Matrix metalloproteinases in cancer: Prognostic markers and therapeutic targets


  • Pia Vihinen,

    Corresponding author
    1. Department of Oncology and Radiotherapy, Turku University Central Hospital, Turku, Finland
    2. MediCity Research Laboratory, University of Turku, Turku, Finland
    • MediCity Research Laboratory, University of Turku, Tykistökatu 6A, FIN-20520 Turku, Finland
    Search for more papers by this author
    • Fax: +358-2-3337000

  • Veli-Matti Kähäri

    1. Turku Centre for Biotechnology, University of Turku, Turku, Finland
    2. Department of Dermatology, Turku University Central Hospital, Turku, Finland
    Search for more papers by this author


Degradation of extracellular matrix is crucial for malignant tumour growth, invasion, metastasis and angiogenesis. Matrix metalloproteinases (MMPs) are a family of zinc-dependent neutral endopeptidases collectively capable of degrading essentially all matrix components. Elevated levels of distinct MMPs can be detected in tumour tissue or serum of patients with advanced cancer and their role as prognostic indicators in cancer is studied. In addition, therapeutic intervention of tumour growth and invasion based on inhibition of MMP activity is under intensive investigation and several MMP inhibitors are in clinical trials in cancer. In this review, we discuss the current view on the feasibility of MMPs as prognostic markers and as targets for therapeutic intervention in cancer. © 2002 Wiley-Liss, Inc.


AP-1, activator protein-1; MMP, matrix metalloproteinase; NSCLC, nonsmall cell lung cancer; SCC, squamous cell carcinoma; SCLC, small cell lung cancer; TACE, TNF-α converting enzyme; TIMP, tissue inhibitor of matrix metalloproteinase.

Matrix metalloproteinases (MMPs) are a family of structurally related zinc-dependent endopeptidases collectively capable of degrading essentially all components of extracellular matrix (ECM). MMPs play an important role in the physiologic degradation of ECM, e.g., in tissue morphogenesis, tissue repair and angiogenesis. MMPs also have important functions in pathologic conditions characterised by excessive degradation of ECM, such as rheumatoid arthritis, osteoarthritis, periodontitis, autoimmune blistering disorders of the skin and in tumour invasion and metastasis.1–3

At present, 21 members of the human MMP gene family are known (Fig. 1, Table I). Based on their structure and substrate specificity, MMPs are classified into subgroups of collagenases, stromelysins and stromelysin-like MMPs, matrilysins, gelatinases, membrane-type MMPs (MT-MMPs) and other MMPs.1–3 MMPs contain several conserved functional domains1 (Fig. 1).

Figure 1.

Structure of human matrix metalloproteinases. The signal peptide directs the proenzyme for secretion. The propeptide contains a conserved sequence (PRCGxPD), in which the cysteine forms a covalent bond (cysteine switch), with the catalytic zinc (Zn2+) to maintain the latency of proMMPs. Catalytic domain contains the highly conserved zinc binding site (HExGHxxGxxHS) in which Zn2+ is coordinated by 3 histidines. The proline-rich hinge region links the catalytic domain to the hemopexin domain, which determines the substrate specificity of specific MMPs. The hemopexin domain is absent in matrilysin (MMP-7) and matrilysin-2 (endometase, MMP-26). Gelatinases A and B (MMP-2 and MMP-9, respectively) contain 3 repeats of the fibronectin-type II domain inserted in the catalytic domain. MT1-, MT2-, MT3- and MT5-MMP contain a transmembrane domain and MT4- and MT6-MMPs contain a glycosylphosphatidylinositol (GPI) anchor in the C-terminus of the molecule, which attach these MMPs to the cell surface. MT-MMPs, MMP-11, MMP-23 and MMP-28 contain a furin cleavage site (RxKR) between the propeptide and catalytic domain, making these proenzymes susceptible to activation by intracellular furin convertases. MMP-23 contains an N-terminal signal anchor, which anchors proMMP-23 to the Golgi complex and has a different C-terminal domain instead of hemopexin-like domain.4

Table I. Human MMPS, their chromosomal localization, substrates, exogenous activators, and activating capacity1
EnzymeChromosomal locationSubstratesActivated byActivator of
  • 1

    FN, fibronectin; 2M, 2-macroglobulin; 1PI, 1-proteinase inhibitor; COMP, cartilage oligomeric matrix protein; ND, not determined; TACE, TNF-converting enzyme; OP, osteopontin.

