Prognostic Impact of Proteolytic Factors

Vol. 7, 2757–2764, September 2001 Clinical Cancer Research 2757

Prognostic Impact of Proteolytic Factors (Urokinase-Type Plasminogen Activator, Plasminogen Activator Inhibitor 1, and Cathepsins B, D, and L) in Primary Breast Cancer Reflects Effects of Adjuvant Systemic Therapy1

Nadia Harbeck,2 Uta Alt, Ursula Berger, 2.5). For all proteolytic factors but Cathepsin L, the uni- Achim Kru¨ger, Christoph Thomssen, variate prognostic impact on DFS was substantial in Fritz Ja¨nicke, Heinz Ho¨fler, Ronald E. Kates, and patients without adjuvant systemic therapy but was diminished if adjuvant therapy had been administered. Manfred Schmitt Cathepsin L maintained its strong prognostic impact on Frauenklinik und Poliklinik, Klinikum rechts der Isar [N. H., U. A., DFS even in patients with adjuvant endocrine therapy .(RR 2.8 ;0.01 ؍ R. E. K., M. S.], Institut fu¨r Medizinische Statistik und Epidemiologie (P (IMSE) [U. B.], Institut fu¨r Experimentelle Onkologie und Therapieforschung [A. K.], and Institut fu¨r Pathologie [H. H.], Conclusions: The observed effect of adjuvant systemic Technische Universita¨t Mu¨nchen, D-81675 Munich, and therapy on their prognostic strength suggests that the as- Universita¨tsfrauenklinik Eppendorf, D-20246 Hamburg [C. T., F. J.], sessed proteolytic factors supply predictive information on Germany therapy response.

ABSTRACT INTRODUCTION Purpose: Prognostic and predictive impact of five pro- Invasion of a malignant tumor into the surrounding extra- teolytic factors associated with tumor invasion and metas- cellular matrix and the basement membrane is a prerequisite for tasis in primary breast cancer were evaluated after long- subsequent systemic spread and metastasis. In breast cancer, term follow-up. distant metastasis is of great clinical relevance because patients Experimental Design: Antigen levels of urokinase-type presenting with distant disease recurrence cannot be cured by plasminogen activator, plasminogen activator inhibitor-1 currently available therapies. Tumor-associated proteolytic fac- (PAI-1), Cathepsins B, D, and L were determined using tors enable tumor cells to disintegrate the stroma in their imme- immunochemical assays in primary tumor tissue of 276 diate vicinity, intravasate into lymphatic or blood vessels, and patients. then spread systemically. Among the key players in the proteo- Results: During follow-up (median 109 months), 119 lytic cascade leading to tumor invasion and metastasis are (43%) patients relapsed, and 117 (42%) died. In the whole factors of the plasminogen activation system, such as the serine 3 collective, lymph node status (P < 0.001; RR 3.8), Cathepsin protease uPA and its type-1 inhibitor PAI-1, or the aspartyl RR 1.7) were protease Cathepsin D and the cysteine proteases Cathepsins B ;0.027 ؍ L(P < 0.001; RR 2.6), and PAI-1 (P the only independent significant factors in multivariate anal- and L (1–3). ysis for disease-free survival (DFS). For overall survival In the late 1980s, the first reports on the clinical relevance (OS), lymph node status (P < 0.001; RR 2.9), Cathepsin L of proteolytic factors in human cancer were published. In 1988, RR 1.9), and grading Duffy et al. (4) reported that in breast cancer, an increased ;0.01 ؍ RR 1.9), PAI-1 (P ;0.017 ؍ P) RR 1.7) were significant. In the node-negative enzymatic activity level of uPA in the tumor tissue is associated ;0.026 ؍ P) subgroup, PAI-1 was the only significant factor for DFS with poor prognosis. In 1989, two groups independently found RR that an increased tumor tissue content of Cathepsin D is also ;0.004 ؍ RR 3.7) and the strongest factor (P ;0.004 ؍ P) RR 3.1). In node- linked to shorter patient survival in breast cancer (5, 6). Since ;0.017 ؍ for OS next to grading (P (3.7 positive patients, Cathepsin L was the only significant factor then, numerous reports have discussed the clinical impact of RR proteases in primary breast cancer. In the early 1990s, our group ;0.003 ؍ for both DFS (P < 0.001; RR 3.2) and OS (P was the first to report that not only the protease uPA but also its inhibitor PAI-1 display significant prognostic strength in node- positive and node-negative breast cancer (7, 8). Convincing confirmatory data by many international research groups have Received 2/9/01; revised 6/5/01; accepted 6/26/01. since been put forward associating high levels of uPA and/or The costs of publication of this article were defrayed in part by the PAI-1 with shorter DFS and OS in primary breast cancer (9– payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported in part by a grant to N. H. by the State of Bavaria (KKF Project 8756159). 2 To whom requests for reprints should addressed, at Frauenklinik und 3 The abbreviations used are: uPA, urokinase-type plasminogen activa- Poliklinik, Klinikum rechts der Isar, Technische Universita¨t Mu¨nchen, tor; DFS, disease-free survival; OS, overall survival; CART, classifica- Ismaninger Strasse 22, D-81675 Munich, Germany. Phone: 49-89-4140- tion and regression trees; PAI, plasminogen activator inhibitor; ER, 2419; Fax: 49-89-4140-7410; E-mail: [email protected]. estrogen receptor; RR, relative risk; PR, progesterone receptor.

