One of the problematic features of hepatocellular carci-Background. Matrix metalloproteinase 9 (MMP-9), a noma (HCC) is its rapid invasion into the portal venous sys-92-kd gelatinase/type IV collagenase, has been impli-tem. Even in small HCCs, microscopic vascular invasion is cated as playing an important role in cancer invasion found in more than 40% of the nodules. 1,2 Macroscopic portal and metastasis. A previous study showed that serum venous invasion is reported to be the worst single prognostic type IV collagenase activity correlated with metastasis factor, after partial hepatectomy, for HCC. 3 Thus the detec-by hepatocellular carcinoma (HCC). The aims of this tion of vascular invasion and its potential is important in study were to determine the plasma levels of immunore-determining therapy and prognosis. At present, there is no active MMP-9 in patients with HCC and to compare the clinical laboratory test to predict the potential for or the ongo-levels with the clinical features including vascular inva-ing activity of vascular invasion by HCC. Theoretically, the sion. Patients and Methods. This study included 100 pa-measurement of the plasma levels of specific proteins that tients with HCC, 21 patients with chronic hepatitis (CH), are released by cancer cells and that are necessary for the 24 patients with liver cirrhosis (LC), and 138 healthy invasive process is a feasible approach to predict the tumor control subjects. Plasma MMP-9 levels were measured behavior. with a specific one-step sandwich enzyme immunoassay. Invasive tumor growth and metastasis involve complex Results. Plasma MMP-9 levels in HCC (62 [33 to 130 ng/ steps. 4,5 The initial steps include the degradation of stromal mL] median [25%, 75%], 13 to 660 ng/mL, minimum, maxi-architecture and of basement membrane components, espe-mum) were significantly elevated compared with those cially type IV collagen. Two kinds of type IV collagenase in in normal controls (36 [25 to 45], range, 2.8-70 ng/mL), in the matrix metalloproteinases family, 6 matrix metallopro-CH (28 [18 to 30], 13 to 66 ng/mL) and in LC (35 [26 to teinase 9 (MMP-9) (92-kd gelatinase/type IV collagenase) and 58], 16 to 86 ng/mL) (P õ .0000001; P Å .0000003; and P Å matrix metalloproteinase 2 (MMP-2) (72-kd gelatinase/type .00205, respectively). When the cutoff level was defined IV collagenase), have been implicated as playing a major role as 60 ng/mL from a receiver operating characteristic in the degradation of the basement membrane in cancer inva-curve, plasma MMP-9 concentrations had a sensitivity sion and metastasis. 4,5 A correlation between the tumor secre-of 53% and a specificity of 89% for the detection of HCC tion of MMP-9 7 as well as MMP-2, 8 and experimental metas-from CH and LC. The levels were significantly higher in tasis has been reported. HCC patients with macroscopic portal venous invasion Previously, type IV collagenase activity (probably a mix-(79 [36 to 160], 15-660 ng/mL) than those without the inva-ture of MMP-2 and MMP-9 activity) in the tissue homogenate...