 Collagenase-1 (MMP-1)11q22.2-22.3Collagen I, II, III, VII, VIII, X, aggregan, serpins, 2MMMP-3, -7, -10, plasmin kallikrein, chymaseMMP-2
 Collagenase-2 (MMP-8)11q22.2-22.3Collagen I, II, III, aggregan, serpins, 2MMMP-3, -10, plasminND
 Collagenase-3 (MMP-13)11q22.2-22.3Collagen I, II, III, IV, IX, X, XIV, gelatin, FN, laminin, large tenascin aggrecan, fibrillin, osteonectin, serpinsMMP-2, -3, -10, -14, -15, plasminMMP-2, -9
 Stromelysin-1 (MMP-3)11q22.2-22.3Collagen IV, V, IX, X, FN, elastin, gelatin, laminin, aggrecan, nidoge fibrillin*, osteonectin*, 1PI*, myelin basic protein*, OP, E-cadherinPlasmin, kallikrein, chymas tryptaseMMP-1, -8, -9, -13
 Stromelysin-2 (MMP-10)11q22.2-3As MMP-3, except *Elastase, cathepsin GMMP-1, -7, -8, -9, -13
Stromelysin-like MMPs
 Stromelysin-3 (MMP-11)22q11.2Serine proteinase inhibitors, 1PIFurinND
 Metalloelastase (MMP-12)11q22.2-22.3Collagen IV, gelatin, FN, laminin, vitronectin, elastin, fibrillin, 1-PI, myelin basic protein, apolipoprotein ANDND
 Matrilysin (MMP-7)11q22.2-22.3Elastin, FN, laminin, nidogen, collagen IV, tenascin, versican, 1PI, O E-cadherin, TNF-MMP-3, plasminMMP-9
 Matrilysin-2 (MMP-26)11q22.2Gelatin, 1PI, synthetic MMP-substrates, TACE-substrateNDND
 Gelatinase A (MMP-2)16q13Gelatin, collagen I, IV, V, VII, X, FN, tenascin, fibrillin, osteonectin, Monocyte chemoattractant protein 3MMP-1, -13, -14, -15, -16, -tryptase?MMP-9, -13
 Gelatinase B (MMP-9)20q12-13Gelatin, collagen IV, V, VII, XI, XIV, elastin, fibrillin, osteonectin 2MMP-2, -3, 7, -13, plasmin, trypsin, chymotrypsin, cathepsin GND
Membrane-type MMPs
 MT1-MMP (MMP-14)14q12.2Collagen I, II, III, gelatin, FN, laminin, vitronectin, aggrecan, tenasci nidogen, perlecan, fibrillin, 1PI, 2M, fibrinPlasmin, furinMMP-2, -13
 MT2-MMP (MMP-15)16q12.2FN, laminin, aggrecan, tenascin, nidogen, perlecanNDMMP-2, -13
 MT3-MMP (MMP-16)8q21Collagen III, FN, gelatin, casein, cartilage proteoglycans, laminin-1, 2MNDMMP-2
 MT4-MMP (MMP-17)12q24Fibrin, fibrinogen, TNF precursorND
 MT5-MMP (MMP-24)20q11.2ProteoglycansNDMMP-2
 MT6-MMP (MMP-25)16p13.3Collagen IV, gelatin, FN, fibrinNDMMP-2
Other MMPs
 MMP-1912q14Gelatin, aggrecan, COMP, collagen IV, laminin, nidogen, large tenastrypsinND
 Enamelysin (MMP-20)11q22Amelogenin, aggrecan, COMPNDND
 MMP-231p36McaPLGLDpaARNh2 (synthetic MMP substrate)NDND

Collagenase-1 (MMP-1), -2 (MMP-8) and -3 (MMP-13) are the principal proteinases capable of cleaving triple helical fibrillar collagens of types I, II, III and V into fragments, which denature into gelatin and are further degraded by other MMPs, such as gelatinases.1 MMP-1 cleaves fibrillar collagens with preference to type III collagen and MMP-8 prefers type I collagen.1 MMP-13 cleaves type II collagen most efficiently and also gelatin4 (Table I). In addition, MMP-13 degrades several other ECM components and serine proteinase inhibitors (Table I). The physiologic expression of MMP-13 in vivo is limited to situations, such as fetal bone development and fetal wound repair, in which rapid remodeling of collagenous ECM is required.4–6 MMP-13 is expressed in pathologic conditions, such as arthritis, chronic dermal and intestinal ulcers, chronic periodontal inflammation and atherosclerotic plaques.1 The expression of MMP-13 is detected in vivo in invasive malignant tumours, breast carcinomas, squamous cell carcinomas (SCCs) of the head and neck and vulva, malignant melanomas, chondrosarcomas and urinary bladder carcinomas.4–11

Stromelysin-1 (MMP-3) and -2 (MMP-10) are closely related with respect to structure and substrate specificity (Table I). This subgroup includes also stromelysin-3 (MMP-11) and metalloelastase (MMP-12). MMP-11 is expressed in stromal compartment of malignant tumours.12 MMP-12 is expressed in macrophages in alveoli of cigarette smokers13 and by tumour cells in cutaneous SCCs.14

Matrilysin (MMP-7) is mainly expressed in epithelial cells in different glandular structures found in endometrium, small intestine, breast, parotis, pancreas, liver, prostate, dermis, bronchus, as well as by epithelial tumours of the gastrointestinal tract, prostate and breast.15 MMP-7 acts in intestinal mucosal defense by activating antibacterial peptides, defensins.16 Matrilysin-2 (endometase, MMP-26) is expressed in uterine endometrium and placenta and in cancers of lung, prostate and breast.17

Gelatinase-A (MMP-2) and gelatinase-B (MMP-9) degrade components of basement membranes (Table I) and they are believed to be crucial in invasion of malignant tumours. In addition, gelatinases are responsible for the final degradation of fibrillar collagens after initial cleavage by collagenases.