17). In addition, a prospective randomized multicenter therapy Table 1 Patient collective (n ϭ 276) trial (“Chemo N0”) validated the independent and significant Prognostic factor Patientsa prognostic impact of uPA and PAI-1 in node-negative breast Lymph node status Negative 130 (47%) cancer patients (18). Positive 146 (53%) Next to uPA and PAI-1, the aspartyl protease Cathepsin D Tumor size Յ2 cm 89 (32%) is the proteolytic factor that has most frequently been implicated Ͼ2 cm 186 (68%) as a prognostic factor in primary breast cancer. A significant Grading 1/2 172 (63%) 3/4 103 (37%) impact of high Cathepsin D antigen levels in the primary tumor Hormone receptor status Negative 59 (21%) on patient prognosis has emerged in the literature over the last Positive 217 (79%) decade (5, 6, 19–22). First data on the prognostic impact of the Menopausal status Pre/perimenopausal 105 (38%) cysteine proteases Cathepsins B and L in human breast cancer Postmenopausal 171 (62%) were put forward by our group in 1995, when we reported that a For some patients, not all information was available. high levels of both proteases were associated with poor patient survival (23). Confirming this initial finding, Foekens et al. (24) showed in a large collective of 1500 breast cancer patients that high levels of both Cathepsin B and L in primary tumor cytosols Laboratory Assays. Tissue extraction for determination were correlated with poor prognosis. of proteolytic factors was performed as follows: deep frozen Although efficient tumor-associated proteolysis requires tumor tissue was pulverized, and the resulting powder was dispersed in buffer [1 ml of Tris-buffered saline, 0.02 M Tris- complex interactions between various extra and intracellular HCl, and 0.125 M NaCl (pH 8.5)] containing 0.1% (volume for proteases and their inhibitors (25), up to now, a relatively few volume) of nonionic detergent Triton X-100. Cytosol fractions authors have looked at the prognostic impact of factors from were prepared accordingly but without detergent. uPA and different proteolytic systems within the same patient collective PAI-1 antigen have been measured in detergent extracts by in breast cancer or other solid tumors (15, 26–28). Even more ELISA (uPA: Imubind # 894, PAI-1: Imubind # 821; American important, data on the predictive value of proteolytic factors, Diagnostica, Greenwich, CT) since 1987 in a prospective fash- such as uPA, PAI-1, or Cathepsins B, D, and L, are still scarce. ion in all breast cancer patients who had their primary surgery We have therefore evaluated the influence of adjuvant systemic performed at our institution (8). Antigen levels ranged for uPA therapy on the prognostic impact of these five different proteo- from 0.07 to 15.17 ng/mg protein (median 2.17) and for PAI-1 lytic factors within the same collective of primary breast cancer from 0.06 to 167.29 ng/mg protein (median 7.69). Antigen Ͼ patients after a long-term median follow-up of 9 years. levels of Cathepsins B, D, and L were determined in breast cancer cytosol fractions obtained from the same tissue speci- MATERIALS AND METHODS mens. Cathepsin D was measured by a radiometric immunoassay (ELSA-CATH-D, CIS Bioindustries, Gif-sur-Yvette, Patients. In a collective of 276 patients with primary France); Cathepsins B and L were measured by ELISA (Diag- breast cancer, the clinical relevance of uPA and PAI-1, as well nostic International, Karlsdorf, Germany; Ref. 23). Levels as Cathepsins B, D, and L, was evaluated. Informed consent for ranged for Cathepsin B from 46 to 5934 ng/mg protein (median determination of tumor-biological factors was obtained before 855), for Cathepsin D from 5.22 to 271.93 pmol/mg protein primary surgery. Patients either had a modified radical mastec- (median 43.39), and for Cathepsin L from 51 to 4619 ng/mg ϭ tomy (n 214) or underwent breast-conserving surgery with protein (median 429.5). ϭ subsequent irradiation of the breast (n 62) at the Department Steroid hormone receptor status (ER and progesterone re- of Obstetrics and Gynecology of the Technische Universita¨t ceptor) was determined and classified as “positive” or “nega- Mu¨nchen, Germany, between 1987 and 1991. Treatment deci- tive” as described (29). sions were based solely on consensus recommendations at the Statistical Analysis. Correlations between continuous time of recruitment. Of the total 276 patients, information about variables were analyzed using the Spearman rank test. Associ- adjuvant systemic therapy was missing from 1 patient; 52 pa- ations between continuous and/or categorical variables were tients received adjuvant chemotherapy, 95 patients received analyzed using the Mann-Whitney U test. Continuous variables adjuvant endocrine therapy, 9 patients received combined were coded as binary variables by using Log-rank statistics to chemo-endocrine therapy, and 119 patients did not receive any determine optimal cutoffs discriminating low- and high-risk adjuvant systemic therapy after primary surgery. Of the 130 patients (8, 30). For univariate analysis