer Res., 51, 5054s–5059s (1991). MANICOURT, D.H., FUJIMOTO, N., OBATA, K. and THONAR, E.J., Levels of circulating collagenase, stromelysin-1, and tissue inhibitor of matrix metalloproteinases 1 in patients with rheumatoid arthritis. Relationship to serum levels of antigenic keratan sulfate and systemic parameters of inflammation. Arthritis Rheum., 38, 1031–1039 (1995). MURRAY, G.I., DUNCAN, M.E., O’NEIL, P., MELVIN, W.T. and FOTHERGILL, J.E., Matrix metalloproteinase-1 is associated with poor prognosis in colorectal cancer. Nature (Med), 2, 461–462 (1996). NAKATSUKASA, H. and 13 OTHERS, Cellular distribution of transcripts for tissue inhibitor of metalloproteinases 1 and 2 in human cellular carcinomas. Hepatology, 24, 82–88 (1996). PAJOUH, M.S., NAGLE, R.B., BREATHNACH, R., FINCH, J.S., BRAWER, M.K. and BOWDEN, G.T., Expression of metalloproteinase genes in human prostate cancer. J. Cancer Res. clin. Oncol., 117, 144–150 (1991). SOLBERG, H.E., Approved recommendations (1987) on the theory of reference values. Part 5. Statistical treatment of collected reference values. Determinations of reference limits. J. clin. Chem. clin. Biochem., 25, 645–656 (1987). STEARNS, M. and STEARNS, M.E., Evidence for increased activated metalloproteinase 2 (MMP-2a) expression associated with human prostate cancer progression. Oncol. Res., 8, 69–75 (1996). STEARNS, M.E. and WANG, M., Type IV collagenase (M2 72,000) expression in human prostate: benign and malignant tissue. Cancer Res., 53, 878–883 (1993). ZENG, Z.S., COHEN, A.M., ZHANG, Z.F., STETLER-STEVENSON, W.G. and GUILLEM, J.G., Elevated tissue inhibitor of metalloproteinase 1 RNA in colorectal cancer stroma correlates with lymph node and distant metastases. Clin. Cancer Res., 1, 899–906 (1995). ZUCKER, S., LYSIK, R.M., DIMASSIMO, B.I., ZARRABI, H.M., MOLL, U.M., GRIMSON, R., TICKLE, S.P. and DOCHERTY, A.J., Plasma assay of gelatinase B: tissue inhibitor of metalloproteinase complexes in cancer. Cancer, 76, 700–708 (1995).