To date, 6 MT-MMPs have been identified. MT1-MMP (MMP-14), MT2-MMP (MMP-15), MT3-MMP (MMP-16) and MT5-MMP (MMP-24) are anchored to cell membrane by a transmembrane domain in the C-terminus, whereas MT4-MMP (MMP-17) and MT6-MMP (MMP-25) are attached to cells by a C-terminal GPI anchor (Fig. 1). All MT-MMPs can activate proMMP-2 and most MT-MMPs degrade ECM components18 (Table I). The expression of MT-MMPs has been noted in several types of malignant tumours mainly in stromal cells.19

MMP-19 is expressed in mammary gland, placenta, lung, pancreas, ovary, spleen, intestine, blood vessels of skin and uterine ligaments and in activated peripheral blood mononuclear cells.1–3, 20 Enamelysin (MMP-20) has a restricted expression pattern in dental tissues and is thought to play a role in tooth enamel formation.1–3 MMP-23 is expressed in adult ovary, testis and prostate, suggesting that MMP-23 plays a role in reproduction. MMP-23 is also expressed in adult heart, intestine, colon, placenta and lung.21 Epilysin (MMP-28) is expressed at high levels in testis and at lower levels in lungs, heart, colon, intestine and brain.1–3 In addition, expression of epilysin is detected in the basal epidermis of intact skin and in basal keratinocytes in wounds.


In general, the level of expression of MMPs by unstimulated cells in culture and in intact tissues in vivo is low. The expression of MMPs is induced by cytokines, growth factors, tumour promoters, physical stress, oncogenic transformation and by cell-matrix and cell-cell interactions.3 The expression of MMPs is primarily regulated at the level of transcription and their proteolytic activity requires zymogen activation.

MMP genes inducible by extracellular stimuli (MMP-1, MMP-3, MMP-7, MMP-9, MMP-10, MMP-12 and MMP-13) harbour an AP-1 (activator protein-1) binding site in the proximal promoter.3 In contrast, promoters of MMP-2, MMP-11 and MMP-14 genes do not contain AP-1 elements.3 Extracellular signals activate the dimeric AP-1 complex, composed of Jun and Fos proteins, which bind to the AP-1 element and activate the transcription. The induction of expression and activity of AP-1 is mediated by 3 classes of mitogen-activated protein kinases (MAPKs): mitogen-activated extracellular signal-regulated kinase1, 2 (ERK1, 2), stress-activated Jun N-terminal kinases and p38 MAPK.3 Another cis-element, PEA3 (polyomavirus enhancer A-binding protein-3) site, is present in the promoters of MMP-1, MMP-3 and MMP-9, in which it can cooperate with the AP-1 element.3, 22

Most MMPs are secreted by cells as inactive zymogens, which are proteolytically activated in the pericellular space by tissue or plasma proteinases, bacterial proteinases or other MMPs (Table I). MMP-11, -23, -28 and MT-MMPs are activated prior to secretion by Golgi-associated furin proteases.21, 23

The proteolytic activity of MMPs is inhibited specifically by tissue inhibitors of metalloproteinases (TIMPs). The TIMP gene family consists of 4 members: TIMP-1, -2, -3 and -4.24 TIMPs inhibit the activity of MMPs by binding to activated MMPs in a 1:1 molar stoichiometry.24 TIMPs can also inhibit the growth, invasion and metastasis of malignant tumours.25 Also nonspecific inhibitors, such as α1-proteinase inhibitor and α2-macroglobulin, can control MMP activity.3


Direct evidence for the role of distinct MMPs in tumour growth and invasion has been provided by studies with knockout mice for specific MMPs. Mice lacking MMP-7 show reduction in intestinal tumorigenesis.26 MMP-11 knockout mice show reduced tumorigenesis in response to chemical mutagenesis27 and MMP-9-deficient mice show reduced formation of melanoma metastases.28 MMP-2-deficient mice show reduced melanoma tumour progression and angiogenesis.29

The levels of MMPs can be determined in patient serum or urine, where levels elevated over a particular threshold can sometimes predict progression or prognosis (Table II). The expression of MMPs in tumour can be determined by immunohistochemical staining, Western blotting, Northern blotting or by reverse transcriptase-polymerase chain reaction analysis. MMP activity can be determined with collagen, gelatin or casein zymography. In numerous work with different types of cancer, the expression levels of particular MMPs and their correlation with clinicopathologic characteristics of the patients have been studied. In general, there are 2 important aspects related to cancer progression in several studies: the association between MMP expression and tumour grade or aggressiveness and the correlation of MMP expression and activity with recurrence or metastasis risk.

Table II. Prognostic Value of MMPs in Cancer (Studies with ≥30 Patients)
Cancer typeMMPTissueExpression levelPatient no.Impact on survivalReference no.
  • BM, bone marrow; DFS, disease-free survival; HNSCC, head and neck squamous cell carcinoma; NSCLC, nonsmall cell lung cancer; SCLC, small cell lung cancer; DSS, disease-specific survival.

  • 1

    Both breast ca and colorectal ca patients were studied.

  • 2

    The expression ratio of MT1-MMP in tumour cells vs. normal epithelial cells or tumour stroma were studied.

  • 3

    MMP/E-cadherin ratio is measured.

  • 4

    Better overall survival but poorer survival of MMP-2-positive male melanoma patients.

  • 5

    MMP-2/TIMP-2 ratio is measured.