of DFS and OS, Kaplan- node-negative patients, 110 (85%) did not receive any adjuvant Meier curves (31) were plotted and then compared using Log- systemic therapy. The median age of the patients at the time of rank statistics. Multivariate analyses were performed in a step- primary surgery was 56 years (range: 33–87 years). Additional wise forward fashion by applying the Cox proportional hazards patient characteristics are displayed in Table 1. At time of model (32) using the SPSS software package (SPSS, Inc., Chi- primary therapy, none of the patients had any clinical or X-ray cago, IL) and by CART analysis (33). The cutoffs determining evidence of distant metastases. Follow-up data were obtained in binary variable coding were maintained during CART analysis. regular intervals (17). Median follow-up time of patients still All test decisions were performed at a significance level of ␣ϭ alive at time of analysis was 109 months (range: 22–142 0.05. Estimated values are provided with a 95% confidence months). Within the follow-up period, 119 patients (43%) ex- interval. perienced disease recurrence, and 117 patients (42%) died. To test for a possible time variation of the prognostic

Table 2 Correlations with proteolytic factors and others a. Correlations among proteolytic factors uPA PAI-1 Cathepsin B Cathepsin D Cathepsin L uPA PAI-1 r ϭ 0.390 P Ͻ 0.001 Cathepsin B n.s.a r ϭ 0.155 P ϭ 0.052 Cathepsin D r ϭ 0.251 r ϭ 0.233 r ϭ 0.404 P Ͻ 0.001 P Ͻ 0.001 P Ͻ 0.001 Cathepsin L r ϭ 0.198 r ϭ 0.129 r ϭ 0.335 r ϭ 0.271 P ϭ 0.005 P ϭ 0.072 P Ͻ 0.001 P Ͻ 0.001

b. Correlations and associations between proteolytic and established factorsb Lymph Hormone node status Tumor size Grading ER PR receptor-status uPA n.s. r ϭ 0.131 r ϭϪ0.180 r ϭϪ0.147 P ϭ 0.03 P ϭ 0.006 P ϭ 0.004 P ϭ 0.018 P ϭ 0.038 PAI-1 n.s. r ϭ 0.138 r ϭϪ0.166 r ϭϪ0.186 P ϭ 0.022 P Ͻ 0.001 P ϭ 0.007 P ϭ 0.003 P Ͻ 0.001 Cathepsin B n.s. n.s. n.s. n.s. n.s. P ϭ 0.009 Cathepsin D n.s. r ϭ 0.140 n.s. n.s. n.s. P ϭ 0.022 P ϭ 0.039 Cathepsin L n.s. n.s. n.s. r ϭϪ0.139 r ϭϪ0.183 P ϭ 0.062 P ϭ 0.013 P ϭ 0.027 a n.s., not significant. b No significant associations were found between proteolytic factors and menopausal status.

influence (relative risk) of the factors (30) and thus detect a general, similar to the correlations among proteolytic factors, no violation of the proportional hazards assumption, the residual strong correlations were observed between proteolytic and es- score tests of Grambsch and Therneau (34) were applied with tablished factors (Table 2b). linear, logarithmic, and quadratic transformations of time. The Univariate and Multivariate Survival Analysis. In uni- Cox model with time-varying effects was determined using variate analysis of the whole collective, uPA, PAI-1, and Cathep- smoothing spline estimates as described by Hastie and Tib- sins D and L were statistically significant for both DFS and OS. shirani (35). These analyses were performed using the authors’ Cathepsin B was not significant, neither for DFS nor OS. Of the software libraries implemented for the statistical programming traditional prognostic factors, lymph node status, grading, and environment S-Plus (MathSoft 1998). tumor size were significant for both DFS and OS, whereas steroid hormone receptor status was significant only for OS. Menopausal RESULTS status failed univariate statistical significance for both DFS and OS. Cutoff Determinations. For Cathepsin D (41 pmol/mg In multivariate analysis (Cox model) for DFS, lymph node status protein) and Cathepsin L (360 ng/mg protein), optimized cutoffs (P Ͻ 0.001; RR 3.8), Cathepsin L (P Ͻ 0.001; RR 2.6), and PAI-1 were determined using Log-rank statistics. For Cathepsin B, the (P ϭ 0.027; RR 1.7) were significant factors in all patients; Ca- median of 855 ng/mg protein was taken as a cutoff because an thepsin D and uPA were not. For OS, lymph node status (P Ͻ optimized cutoff could not be found. For uPA (3 ng/mg protein) 0.001; RR 2.9), Cathepsin L (P ϭ 0.017; RR 1.9), and PAI-1 (P Ͻ and PAI-1 (14 ng/mg protein), previously calculated and reeval- 0.01; RR 1.9) were again statistically significant next to grading uated optimized cutoffs were used (17). Established prognostic (P ϭ 0.026; RR 1.7). factors were dichotomized as described elsewhere (30). Multivariate Cox analysis in node-negative patients (n ϭ Correlations and Associations between Proteolytic 130) showed that PAI-1 was the only significant factor for DFS Factors and Established Factors. Except for uPA and Ca- (P ϭ 0.004; RR 3.7) and the strongest significant factor (P ϭ thepsin B, all proteolytic factors were correlated with each other 0.004; RR 3.7) for OS next to grading (P ϭ 0.017; RR 3.1). In (Table 2a). However, strong correlations (exceeding r ϭ 0.5) node-positive patients (n ϭ 146), Cathepsin L was the only between factors were not observed. With regard to correlations significant factor both for DFS (P Ͻ 0.001; RR 3.2) and OS between proteolytic and established factors, it is interesting that (P ϭ 0.003; RR 2.5). All of the other