Chemotherapy in N0 Breast Cancer Patients by Urokinase-type Plasminogen Activator

Abstract & Commentary

Synopsis: Among the approximately one half of breast cancer patients who are identified as lymph node negative at the time of breast surgery, nearly one third relapse distantly. Presently, traditional prognostic factors like primary tumor size, estrogen/progesterone receptor status, menopausal status, tumor grade, and histomorphology are used to form the basis for recommendations regarding adjuvant chemotherapy. However, use of such factors provides only a rough guide and does not enable clinicians consistently to pinpoint which node-negative women fall into the high-risk category that may benefit from adjuvant chemotherapy. This study from the German Chemo N0 Study Group stratified N0 women according to tumor levels of uPA and PAI-1 and then randomized them to chemotherapy or observation. They concluded that their study validated the use of these 2 tumor biologic factors for assessing risk, and that risk-adapted chemotherapy appears to offer a substantial benefit to women at high risk while enabling low-risk women to avoid chemotherapy.

Source: Janicke F, et al. J National Cancer Inst. 2001;93:913-920.

Janicke and colleagues recently reported the first planned interim results of the German multicenter Chemo-N0 trial. He was among those researchers about 10 years ago who first showed that tumor levels of urokinase-type plasminogen activator (uPA) and its inhibitor, plasminogen activator inhibitor type 1 (PAI-1), were prognostic factors for lymph node-positive and lymph node-negative breast cancer. His group noted that, by uPA and PAI-1 criteria, approximately 45% of lymph node-negative women belonged to the high-risk category, where high risk was defined as tumor uPA levels > 3 ng/mg of tumor protein and/or PAI-1 levels > 14 ng/mg tumor protein. The percentage of high-risk women derived from their prognostic model roughly correlates with the proportion of node-negative women who eventually relapse outside the breast. Janicke et al thus reasoned that a randomized prospective trial of risk-based chemotherapy might validate the use of these 2 tumor biologic factors as a new method for sorting out chemotherapy candidates, and that those candidates might benefit from systemic therapy.

From June 1993 through March 1997, 556 women younger than the age of 70 who underwent breast surgery at 14 centers in Germany and Slovenia were staged as pT1-2N0M0 based on assessment of > 10 lymph nodes and a negative-staging work-up. Three-hundred milligrams of fresh tumor tissue was snap-frozen for each patient and stored in liquid nitrogen. An ELISA assay was run and uPA and PAI-1 levels were determined. Based on previously published optimized cut-offs, high- and low-risk status was assigned to each patient. Patients were then offered trial participation. The trial was intended to have 86% power to detect a 30% reduction in distant recurrences based on enrollment of 900 patients. Whether a patient agreed to be randomized, her clinical course and progress were followed and recorded. The median age was 54 years (r 28-71), and median follow-up was 32 months (r 0-53). One-hundred and sixty women underwent modified radical mastectomy, and 396 selected breast conservation. Although patient accrual continued until December 1998, the protocol called for a first interim analysis 4.5 years after accrual was begun.

Among the 315 high-risk patients identified, 182 agreed to be randomized to 6 cycles of CMF chemotherapy vs. observation. Eighty-eight patients were randomized to CMF, and 94 to the observation arm. Eighty percent of the patients randomized to CMF received it, while 20% (18/88) did not. Of the 70 patients randomized to CMF who actually received it, 66 (94%) received all 6 cycles. Fourteen of the 94 high-risk patients randomized to observation (15%) received CMF anyway, and 72 of the remaining 80 observed patients (90%) completed the study. Two-hundred and eight low-risk patients were observed, and those high-risk patients who refused trial enrollment were followed. Eighteen percent of the high-risk patients who refused treatment on protocol went on to receive CMF chemotherapy off protocol (n = 24). Tamoxifen was prohibited.

Sixty patients among the 556 protocol candidates (11%) developed recurrences outside the breast. On an intention-to-treat basis, patients randomized to CMF had a 12% 3-year distant recurrence rate compared with an 18% recurrence rate in high-risk patients who were followed with observation only (P = NS). Analysis of the results based on actual treatment delivered showed that, among high-risk patients receiving CMF, the 3-year recurrence rate was 9% vs. 19% for those who did not receive the chemotherapy (P = 0.016). Multivariate analysis evaluated the effect of tumor grade, uPA levels, PAI-1 levels, surgical procedure used, age < or >50, T-stage, and ER/PR status. Only uPA, PAI-1, and grade were found to be statistically significant prognostic factors.

Despite a lack of statistical significance in the intention-to-treat analysis, Janicke et al concluded that their data provide compelling evidence favoring large-scale testing of uPA and PAI-1 in assignment of risk-based adjuvant chemotherapy. A total of 689 patients was accrued by December 1998 when Janicke et al closed protocol entry before their goal of 900 patients was reached. Accrual was ceased because of the conviction that the treatment effect will become statistically significant once all of the patient data are available for future interim analyses scheduled for 6.5, 8.5, and 10 years from the start of patient recruitment.

Comment by Edward J. Kaplan, MD

uPA is involved in degradation of the extracellular matrix during tumor cell invasion. It is responsible for the conversion of plasminogen to plasmin which facilitates stromal breakdown. In addition to the cited data presented by Janicke et al, uPA has been shown to be of prognostic value in other tumors such as ovarian cancer, colorectal cancer, prostate cancer, and gastric cancer.1-4 Other studies have also reported on its prognostic role in breast cancer. High uPA values predicted a poor prognosis in an evaluation of 878 breast cancers in one Dutch series, and in a Swedish series of 237 N0 patients.5,6 Janicke et al commented that contradictory data on uPA’s and PAI-1’s prognostic role in breast cancer has not been forthcoming, and I could not find any contrary data either. A meta-analysis of studies assessing the use of uPA and PAI-1 in breast cancer is currently being conducted by the EORTC, according to Malmstrom.6 In other tumor systems, such as non-small-cell lung cancer, data have been published that did not confirm any prognostic value for uPA or PAI-1 or 2.7

Although the German Chemo N0 Study Group data are intriguing, the early data did not show a significant benefit for risk-based chemotherapy administration using uPA and/or PAI-1 as prognostic factors. It is tempting to focus on the statistically significant results gleaned from the analysis by treatment delivered rather than intended treatment, but this is not a scientifically rigorous approach. Presumably, Janicke et al are working on their next planned interim analysis at 6.5 years for data accrued through December 1999. Meanwhile, a follow-up trial to determine the optimal chemotherapy protocol for high-risk N0 women, known as the Euro Chemo N0-European Node-Negative Breast Cancer Trial, has been approved. It involves anthracycline-containing regimens as well as taxanes, and uses tamoxifen for steroid hormone receptor-positive patients. uPA and PAI-1 may represent important prognostic factors for breast cancer and other tumors, but more studies are needed before these biologic factors can be adopted for clinical use.


1. Konecny G, et al. Clin Cancer Res. 2001;7:1743-1749.

2. Yang JL, et al. Int J Cancer. 2000;89:431-439.

3. Miyake H, et al. Prostate. 1999;39:123-129.

4. Plebani M, et al. Clin Exp Metastasis. 1997;15:

5. Witte JH, et al. Br J Cancer. 2001;85:85-92.

6. Malmstrom P, et al. J Clin Oncol. 2001;19:2010-2019.

7. Salden M, et al. Ann Oncol. 2000;11:327-332.