Acute myeloid leukaemiaMMP-2BM blast cellsPositive54Good prognosis89
Brain tumoursMMP-2Tumour cellsPositive101Poor survival152
Breast caMMP-2Tumour cellsPositive177Poor survival101
MMP-11Tumour cellsHigh111Poor survival & DFS153
Cervical caMMP-2Tumour cellsHigh49Poor prognosis37
Colorectal caMMP-1Tumour cellsPositive64Poor prognosis93
MMP-2SerumHigh158Poor survival103
MMP-9Tumour cellsHigh71Poor survival & DFS64
MMP-9SerumHigh1221Poor prognosis154
Gastric caMMP-1Tumour cellsPositive103Poor prognosis60
MMP-2Tumour cellsPositive100Poor prognosis95
MMP-2Tumour cellsHigh50Poor survival155
MMP-2Tumour cellsHigh203Poor prognosis156
MMP-9Tumour cellsHigh50Poor survival155
MT1-MMPTumour/normal tissue ratio ≥4.8268Poor prognosis157
MT1-MMPTumour/stroma2Positive127Poor prognosis48
Hepatocellular caMT1-MMPTumour cellsHigh36Poor prognosis158
HNSCCMMP-9Tumour cellsPositive52Poor survival72
Lung adenocaMMP-2, -9Tumour cellsHigh79Poor prognosis31
Lung caMMP-9SerumHigh90Poor survival159
NSCLCMMP-2Tumour cellsHigh193Poor prognosis98
MMP-2,3 -93Tumour cellsLow60Longer DFS160
MMP-9Tumour cellsPositive169Poor prognosis161
SCLCMMP-3, -11, MT1-MMPTumour cellsPositive46Poor survival107
 SkinMMP-2Tumour cellsHigh50Better/poor survival4162
 MetastaticMMP-1, -3Tumour cellsHigh56Poor DFS90
NeuroblastomaMT1-MMPTumour cellsHigh30Poor survival163
Oesophageal caMMP-1Tumour cellsPositive46Poor survival164
MMP-1Tumour cellsHigh51Poor prognosis92
MMP-7Tumour cellsHigh48Poor prognosis108
MMP-13Tumour cellsHigh45Poor prognosis40
Ovarian caMMP-2Tumour cellsHigh33Poor prognosis54
MMP-9Tumour cellsHigh45Poor prognosis165
Pancreatic adenocaMMP-1Tumour cellsPositive46Poor prognosis94
MMP-7Tumour cellsPositive70Poor survival109
Tongue SCCMMP-2, MT1-MMPTumour cellsPositive51Poor prognosis62
Urothelial caMMP-2, MT1-MMPTumour cellsHigh41Poor survival49
MMP-2, -3SerumHigh117Poor DFS104
MMP-25SerumHigh53Poor DFS53
MMP-25SerumHigh97Poor prognosis105
MMP-2Tumour cellsHigh61Poor DSS99

MMP-2 and MMP-9 can both degrade type IV collagen of basement membrane—the first barrier for cancer invasion. Expression of MMP-2 and MMP-9 is elevated in carcinomas in association with low differentiation grade and accelerated tumour progression in oral carcinoma,30 lung adenocarcinoma,31 bladder32 and ovarian carcinoma33 and papillary thyroid carcinoma.34 MMP-2 expression correlates also with advanced disease in neuroblastoma.35 MMP-236, 37 or MT1-MMP are not expressed in cervical neoplasia but their expression levels are high in cervical carcinoma and correlate with advanced stage.37 Similarly in endometrial sarcoma, staining for MMP-3 and MMP-9 is more pronounced in high-grade than low-grade tumours.38 Increased MMP-13 or MT1-MMP production is associated with tumour aggressiveness in laryngeal39 or eosophageal carcinomas.40 In eosophageal carcinoma, MMP-7 expression correlates to tumour aggressiveness.41


Tumour invasion is a multistep process in which cell motility is coupled to proteolysis and which involves interactions of cells with the ECM. During invasion, malignant cells detach from the primary tumour and invade through basement membranes and stromal ECM. In spite of observations on abundant expression of specific MMPs in invasive primary tumours or in their metastases, the evidence for the activity of distinct MMPs in tumour tissues in vivo is limited.3, 25, 42 In most malignant tumours, stromal fibroblasts are the primary source of MMPs. Infiltration of inflammatory cells is a prominent feature of many tumours and they also produce MMPs to the peritumoural environment. Inflammatory cells also produce cytokines, which enhance expression of MMPs by tumour and stromal cells. Tumour cells produce factors, which enhance production of MMPs by fibroblasts.43, 44 It is likely that MMPs form a network in which a single MMP cleaves certain native or partially degraded matrix components and activates other latent MMPs. It is also likely that distinct MMPs play a role at different stages of tumour development.

High expression levels of certain MMPs are related to tumour invasion capacity in vivo. This has been shown in laryngeal carcinoma with MMP-13,45 in oesophageal carcinoma with high MMP-7, MMP-9 and MT1-MMP expression levels46 and in oral SCCs with high MMP-2 and MMP-930, 47 and MMP-1, MMP-3 and MT1-MMP expression.47 Invasive behavior of gastric carcinoma is associated with MT1-MMP expression48 and bladder carcinoma with both MMP-2 and MT1-MMP expression.49 In papillary thyroid carcinoma, high MMP-2 and MMP-9 expression levels correlate with invasion capacity and lymph node metastasis.34

Recent observations show that insertion of G nucleotide at −1607 bp in MMP-1 promoter generates a new ETS binding site and increases the transcription of the MMP-1 gene.50 Homozygosity for this single nucleotide insertion polymorphism (2G allele) has been shown to correlate with invasiveness of colorectal cancer51 and cutaneous melanoma.52 In addition, homozygosity for this 2G allele has been detected in metastases of malignant melanoma. Together these observations show that single nucleotide polymorphisms in the regulatory regions of MMP genes may serve as prognostic markers.