factors failed statistical Cathepsin B was the only proteolytic factor that was signifi- significance in this subgroup. The qualitative weighting of the cantly associated with nodal status. In node-negative patients, a factors in our subgroup analyses for DFS using the Cox model mean Cathepsin B value of 1256 ng/mg protein was determined is consistent with the picture obtained by another multivariate in primary tumor tissue compared with 881 ng/mg protein in model, CART analysis. In the CART model, lymph node status node-positive patients. There was no significant correlation be- was the strongest discriminator between high-risk and low-risk tween Cathepsin B and the number of involved lymph nodes. In patients (P Ͻ 0.001). Among node-negative patients, PAI-1 was

Table 3 Univariate Cox analysis for DFS within patient subgroups defined by adjuvant systemic therapy: relative risks (95% confidence intervals) for relapse associated with each of the five proteolytic factors Prognostic factors No adjuvant therapy (n ϭ 119) Adjuvant chemotherapy (n ϭ 52) Adjuvant hormone therapy (n ϭ 95) uPA P ϭ 0.008 RR 2.4 (1.3–4.7) P ϭ 0.060 RR 2.0 (0.97–3.9) P ϭ n.s.a RR 1.4 (0.8–2.6) PAI-1 P Ͻ 0.001 RR 4.3 (2.1–8.4) P ϭ n.s. RR 1.5 (0.8–3.1) P ϭ n.s. RR 1.6 (0.8–3.1) CathepsinB P ϭ 0.085 RR 2.7 (0.9–8.1) P ϭ n.s. RR 1.7 (0.7–4.3) P ϭ n.s. RR 1.3 (0.6–2.9) CathepsinD P ϭ 0.001 RR 4.0 (1.8–9.2) P ϭ n.s. RR 1.2 (0.6–2.3) P ϭ 0.061 RR 1.9 (0.97–3.6) CathepsinL P ϭ 0.030 RR 3.0 (1.1–8.0) P ϭ 0.068 RR 2.5 (0.9–6.9) P ϭ 0.01 RR 2.8 (1.3–6.1) a n.s. ϭ not significant.

the strongest prognostic factor for DFS (P Ͻ 0.001). Node- Kaplan-Meier curves for DFS of the two strongest proteolytic negative breast cancer patients whose tumors had low levels of factors, PAI-1 and Cathepsin L, in patient subgroups defined by both PAI-1 and Cathepsin D had a 5-year relapse rate of 9.3% administration of adjuvant systemic therapy. compared with 31.2% in patients with either or both factors After merging the two therapy groups (chemotherapy and high. Among the node-positive patients, Cathepsin L was the endocrine therapy) to increase the patient number in the “treat- strongest prognostic factor (P ϭ 0.001). ment category,” only Cathepsin L (P ϭ 0.001; RR 2.7), but none Time Varying Analysis. No significant time variation of of the other proteolytic factors, was significant in univariate Cox the prognostic impact on DFS was found for either lymph node analysis for DFS in patients who received adjuvant systemic status, uPA, or for any of the cathepsins using the univariate test therapy. of Grambsch and Therneau (34). In contrast, for PAI-1, the previously described (29) time variation in its effect on DFS (P ϭ 0.025 for linear time dependency) was confirmed even DISCUSSION after prolonged time of follow-up. Modeling the univariate To date, once breast cancer has recurred after primary relative risk of PAI-1 by smoothing splines, one finds that in the therapy, it cannot be cured by currently available therapy regi- first 2.5 years after surgery, its prognostic impact is higher mens. Thus, identification of patients at high risk for recurrence Ϸ (RRVar 3.4) than that determined by a proportional hazards after primary surgery is a prerequisite for individualizing con- ϭ model (RRPH 2.30) but that it declines steadily over time. The cepts for adjuvant systemic therapy in primary breast cancer. main contribution to the relative risk associated with PAI-1 is Proteolytic factors are essential for a tumor’s invasive and attributed to the first 5 years of follow-up. Hence, we recalcu- metastatic capacity and have thus been implicated as clinically lated a multivariate survival model for DFS, including those relevant prognostic factors in a variety of solid tumors, includ- covariates already selected in the proportional hazards model ing breast cancer. To assess and rank their clinical usefulness, (lymph node status, Cathepsin L, and PAI-1), but also taking we have studied the prognostic and predictive impact of five into account time variation for PAI-1. This multivariate model different proteolytic factors, uPA, PAI-1, and Cathepsins D, B, resulted in a better fit, whereas the (constant) impact of lymph and L, in a single collective of primary breast cancer patients node status and Cathepsin L rendered comparable results after long-term follow-up of Ͼ9 years. ϭ ϭ (RRLNstatus 3.32, RRCath-L 2.46). As for traditional prog- All five proteolytic factors were determined in primary nostic factors, the time-dependent effect of hormone receptor tumor tissue extracts using immunochemical assays. For uPA status (29) was confirmed (P Ͻ 0.001 for logarithmic time and PAI-1 (36), as well as for the cathepsins (1, 2, 37), the assay dependency). In addition, a time-varying impact was noted for procedures are standardized, and international quality control grading (P ϭ 0.034 for logarithmic time dependency). data are available. In the present investigation, clinically vali- Effect of Adjuvant Systemic Therapy on Prognostic dated cutoffs for uPA and PAI-1 