High level of urokinase‐type plasminogen activator is a new prognostic marker in patients with gastric carcinoma

Prognosis of gastric carcinoma is related to invasion and metastasis. Evidence has accumulated that invasion and metastasis in solid tumors require the action of tumor‐associated proteases, which promote the dissolution of the surrounding tumor matrix and the basement membrane. The serine protease urokinase‐type plasminogen activator (uPA) and its inhibitor, plasminogen activator inhibitor‐1 (PAI‐1), appear to have a major function in these processes. Recent reports have demonstrated that expression of these proteolytic enzymes is elevated in breast and colon carcinoma and that it can be associated with invasiveness and poor prognosis. Therefore, the authors evaluated whether the expression and activation of uPA and PAI‐1 might be of clinical value as a tumor/biologically defined risk factor in patients with gastric carcinoma.

have advanced beyond Stage III. 1,2 The majority of pa-Characteristics of Patients (n Å 160) tients in the treatment failure group die due to the direct effects of the metastasis or complications asso- vation of some procollagenases and can activate latent growth factors such as latent transforming growth factor-b. 8-10 Inhibition of uPA activity leads to inhibition of invasion in several experimental systems, 11-14 and The postoperative adjuvant chemotherapy was performed for advanced stage (Stages II, III, and IV) uPA is selectively expressed at invasive foci in some experimental and human cancers. [15][16][17] The activity of patients. Stage II and III patients were treated with 5fluorouracil and doxorubicin with or without nonspe-uPA is controlled not only through synthesis and secretion but also by its specific inhibitors: plasminogen cific immune stimulator OK-432, and Stage IV patients were treated with 5-fluorouracil, doxorubicin, and mi-activator inhibitor-1 (PAI-1) and -2 (PAI-2). Recently it has been proposed that PAI-1 plays a role in pro-tomycin C or 5-fluorouracil, etoposide, and cisplatin regimens with or without OK-432. tecting the tumor against degrading itself and is a potentially important prognostic factor in breast carci- The median follow-up was 24 months (maximum 48 months) at the time of the study. The follow-up noma. [18][19][20][21][22] The purpose of this study was to analyze the anti-was comprised of a clinical examination every 2 -3 months for the first 2 years and every 6 months there-gen levels of uPA and PAI-1 by means of an enzymelinked immunoadsorbent assay (ELISA) method and after. Recurrence, as confirmed by biopsy and/or other relevant diagnostic procedures, was defined as the ap-to evaluate their relative value in predicting disease free survival rates in patients with primary gastric car-pearance of new lesions in patients with no previous evidence of disease. At the time of analysis, 53 patients cinoma.
had recurred. Data regarding histopathologic tumor size, lymph node metastasis, and time to relapse were

Patient Characteristics
available for nearly all patients ( Table 1). The 160 patients chosen for this study at the Yonsei University Cancer Center (Seoul, Korea) were diag-Methods Tissue specimens nosed with primary gastric carcinoma between 1992 and 1995. These patients, whose ages ranged from 27 Immediately after resection, fresh tumor specimens and normal mucosa (both approximately 1 cm 3 ) were to 86 years (median, 54 years), all underwent resection with curative intent. Curative resection was defined selected, snap-frozen, and stored in liquid nitrogen. Cryostat sections (5-mm thickness) were prepared by the General Rules for the Gastric Cancer Study in Surgery and Pathology of the Japanese Research Soci-from the tumor specimens and normal mucosa, which were then stained with hematoxylin and eosin to con-ety for Gastric Cancer as 1) no involvement of surgical stumps; 2) sufficient lymphatic dissection (R number firm the presence or absence of tumor cells. Subsequently, 20 to 30 sections of each tumor and normal ¢ N number); 3) no distant metastasis; 4) removal of involved adjacent organs and structures by combined tissue sections were cut (total wet tissue weight was between 100 and 150 mg) for tissue extraction. The en bloc resection; and 5) no gross residual disease. 23 Of the 160 patients, 27 had distant lymph node metas-last section was also stained by hematoxylin and eosin to demonstrate the presence or absence of tumor cells. tasis (¢ N3 lymph nodes) and were regarded as M1 and Stage IV.
uPA and PAI-1 ELISAs uPA in the cytosolic extracts was measured by a sandwich ELISA, using a polyclonal catching antibody and a mixture of three different biotinylated monoclonal detecting antibodies, which in combination recognize free uPA as well as receptor and PAI-1-bound uPA.
The assay was performed as previously described. 21