MMPs can be used as markers to predict tumour recurrence in several cancer types. High preoperative serum levels of MMP-2 or MMP-3 predicts recurrence in patients with advanced urothelial carcinoma.53 Similarly in ovarian cancer, high expression levels of MMP-2 in tumour cells can predict tumour recurrence.54 Kuniyasu et al.55 found that a high ratio of gelatinase expression (MMP-2 or MMP-9) to E-cadherin expression in tumour cells can predict recurrence and death in pancreatic cancer. Similarly, expression of activated MMP-2 is related to regional lymph node and distal metastasis as well as to postresection recurrence of the same tumour.56 Concomitant overexpression of MMP-2 and MMP-7 is associated with recurrence in hepatocellular cancer,57 and elevated MMP-9 expression in tumor cells is related to recurrence in superficial carcinoma of the urinary bladder.58

The expression of certain MMPs in primary tumour can predict the risk of metastasis. Expression of MMP-1 is associated with lymphvascular invasion and lymph node metastasis in stage IB cervical cancer59 and peritoneal metastasis in gastric cancer.60 Expression of MMP-2 in tumour cells can indicate increased risk of metastasis in uveal melanoma61 and in SCC of tongue.62 The expression of MMP-7 in tumour cells is associated with liver and lymph node metastasis in gastric carcinomas.63 Similarly, increased MMP-9 expression by tumour cells in colorectal cancer is associated with advanced Dukes stage and presence of distant metastases.64 Interestingly, MMP-2 and MMP-9 expression levels are especially high in lung carcinomas and melanomas metastasizing to the spine, suggesting that they contribute to enhanced invasive properties of metastatic spinal tumours.65

MMP expression in certain tumours can also predict hematogenous metastasis. This has been shown with MMP-2 in breast cancer66 and with MMP-1 in colorectal cancer.67 In some studies, MMP determinations from patient serum have shown predictive value in estimation of metastasis risk. High serum levels of MMP-2 correlate with the presence of metastases in lung cancer68 or to disease progression in patients with prostate cancer,69 and a high serum MMP-9/E-cadherin ratio can predict metastasis of renal cell carcinoma.70


In addition to hematogenous spread, certain MMPs may play a role in tumour vascularisation.71 Riedel et al. have suggested that head and neck carcinomas negative for MMP-9 have smaller microvessel density than positive tumours.72 The role of MMPs in angiogenesis has also been implicated in studies with knockout mice.28, 29 Disturbed angiogenesis has been noted in mice lacking MMP-2, -9 and MT1-MMP.28, 29, 73–75 Inhibition of angiogenesis has also been detected in in vivo studies by treatment of tumour-bearing animals with MMP inhibitors.76 Inhibition of angiogenesis in vivo has been shown at least with MMP inhibitors prinomastat,76 BAY 12-9566,77 batimastat,78 BMS-275291,79 neovastat80 and metastat.81 MMPs can also be highly expressed in blood vessels as MMP-2 and MMP-9 in abdominal aortic aneurysms.82 In this disease, MMPs can function in both, modulating vascularisation and structure of ECM.

It is interesting that MMP-3, -7, -9 and -12 can generate angiostatin from plasminogen, indicating that their activity in peritumoural area may inhibit tumour-induced angiogenesis.83–86 Recent observations also show that endostatin can inhibit catalytic activities of both MMP-2 and MT1-MMP.87 Furthermore, at least in vitro the release of MMP-2 from tumour cells can induce platelet aggregation, which might affect the tumour spread.88 Regardless of the regulatory mechanisms, it is important to elucidate whether MMP expression can serve as a marker to predict metastasis formation and help in determining whether adjuvant therapy is indicated for patients at high risk of recurrence and metastasis.


In the majority of prognostic studies, increased expression levels of different MMPs are related to poor survival. However, in certain situations, increased MMP expression appears to be associated with better survival or treatment response. In acute myeloid leukaemia, the positivity of bone marrow blast cells for MMP-2 is associated with better prognosis.89 We have shown that high expression level of MMP-1 in tumour cells correlates to a favourable treatment response in human metastatic melanoma.9 We have also found that high expression levels of MMP-1 or MMP-3 are associated with shorter disease-free survival in the same disease.90 On the other hand, Arenas-Huertero et al.91 have found that low activity of MMP-2 or MMP-9 are associated with better response to neoadjuvant chemotherapy in SCC of oral cavity.