were used (17, 18), whereas Impact of Proteolytic Factors. The prognostic impact of the cutoff values for the cathepsins were reevaluated. The cutoffs proteolytic factors on DFS and, in particular, that of PAI-1 and for Cathepsins D and L have remained remarkably stable after Cathepsin L differed rather substantially between node-negative prolonged follow-up, whereas no optimized cutoff was found and node-positive patients. These patient subgroups differed for Cathepsin B. For Cathepsin D, our optimized cutoff corre- substantially with regard to the frequency of adjuvant systemic sponded well to that found by other researchers using the same therapy. We therefore looked at the influence of adjuvant sys- immunochemical assay (17, 22, 38). For Cathepsin L, we con- temic therapy on prognostic impact (DFS) of proteolytic factors firmed our previously optimized cutoff of 360 ng/mg protein using univariate Cox analysis (Table 3). In patients who did not (26). The observed significant correlations between tumor tissue receive any adjuvant systemic therapy, uPA, PAI-1, and Cathep- levels of the different proteolytic factors are consistent with the sins D and L had a significant and Cathepsin B a borderline known complex interactions between various extra and intracel- significant prognostic impact. In patients who received adjuvant lular proteases and their inhibitors (25). systemic therapy, the prognostic significance disappeared for all After a long-term follow-up, the tumor-associated proteo- factors except Cathepsin L, which retained a significant prog- lytic factors uPA and PAI-1, as well as Cathepsins D and L, still nostic impact on DFS even in patients who received adjuvant provided significant prognostic information in primary breast hormone therapy. This impact was just as strong as in the cancer patients, whereas Cathepsin B did not retain its previ- patients who did not receive adjuvant therapy. Fig. 1 shows the ously reported prognostic impact on DFS (23, 24, 38). In our

Fig. 1 a, effect of adjuvant systemic therapy on prognostic impact of PAI-1 on DFS in primary breast cancer. With regard to DFS, events are relapses (all sites). b, effect of adjuvant systemic therapy on prognostic impact of Cathepsin L on DFS in primary breast cancer. With regard to DFS, events are relapses (all sites).

whole patient collective, PAI-1 outperformed uPA and Cathep- relevant risk group assessment, in particular with regard to sin D in multivariate analysis for DFS in primary breast cancer, identification of low-risk node-negative patients. Our findings although all three factors were significant in univariate analysis. after long-term follow-up are in agreement with data by Kute et This finding is in concordance with other retrospective studies al. (15) that the combination of PAI-1 and Cathepsin D is well (9, 10). The strong prognostic impact of the individual factors is suited for identifying node-negative breast cancer patients at enhanced by a combination of factors, thus allowing clinically low risk of recurrence. In a cohort of node-negative breast

cancer patients without any adjuvant systemic therapy, we had A beneficial effect of adjuvant tamoxifen in patients with demonstrated previously that with regard to selection of low- elevated Cathepsin D tumor levels has also been observed in risk patients, another combination of proteolytic factors, i.e., other retrospective analyses (20, 45). It also agrees well with the uPA and PAI-1, is superior to either factor taken alone as well fact that estrogens do induce Cathepsin D overexpression in as to established prognostic or other tumor-biological factors ER-positive cell lines (1). (30). The clinical relevance of this combination, uPA and PAI-1, Our data support the conclusion that high-risk primary for risk group stratification in node-negative breast cancer has breast cancer patients as identified by high levels of uPA, PAI-1, already been validated by the German “Chemo N0” trial in a and Cathepsins B and D do benefit from adjuvant systemic prospective randomized multicenter fashion (18). therapy with regard to DFS. Whether this benefit extends to OS In our subgroup of node-negative breast cancer patients, needs to be evaluated prospectively considering that systemic PAI-1 maintained its role as the strongest prognostic factor for treatment after first relapse is quite heterogeneous, thus ham- DFS, whereas in node-positive patients, Cathepsin L was the pering retrospective analysis. strongest. Interestingly, these nodal subgroups also differed Concerning Cathepsin L, our data and the above discussion substantially with regard to the frequency of adjuvant systemic indicate that the strong prognostic impact of Cathepsin L in therapy. After general recommendations at the time of recruit- node-positive patients is in fact attributable to its particular ment, the majority (85%) of our node-negative patients did not prognostic strength in patients treated by adjuvant systemic receive any adjuvant systemic therapy, whereas in node-positive