Proteinase inhibition in invasive cancer therapy: four control levels of matrix degradation (Larsen, 1993;Fontanini et al., 1996). Second, the understanding of the control of proteolysis at the invasive front of a malignant tumor is rapidly growing and is generating concepts with practical implications for the treatment of cancer.The concept that particular proteases, e.g., plasminogen activators, contribute to invasion and metastasis is not new (reviewed by Danø et al., 1985;Murphy et al., 1989;Liotta and Stetler-Stevenson, 1991;Mignatti and Rifkin, 1993). Like the blood-clotting cascade, remodeling of the extracellular matrix at the invasion front is governed by a strictly regulated cascade of proteolytic enzymes (Woessner, 1991;Matrisian, 1992). Serine proteases and metalloproteases (e.g., collagenases, stromelysins, gelatinases) are examples of such enzymes. In Figure 1 (Opdenakker and Van Damme, 1992a). A high local protease load may be obtained by concentration of proteases on specific membrane receptors (Blasi, 1988;Emonard et al., 1992) or by anchoring through an intrinsic hydrophobic transmembrane domain (Sato et al., 1994;Puente et al., 1996).Recent studies illustrate that intervention at each of these levels promises to become a new target for cancer therapy. First, transcriptional inhibition of gelatinase B production is possible (Houde et al., 1996;Tremblay et al., 1995;McMillan et al., 1996). Remarkably, in the study of McMillan et al. (1996), the anti-inflammatory theophylline derivative (CT-2519) Paemen et al., 1996) interleukin-8 and granulocyte chemotactic protein-2 (Proost et al., 1993), which stimulate gelatinase B release from recruited leukocytes and thus assist in the generation of a peritumoral protease load and in a countercurrent migration of cancer cells (Opdenakker and Van Damme, 1992b). Particular subsets of these chemokines possess angiogenic activities (Strieter et al., 1995). Inhibition of such chemokines may, predictably, result in lower protease secretion levels and might reverse the invasive phenotype. was also shown to inhibit tumor-cell invasion. Since many proteases depend on transcriptional induction by cytokines, we may expect more results, of relevance for tumor-cell invasion control, from studies on anti-inflammatory drugs and cytokineinhibitors. In line with this, the control of endotoxinemia in patients with invasive cancers may be crucial. Endotoxin is a strong inducer of matrix proteases such as gelatinase B, interstitial collagenase and stromelysin, and also an inducer of many cytokines which are known to induce these enzymes. At the levels of activation and enzyme inhibition, antiinflammatory drugs also play a role. For instance, the antirheumatic, D-penicillamine, was found to inhibit gelatinase B (and A) and-at a higher dose-to inhibit the 2 plasminogen activators. This implies that D-penicillamine might block the activation of gelatinase B by the inhibition of plasminogen activitor or gelatinase A (Norga et al., 1995). At the same time, Dpenicillamine dire...