Presence of MMP-1 in tumour cells has unfavourable prognostic value in cancers of the gastrointestinal tract. MMP-1 expression indicates poor prognosis in oesophageal,92 gastric,60 colorectal93 and pancreatic cancer.94 Increased expression levels of MMP-2 in tumour cells is related to poor survival in gastric cancer,95 hypopharyngeal SCC,96 ovarian cystadenocarcinoma,97 adenocarcinoma31 and nonsmall cell carcinoma of the lung,98 ovarian54 and cervical carcinoma,37 bladder carcinoma,49 urothelial99 and renal carcinoma,100 breast carcinoma101 and uveal melanoma.61 However, Ring et al.102 found no correlation between positivity of tumour cells for MMP-2 or MMP-9 and tumour stage or overall survival in 212 patients with colorectal carcinoma.

High serum MMP-2 levels are related to poor survival in ovarian cystadenocarcinoma,97 colorectal carcinoma103 and urothelial carcinoma.53 In urothelial cancer, a high serum MMP-2/TIMP-2 ratio predicts poor prognosis104 or recurrence.105 On the other hand, Vuoristo et al.106 found no prognostic value for serum total MMP-2 levels in patients with advanced cutaneous melanoma. Increased expression of MMP-3 in tumour cells of SCLC (107) and in serum of urothelial carcinoma patients53 have shown only limited prognostic value. Increased MMP-7 expression levels in tumour cells of patients with oesophageal carcinoma108 or pancreatic adenocarcinoma109 may be associated with poor prognosis.

On the basis of these studies, it is evident that measurement of tumour/serum MMP levels may provide data for selecting and following patients considered for treatment with drugs that interfere with MMP activity.110


Inhibition of MMP activity in the extracellular space has been extensively studied as an approach to inhibit growth and invasion of neoplastic cells. At present, several MMP inhibitors (MMPIs) are in clinical trials (Table III) and have been expected to represent a new approach to cancer treatment in addition to traditional cytotoxic drugs. MMPIs currently in clinical trials are synthetic peptides or nonpeptidic molecules, chemically modified tetracyclines, bisphosphonates or natural MMP inhibitors (Table III). MMPIs may inhibit tumour growth by enhancing development of fibrotic capsule around the tumour thereby preventing tumour invasion, by inhibiting tumour-induced angiogenesis or by inducing apoptosis in tumour cells. The principal side effect of these drugs is musculoskeletal pain, especially in tendons and joints possibly due to inhibition of tumour necrosis factor-α (TNF-α) converting enzyme (TACE) and shedding of TNF-α receptor II (TNF-RII).111 MMPIs have shown efficiency against malignant tumours in preclinical studies.111 However, the outcome of the clinical MMPI trials exceeding phase II have been disappointing thus far.

Table III. Clinical Cancer Trials with MMP Inhibitors
CompoundCompanyCancer typePhase of clinical trial
  • 1

    Studies with gastrointestinal tract tumours are continued.

  • 2

    Further studies in cancer are cancelled.

  • 3

    Phase III studies in prostate cancer and stage IV NSCLC have been discontinued.

  • 4

    Survival benefit has been shown in breast cancer and multiple myeloma.

Batimastat (BB-94)British BiotechSeveralPhase II/cancelled
Marimastat (BB-2516)British BiotechPancreatic, SCLC, NSCLC, ovarian, breast, glioblastoma, gastricPhase III/cancelled1
Solimastat (BB-3644)British Biotech/Schering-PloughNot availablePhase I2
Prinomastat (AG3340)Agouron/PfizerBreast, prostate, NSCLC, glioblastomaPhase II/III3
BAY12-9566BayerPancreatic, SCLC, NSCLC, ovarianPhase III/cancelled
BMS-275291 (D 2163)Chiroscience/Bristol Myers-SquibbNSCLCPhase II
CGS27023A (MMI 270B)NovartisSeveralPhase II/cancelled
Neovastat (Æ-941)AeternaMultiple myelomaPhase II
NSCLC, prostate, renalPhase III
Metastat (Col-3, CMT-3)CollaGenexKaposi's sarcoma, malignant gliomasPhase I/II
Bisphosphonates (clodronate, pamidronate, zelodronate, ibandronate)SeveralSeveral4Phase III


Peptidomimetic MMPIs are pseudopeptides that mimic the structure of MMP substrates and function as reversible competitive inhibitors.111 Hydroxyamate inhibitors are small (Mr<600) peptide analogs of fibrillar collagens, which inhibit MMP activity by specifically interacting with the Zn2+ in their catalytic site. Batimastat (BB-94) is a low-molecular-weight broad-spectrum hydroxamate-based inhibitor that inhibits MMPs and members of the adamalysin family of metalloproteinases, such as TACE (ADAM-17). Batimastat was the first MMPI to enter clinical trials. It is well tolerated, but its utility is limited by poor water solubility, which requires intraperitoneal administration.112 Surprisingly, a recent study showed that batimastat promotes liver metastasis in a mouse model.113 In a phase I study with patients having malignant pleural effusions, intrapleural batimastat reduced the need for pleural aspirations.114 However, phase I and II clinical trials with intraperitoneally administered batimastat have not shown any marked responses and there is no further development of batimastat for cancer therapy at the moment.115

Marimastat (BB-2516) is an orally bioavailable low-molecular-weight MMPI. Marimastat is a broad-spectrum MMPI, which inhibits the activity of MMP-1, -2, -3, -7, -9 and -12. In a phase I clinical trial, in which marimastat was administered orally twice daily to 12 patients with advanced lung cancer, no consistent change was seen in blood MMP levels.116 In another phase I study with patients with advanced, inoperable gastric cancer, marimastat twice or once daily for 28 days led to endoscopically measured favourable changes in 18/31 tumours.117 In these studies the dose-limiting side effect was mild to severe joint and muscle pain. Marimastat has been studied in phase II trials in patients with recurrent colorectal cancer,118 advanced pancreatic cancer119 and as adjuvant treatment in patients with early breast cancer.120 In these studies, response to treatment has been measured as a decrease in serum tumour marker CEA118 and CA19/9119 expression levels.