Distribution of uPA and PAI-1 Levels
tively, in cancerous tissue. Both uPA and PAI-1 levels were significantly higher in cancerous tissue than nor- Figures 1 and 2 show the distribution of uPA and PAI-1 levels according to tumor progression in the cytosolic mal tissue (P õ 0.001, respectively) ( Table 2). extracts and the relationship between these two parameters. There was a strong positive correlation be-

Prognostic Significance of uPA and PAI-1 Levels
The prognostic significance of uPA and PAI-1 levels tween uPA and PAI-1 levels in the cytosols (P õ 0.001, correlation coefficient Å 0.57).
was studied by univariate analysis for each of the pa-/ 7b4e$$0917 02-06-97 00:47:36 canal W: Cancer relapse free survival rate was significantly lower in the FIGURE 3. Relapse free survival of the gastric carcinoma patients ac-uPA and PAI-1 high expression groups than in the low cording to urokinase-type plasminogen activator (uPA) expression. expression groups (P Å 0.005 and P Å 0.004, respectively).

Correlation of uPA and PAI-1 with Other Prognostic Variables
The correlation between the cytosolic uPA and PAI-1 levels and other prognostic variables was studied using the chi-square test (Table 3). A high grade of anaplasia was significantly associated with high uPA and PAI-1 (P Å 0.04 and P Å 0.004, respectively) and an elevated PAI-1 level was associated with lymph node metastasis and advanced stage (P Å 0.003 and P Å 0.04, respectively).

Multivariate Analysis FIGURE 4. Relapse free survival of the gastric carcinoma patients ac-
To compare the prognostic significance of cytosolic cording to plasminogen activator inhibitor-1 (PAI-1) expression.
uPA and PAI-1 levels with that of other parameters, variables were singly eliminated from the model in a backwards fashion and reincluded only if the P value was õ 0.05. uPA level was a statistically significant rameters for all patients. At the time of the study 53 patients had experienced recurrence. The relapse free independent variable parameter for relapse free survival, with a relative risk of 3.6, but the PAI-1 level was survival was compared for patients with both low and high expression of uPA and PAI-1. The cutoff value of a stage-dependent prognostic factor ( Table 4).
The authors subsequently analyzed the prognostic both uPA and PAI-1 were defined as the mean value of all normal tissue cytosol levels of uPA and PAI-1 / significance of high uPA expression according to disease stage. Subgroup analysis according to each stage 2X standard deviation at 11.5 and 11.6 ng/mg protein cytosol, respectively. As seen in Figures 3 and 4   sive potential by using a comparison of the uPA level this study will answer this question. The current investigation also reveals that the PAI-1 antigen level of tumor tissue extracts as determined by ELISA is higher in gastric carcinoma than in normal stage patients (Stage II-IV); the relapse free survival of the patients with high uPA expression was lower than tissues. The presence of PAI-1 in gastric carcinoma at a level correlating with that of uPA suggests that PAI-those with low uPA expression (P Å 0.037, P Å 0.048, and P Å 0.039, respectively). However, there was no 1 may play a role in the regulation of plasminogen activation, and thereby in the process of gastric carci-significant difference between the two groups in Stage I patients (survival curves not included).
noma invasion and metastasis.
Recently, Soff et al. reported that PAI-1 expression in human prostate carcinoma cells reduced primary

DISCUSSION
The involvement of uPA in the invasion and metastatic tumor growth and tumor-associated microvasculature, and resulted in a tenfold inhibition of lung metas-mechanisms of gastric carcinoma is well documented, 8-12 but the exact role of uPA and PAI-1 is less tasis and a significant inhibition of liver metastasis. 27 However, these findings are quite contradictory to the well known. To better analyze their prognostic value, it is important to evaluate the relative importance of previous hypothesis of PAI-1 function in vivo. Mapping of the uPA-and PAI-1-producing cells in gastric these parameters and their interactions. To this end, the uPA and PAI-1 antigen levels were measured in a carcinoma by in situ hybridization will, in combination with immunohistochemical stainings, shed more series of paired 160 gastric normal and tumor specimens and the relationships to their prognostic value light on this issue.
In conclusion, uPA can be added to the prognostic studied. The data support the notion that the uPA antigen content of the tumor tissue is an independent and markers in gastric carcinoma patients. However, further study with plasma antigen levels for detection of strong prognostic factor in gastric carcinoma patients. Because uPA antigen is an independent prognostic early recurrence or metastasis is needed, suggesting that pharmacologic or molecular inhibition of uPA can variable, subgroups of patients at high risk for recurrence (even within the risk groups that were until now be a new target for antiinvasion and antimetastasis therapy. ease. 26 Therefore, it is more accurate to test the inva-/ 7b4e$$0917 02-06-97 00:47:36 canal W: Cancer