therapy. This finding suggests that high-risk patients as identi- patients, only 6% remained untreated after primary locoregional fied by Cathepsin L do not necessarily benefit from adjuvant therapy. For all of the proteolytic factors assessed, the univariate systemic treatment. This statement holds in particular for the prognostic impact on DFS was substantial in patients without patient group treated by tamoxifen. adjuvant systemic therapy (Table 3). This underlines the strong Different tumor-biological properties of the proteolytic fac- association of the proteolytic factors uPA, PAI-1, and Cathep- tors studied here may well be responsible for their different sins D, B, and L with an aggressive tumor phenotype leading to prognostic and predictive impact, although in vitro data do not invasion and metastasis (2, 3). However, except for Cathepsin L, necessarily reflect the in vivo situation. Whereas lysosomal the prognostic strength of the proteolytic factors is diminished in cathepsins are predominantly involved in degradation of the patients who received adjuvant systemic therapy. This dimin- extracellular matrix (2), proteolytic factors uPA and PAI-1 are ished univariate significance in patients with adjuvant systemic therapy might be attributed to their node-positive status. How- also associated with tumor cell migration or angiogenesis (3). ever, in a different patient collective in which about half of the Yet, how these biological differences translate into differential node-positive patients did not receive any adjuvant systemic therapy response has not been studied up to now and is an therapy, uPA and PAI-1 had a strong, significant prognostic interesting subject for additional experimental research. impact that was similar for node-negative and node-positive In conclusion, the present study clearly demonstrates a patients (39). Hence, at least with regard to uPA and PAI-1, the substantial influence of adjuvant systemic therapy on the prog- apparent dichotomy in prognostic impact of proteolytic factors nostic impact of proteolytic factors. Thus, the apparent dichot- with respect to nodal status is best explained by the effect of omy in their prognostic strength within node-negative or node- adjuvant systemic therapy. This explanation is also consistent positive patient subgroups may merely reflect effects of with the well-established evidence that the relative risk reduc- adjuvant systemic therapy and hence indicate a predictive po- tion achieved by adjuvant therapy is independent of nodal status tential for therapy response. Consequently, the influence of (40, 41). In addition, there was no significant association be- adjuvant systemic therapy needs to be accounted for in any tween tissue levels of proteolytic factors and nodal involvement prognostic factor study to put the results into a clinically valid in our study, except for Cathepsin B. perspective. Prospective studies addressing the predictive im- For uPA and PAI-1, the apparent benefit of adjuvant sys- pact of proteolytic factors in primary breast cancer are still temic therapy in our retrospective study agrees with data from scarce. However, in view of the convincing data on significant and clinically relevant risk group selection based on proteolytic the prospective randomized multicenter “Chemo N0” therapy trial demonstrating benefits from adjuvant CMF chemotherapy factors, such as uPA, PAI-1, or Cathepsin D or L, such prospec- in node-negative patients with high uPA and/or PAI-1 levels in tive data are urgently needed to individualize adjuvant therapy their primary tumors (18). A European follow-up study will now accordingly. The evidence in this paper suggests that there could compare different chemotherapy regimens for these high-risk be a group of patients that are not at all, or only temporarily, patients. In another prospective study, no correlation was found benefiting from standard adjuvant therapy regimens. These pa- between PAI-1 levels in core biopsies of the primary tumor and tients would either be candidates for “more aggressive” chemo local response to anthracycline containing neoadjuvant chemo- or endocrine regimens or for therapies specifically targeted therapy (42). In that study, PAI-1 tumor levels were not altered toward the invasive phenotype of their tumor (3). by chemotherapy; however, the authors did not present any data on correlation of PAI-1 levels with patient outcome. Retrospec- REFERENCES tive data from metastatic breast cancer, that high PAI-1 in the 1. Rochefort, H., and Cathepsin, D. Breast cancer. a tissue marker primary tumor is associated with poor response to palliative associated with metastasis. Eur. J. Cancer, 28: 1780–1783, 1992. endocrine therapy (43, 44), cannot be directly compared with 2. Kos, J., and Lah, T. T. Cysteine proteinases and their endogenous data derived from the adjuvant setting and additionally under- inhibitors: target proteins for prognosis, diagnosis and therapy in cancer lines the need for prospective studies. (review). Oncol. Rep., 5: 1349–1361, 1998.