Marimastat has been studied in several different phase III clinical trials for treatment of pancreatic, ovarian, gastric and breast cancers as well as in NSCLC, SCLC and glioblastoma (Table III). In a phase III randomised trial in unresectable pancreatic cancer, no difference in 1-year survival was found between patients treated with 25 mg of marimastat or gemcitabine.121 In a study by British Biotech (study 186), treatment of advanced ovarian cancer patients with the combination of carboplatin and marimastat showed no statistically significant advantage over carboplatin alone.122 Two SCLC clinical trials with marimastat (studies 140 and 117) have been completed. Shepherd et al.123 have recently reported results from a randomised, double-blind, placebo-controlled trial in which they wanted to determine if adjuvant treatment with marimastat could prolong the duration of remission and overall survival in patients with SCLC in whom complete or partial response had been achieved by first-line chemotherapy. In this study, no difference was found between marimastat and placebo. In another randomised, double-blind, placebo-controlled trial in patients with glioblastoma multiforme or gliosarcoma, marimastat did not improve survival in patients who have completed conventional, first-line treatment.124 However, marimastat improved the survival of stable gastric cancer patients who had received prior chemotherapy.117 Preliminary results from another study suggest that treatment of patients with stable advanced gastric cancer with marimastat can increase survival and time to disease progression compared to placebo.125 Follow-up results of gastric cancer patients treated with marimastat suggest that they are having a long-term survival benefit compared to those who received placebo.126 Based on the outcome of these phase III trials, further clinical trials with marimastat will be performed only in cancers of the gastrointestinal tract.126

Solimastat (BB-3644) is a follow-on to marimastat and it was initially targeted at cancer. Solimastat has shown similar anticancer properties to marimastat but failed to show better tolerability than marimastat in a phase Ib trial (Table III).


Nonpeptidic MMPIs have been synthesized on the basis of the 3-dimensional conformation of MMP zinc-binding site. They are more specific and have better oral bioavailability than peptidomimetic inhibitors.

Prinomastat (AG3340) is a synthetic, low-molecular-weight collagen-mimicking MMP inhibitor. It is lipophilic and inhibits the activity of MMP-1, -2, -3, -7, -9 and -14. Prinomastat inhibits tumour growth and angiogenesis in several xenograft models.115 In a phase II study in patients with progressive breast cancer, no objective disease responses were observed.127 Two phase III trials with prinomastat were discontinued after preliminary results failed to meet primary efficacy objectives in advanced hormone refractory prostate cancer128, 129 and in stage IV NSCLC patients.129, 130 There were no beneficial effects of the combination of prinomastat with standard chemotherapy observed. Prinomastat is currently studied in earlier stage NSCLC and in a phase II trial for glioblastoma multiforme (Table III). The effect of prinomastat in combination with temozolomide in patients who have received surgery and radiotherapy for glioblastoma multiforme will be studied.131

BAY 12-9566 is a synthetic MMPI, which inhibits the activity of MMP-2, MMP-3, MMP-9 and MMP-13.132 In a trial with advanced pancreatic cancer patients, gemcitabine was significantly superior to BAY 12-9566 when disease-free and overall survival were studied.133 Phase III trials in SCLC, NSCLC, ovarian and pancreatic cancers were cancelled because in the SCLC trial, BAY-12-9566 was performing worse than placebo.134 Preliminary data from the cancelled phase III study of BAY 12-9566 vs. placebo in patients with advanced ovarian cancer responsive to primary therapy show no evidence of an impact of BAY 12-9566 on progression-free or overall survival.135

BMS-275291 (D2163) is an orally bioavailable inhibitor of MMP-2 and MMP-9, which inhibits angiogenesis. In rodent models BMS-275291 has shown inhibition of melanoma growth and reduced size and metastases of mammary carcinoma.79 BMS-275291 does not inhibit TNF–RII shedding and thus does not cause joint pain as a side effect. BMS-275291 is currently in phase II/III clinical trial with advanced or metastatic NSCLC in combination with standard chemotherapy, paclitaxel and carboplatin.136

MMI 270B/CGS27023A is a nonpeptidic MMPI and a potent inhibitor of MMP-1, MMP-2 and MMP-3 activity. In a phase I trial, 92 patients with advanced solid cancer were treated and 20% of them reached stable disease.137 Cutaneous rash, arthralgias and myalgias were seen as side effects at high CGS27023A doses.137 Due to lack of clear efficacy in phase II trials and to cutaneous side effects, further trials with CGS27023A have been cancelled.86


