The prognostic impact of the proteolytic factors urokinase-type plasminogen activator (uPA) and plasminogen activator inhibitor type 1 (PAI-1) was evaluated in 76 completely resected gastric cancer patients enrolled in a prospective study. All patients underwent macroscopically and microscopically residual tumor-free resection (category Ro, Union International Contre Cancer, 1987). uPA and PAI-1 levels were quantified in detergent-extracted (Triton X-100) specimens of primary gastric tumors by enzyme-linked immmunosorbent assays. Median values of 1.57 ng uPA/mg protein were determined in tumor tissue extracts compared to 0.14 ng uPA/mg protein in normal mucosa. For PAI-1, 0.93 ng PAI-1/mg protein versus 0.09 ng PAI-1/mg protein was calculated. uPA levels in tumor tissue extracts were significantly correlated with vascular invasion, Laurén classification, and WHO classification, whereas PAI-1 levels showed a significant correlation with advanced lymph node involvement, depth of invasion, tumor stage, site of tumor, and the Laurén, Borrmann, and WHO classifications.
Elevated uPA and PAI-1 levels were found to be associated with poor prognosis. The optimal cutoff values indicating a group of patients with shorter survival were 1.5 ng uPA/mg protein and 1.25 ng PAI-1/mg protein, respectively (Classification and Regression Tree analysis). Patients with either high uPA or PAI-1 values were significantly associated with decreased survival (median time of survival was 23 months (high) versus 44 months (low). By univariate Cox regression analysis, it was shown that TNM categories, WHO classification, size of tumor, uPA and PAI-1 levels were all significantly associated with survival. However, in multivariate Cox regression analysis of these grouped variables, nodal status, PAI-1 levels, and WHO classification were the only independent prognostic factors. The relative risks of failure were 5-, 2.9-, and 2.4-fold, respectively. We conclude that PAI-1 and uPA positivity may serve as new prognostic factors in gastric cancer, predicting shorter survival even in clinically important subgroups of patients.
Supported by the Deutsche Forschungsgemeinschaft (Klinische Forschergruppe GR 280/4-1) and the Wilhelm-Sander Stiftung No. 92.016.01.