The serine protease urokinase-type plasminogen activator (uPA) plays a key role in tumor-associated proteolysis in malignant solid tumors. Proteolytic activity of uPA is controlled by its naturally occurring plasminogen activator inhibitor type 1. As an initial observation, a correlation of enzymatic uPA activity in breast cancer cytosols with prognosis was described in 1988 (Duffy et al., Cancer (Phila.), 62: 531-533, 1988). A pronounced prognostic impact of uPA, independent of classical risk parameters, was then first demonstrated in detergent-extracted (Triton X-100) breast cancer tissues by applying enzyme-linked immunosorbent assay techniques (Jänicke et al., Lancet, 2: 1049, 1989; Fibrinolysis, 4:69-78, 1990; Duffy et al., Cancer Res., 50: 6827-6829, 1990). In addition, not only uPA but also plasminogen activator inhibitor type 1 were shown to be of prognostic value in breast cancer (Jänicke et al., Semin. Thromb. Hemostasis, 17: 303-312, 1991; Breast Cancer Res. Treat., 24: 195-208, 1993). Subsequently, the prognostic value of uPA and plasminogen activator inhibitor type 1 was also confirmed in studies using archived "cytosol fractions" of breast cancer tissues (Foekens et al., Cancer Res., 52: 6101-6105, 1992; Spyratos et al., J. Natl. Cancer Inst., 84: 1266-1272, 1992; Grondahl-Hansen et al., Cancer Res., 53: 2513-2521, 1993; Sumiyoshi et al., Int. J. Cancer, 50: 345-348, 1992). A direct comparison of both methods with regard to prognosis, however, was lacking. We therefore prepared both the detergent-treated tissue extracts and the cytosol fractions from the same breast cancer specimens to allow a direct comparison of both methods. In 247 breast cancer patients investigated, the Triton X-100-extracted tissues revealed about twice as much uPA antigen (uPATx: median, 2.32 ng/mg protein) than the cytosol fractions (uPAcyt: median, 1.07 ng/mg protein). In contrast, the presence of Triton X-100 did not result in an increase of PAI-1 (PAI-1Tx: median, 6.34 ng PAI-1/mg protein) compared to the cytosol fractions (PAI-1cyt: median, 7.15 ng PAI-1/mg protein). Good correlations between uPATx and uPAcyt (R = 0.72) and between PAI-1Tx and PAI-1cyt (R = 0.88) were observed. Furthermore, PAI-1 and uPA are moderately correlated with each other (uPATx versus PAI-1Tx: R = 0.40; uPAcyt versus PAI-1cyt: R = 0.39). The prognostic power of uPA showed its best advantage in Triton X-100-extracted tissues (RR = 3.22), most pronounced in the subgroups of node-negative and premenopausal patients, respectively.(ABSTRACT TRUNCATED AT 400 WORDS)