Neovastat (Æ-941) is an orally bioavailable extract from shark cartilage. The function of neovastat is based on multifunctional antiangiogenic effects. Neovastat can inhibit the activity of elastase, MMP-2, MMP-9, MMP-12 and MMP-13. Neovastat also inhibits the function of vascular endothelial growth factor receptor-2 and induces endothelial cell apoptosis.80 The antitumour activity of neovastat has been demonstrated in experimental mouse models with human glioblastoma cell graft, Lewis Lung Carcinoma, as well as in melanoma, colon and breast cancer models.138 Preliminary results of a phase II study suggest improvement in survival in a small group of patients with unresectable NSCLC who received neovastat.138 The most common side effects have been musculoskeletal pain and joint stiffness. There is some evidence about disease stabilization in metastatic prostate cancer as indicated by reduction in PSA levels or pain in patients treated with neovastat.139 Evaluation of another phase I/II trial in patients with metastatic renal cell carcinoma suggests that neovastat treatment can increase median survival time.139 Neovastat is currently evaluated in phase III clinical trials as a monotherapy in patients with refractory renal cell carcinoma who have progressed following a first line of immunotherapy. Similarly, advanced unresectable NSCLC (stage IIIA and IIIB) patients will be treated with neovastat or placebo in combination with induction platinum-based chemotherapy followed by chemo-radiotherapy. A phase II clinical trial is underway to evaluate the efficacy of neovastat as monotherapy treatment for patients with multiple myeloma not responding to standard therapies139 (Table III).


Metastat (Col-3, CMT-3) is a synthetic MMPI, which belongs to nonantimicrobial chemically modified tetracyclines (CMTs), with limited systemic toxicity but significant tumour cell toxicity and antimetastatic activity.81 Metastat inhibits the activity of MMP-1, MMP-2, MMP-8, MMP-9, MMP-13 and elastase and it downregulates expression of various inflammatory cytokines. In preclinical studies, metastat inhibited malignant cell invasion into normal lung tissue in rodent models of metastasis and it also inhibited tumour-induced angiogenesis. Metastat is being examined in the treatment of Kaposi′s sarcoma.140 In a phase I trial, metastat was found to induce disease stabilisation in several patients who had a nonepithelial type of malignancy and it was also able to lower plasma MMP-2 levels.141 Recently Munoz-Mateu et al.142 have also shown that metastat can suppress the production of both MMP-2 and MMP-9 when measured in plasma or skin biopsies in cancer patients. In both these studies, phototoxicity was found to be dose limiting. In some patients, treatment with metastat has been associated with reversible sideroblastic anaemia.143 Metastat is currently in a phase I/II study in patients with progressive or recurrent high-grade anaplastic astrocytoma, anaplastic oligodendroglioma or glioblastoma multiforme (Table III).


Bisphosphonates are a group of pharmacologic substances recently identified as MMP inhibitors.144 They were developed as inhibitors of bone resorption and have been used to treat patients with bone metastases.145 Their use in treatment of skeletal metastases in breast cancer and multiple myeloma has been established.146 Bisphosphonates have also been shown to inhibit secretion of MMP-2,147 and they prevent the inhibitory effect of TIMP-2 on MMP-2 degradation by plasmin and thereby enhance inactivation of MMP-2.148


The limited performance of MMPIs in phase III clinical trials so far has been a disappointment. There may be several explanations for this. MMPIs act as cytostatic rather than cytotoxic compounds and thus measurements of radiologic response for assessment are not applicable. Their effectiveness has been shown mainly as effects on levels of tumour markers as surrogate markers of tumour response and by fibrotic stromal reactions especially in pancreatic cancer.149 In many clinical trials, MMPIs have been compared to placebo or standard chemotherapy.150 The efficacy of MMPIs is probably best in maintenance of remission after traditional chemotherapy or in combination with cytotoxic agents and perhaps with combinations of MMPIs. In addition, the best effects can perhaps be gained in adjuvant therapy or in early stages of cancer, when tumour burden is minimal.151 In situations with large tumour mass or rapid cancer progression, there can also be extensive ECM remodelling and thus a long-term or a high-dose administration of an MMPI would be needed to gain high enough local concentration for proper inhibition of tumour spread. The expression and activity of MMPs are especially abundant in the peritumoural area, where tumour angiogenesis also takes place. It has been shown with many MMPIs that their serum concentrations are high after oral administration,111 suggesting that they reach the peritumoural area to some extent. The use of different MMPIs with maximum tolerated doses are known to be limited with side effects such as arthralgias and cutaneous rash.86

Selective targeting of certain MMPs instead of using broad-spectrum inhibitors may be preferable, for example, due to the fact that certain MMPs (MMP-3, -7, -9 and -12) can inhibit angiogenesis by generating angiostatin from plasminogen and thereby suppress tumour growth.83–85 What is clearly needed are specific inhibitors and more information on the expression of distinct MMPs in different types of malignant tumours. With this combination it could be possible to choose the right inhibitor for a malignant tumour with a specific MMP profile.

In conclusion, there is a considerable body of evidence that MMPs play an important role in the invasion and growth of various malignant tumours and that the expression of certain MMPs can be used to estimate metastatic capacity and recurrence of malignant tumours as well as prognosis of patients. The clinical trials so far have failed to prove that the use of MMPIs can improve disease-free or overall survival of cancer patients. The ongoing phase III trials with synthetic and natural MMP inhibitors are expected to show whether the concept of MMP inhibition has a place in the therapeutic arsenal aimed at inhibiting growth, invasion and metastasis of malignant tumours.