3. Schmitt, M., Wilhelm, O. G., Reuning, U., Kru¨ger, A., Harbeck, N., 18. Ja¨nicke, F., Prechtl, A., Thomssen, C., Harbeck, N., Meisner, C., Lengyel, E., Graeff, H., Ga¨nsbacher, B., Kessler, H., Bu¨rgle, M., Stu¨r- Untch, M., Sweep, C. G., Selbmann, H. K., Graeff, H., and Schmitt, M. zebecher, J., Sperl, S., and Magdolen, V. The plasminogen activation The German Chemo N0 Study Group. Randomized adjuvant therapy system as a novel target for therapeutic strategies. Fibrinolysis, 14: trial in high-risk node-negative breast cancer patients identified by 114–132, 2000. urokinase-type plasminogen activator and plasminogen activator inhib- 4. Duffy, M. J., O’Grady, P., Devaney, D., O’Sioran, L., Fennelly, J. J., itor-1. JNCI, 93: 913–920, 2001. and Lijnen, H. Urokinase-plasminogen activator, a marker for aggres- 19. Tandon, A. K., Clark, G. M., Chamness, G. C., Chirgwin, J. M., and sive breast carcinomas. Cancer (Phila.), 62: 531–533, 1988. McGuire, W. L. Cathepsin D and prognosis in breast cancer. N. Engl. 5. Spyratos, F., Brouillet, J. P., Defrenne, A., Hacene, K., Rouess, J., J. Med., 322: 297–302, 1990. Maudelonde, T., Brunet, M., Andrieu, C., Desplaces, A., and Rochefort, 20. Ferno¨, M., Baldetorp, B., Borg, Å., Brouillet, J. P., Olsson, H., H. Cathepsin D: an independent prognostic factor for metastasis of Rochefort, H., Sellberg, G., Sigurdsson, H., and Killander, D. The South breast cancer. Lancet, 2: 1115–1118, 1989. Sweden Breast Cancer Group. Cathepsin D, both a prognostic factor and 6. Thorpe, S. M., Rochefort, H., Garcia, M., Freiss, G., Christensen, a predictive factor for the effect of adjuvant tamoxifen in breast cancer. Eur. J. Cancer, 30: 2042–2048, 1994. I. J., Khalaf, S., Paolucci, F., Pau, B., Rasmussen, B. B., and Rose, C. 21. Ferrandina, G., Scambia, G., Bardelli, F., Benedetti Panici, P., Association between high concentrations of Mr 52,000 cathepsin D and poor prognosis in primary human breast cancer. Cancer Res., 49: Manusco, S., and Messori, A. Relationship between cathepsin D content 6008–6014, 1989. and disease-free survival in node-negative breast cancer patients: a meta-analysis. Br. J. Cancer, 76: 661–666, 1997. 7. Ja¨nicke, F., Schmitt, M., and Graeff, H. Clinical relevance of the urokinase-type and tissue-type plasminogen activators and of their type 1 22. Foekens, J., Look, M. P., Bolt-de-Vries, J., Meijer-van Gelder, inhibitor in breast cancer. Semin. Thromb. Hemost., 17: 303–312, 1991. M. E., van Putten, W. L. J., and Klijn, J. G. M. Cathepsin D in primary breast cancer: prognostic evaluation involving 2810 patients. Br. J. 8. Ja¨nicke, F., Schmitt, M., Pache, L., Ulm, K., Harbeck, N., Ho¨fler, H., Cancer, 79: 300–307, 1999. and Graeff, H. Urokinase (uPA) and its inhibitor PAI-1 are strong, 23. Thomssen, C., Schmitt, M., Goretzki, L., Oppelt, P., Pache, L., independent prognostic factors in node-negative breast cancer. Breast Dettmar, P., Ja¨nicke, F., and Graeff, H. Prognostic value of the cysteine Cancer Res. Treat., 24: 195–208, 1993. proteases cathepsin B and L in human breast cancer. Clin. Cancer Res., 9. Foekens, J. A., Schmitt, M., van Putten, W. L. J., Peters, H. A., 1: 741–746, 1995. Kramer, M. D., Ja¨nicke, F., and Klijn, J. G. M. Plasminogen activator 24. Foekens, J., Kos, J., Peters, H. A., Krasovec, M., Look, M. P., inhibitor-1 and prognosis in primary breast cancer. J. Clin. Oncol., 12: Cimerman, N., Meijer-van-Gelder, M., Henzen-Xogmans, S. C., Putten, 1648–1658, 1994. W. L. J., and Klijn, J. G. M. Prognostic significance of cathepsins B and 10. Grøhndahl-Hansen, J., Peters, H. A., van Putten, W. L. J., Look, L in primary human breast cancer. J. Clin. Oncol., 16: 1013–1021, 1998. M. P., Pappot, H., Rønne, E., Danø, K., Klijn, J. G. M., Bru¨nner, N., and 25. Schmitt, M., Ja¨nicke, F., and Graeff, H. Tumor-associated pro- Foekens, J. A. Prognostic significance of the receptor for urokinase teases. Fibrinolysis, 6: 3–26, 1992. plasminogen activator in breast cancer. Clin. Cancer Res., 1: 1079– 1087, 1995. 26. Thomssen, C., Oppelt, P., Ja¨nicke, F., Ulm, K., Harbeck, N., Ho¨fler, H., Kuhn, W., Graeff, H., and Schmitt, M. Identification of low-risk 11. Ferno¨, M., Bendahl, P. O., Borg, Å., Brudell, J., Hirschberg, L., node-negative breast cancer patients by tumor-biological factors PAI-1 Olsson, H., and Killander, D. Urokinase plasminogen activator, a strong and cathepsin L. Anticancer Res., 18: 2173–2180, 1998. independent prognostic factor in breast cancer, analysed in steroid receptor cytosols with a luminometric immunoassay. Eur. J. Cancer, 32: 27. Herszenyi, L., Plebani, M., Carraro, P., De Paoli, M., Roveroni, G., 793–801, 1996. Cardin Tulassay, Z., Naccarato, R., and Farinati, F. The role of cysteine and serine proteases in colorectal carcinoma. Cancer (Phila.), 86: 1135– 12. Eppenberger, U., Kueng, W., Schlaeppi, J. M., Roesel, J. L., Benz, 1142, 1999. C., Mueller, H., Matter, A., Zuber, M., Luescher, K., Litschgi, M., Schmitt, M., Foekens, J., and Eppenberger-Castori, S. Markers of tumor 28. Russo, A., Bazan, V., Migliavacca, M., Zanna, I., Tubiolo, C., angiogenesis and proteolysis independently define high- and low-risk Tumminello, F. M., Dardanoni, G., Cajozzo, M., Bazan, P., Modica, G., subsets of node-negative breast cancer patients. J. Clin. Oncol., 16: Latteria, M., Tomasino, R. M., Colucci, G., Gebbia, N., and Leto, G. 3129–3136, 1998. Prognostic significance of DNA ploidy, S-phase fraction, and tissue levels of aspartic, cysteine, and serine proteases in operable gastric 13. Kim, S. J., Shiba, E., Kobayashi, T., Yayoi, E., Furukawa, J., Takat- cancer. Clin. Cancer Res., 6: 178–184, 2000. suka, Y., Shin, E., Hiroki, K., Inaji, H., and Takai, S. Prognostic impact of urokinase-type plasminogen activator (PA), PA inhibitor type-1, and tissue- 29. Schmitt, M., Thomssen, C., Ulm, K., Seiderer, A., Harbeck, N., type PA antigen levels in node-negative breast cancer: a prospective study Ho¨fler, H., Ja¨nicke, F., and Graeff, H. Time-varying prognostic impact on multicenter basis. Clin. Cancer Res., 4: 177–182, 1998. of tumor biological factors urokinase (uPA), PAI-1, and steroid hormone receptor status in primary breast cancer. Br. J. Cancer, 76: 14. Knoop, A., Andreasen, P. A., Andersen, J. A., Hansen, S., Lænk- 306–311, 1997. holm, A. V., Simonsen, A. C. W., Andersen, J., Overgaard, J., and Rose, C. Prognostic significance of urokinase-type plasminogen activator and 30. Harbeck, N., Dettmar, P., Thomssen, C., Berger, U., Ulm, K., plasminogen activator inhibitor-1 in primary breast cancer. Br. J. Can- Kates, R., Ja¨nicke, F., Ho¨fler, H., Graeff, H., and Schmitt, M. Risk- cer, 77: 932–940, 1998. group discrimination in node-negative breast cancer using invasion and proliferation markers: six-year median follow-up. Br. J. Cancer, 80: 15. Kute, T. E., Grøhndahl-Hansen, J., Shao, S. M., Long, R., Russell, 419–426, 1999. G., and Bru¨nner, N. Low cathepsin D and low plasminogen activator type 1 inhibitor in tumor cytosols defines a group of node negative 31. Kaplan, E. L., and Meier, P. Nonparametric estimation from incom- breast cancer patients with low risk of recurrence. Breast Cancer Res. plete observations. J. Am. Stat. Assoc., 53: 457–481, 1958. Treat., 47: 9–16, 1998. 32. Cox, D. R. Regression models and life-tables. J. R. Stat. Soc. (B), 16. Bouchet, C., Hace`ne, K., Martin, P. M., Becette, V., Tubiana-Hulin, 34: 187–200, 1972. M., Lasry, S., Oglobine, J., and Spyratos, F. Dissemination risk index 33. Breiman, I., Friedman, J., Olshen, R., and Stone, C. (eds.). Classi- based on plasminogen activator system components in primary breast fication and Regression Trees. New York: Chapman and Hall, Inc., cancer. J. Clin. Oncol., 17: 3048–3057, 1999. 1984. 17. Harbeck, N., Thomssen, C., Berger, U., Ulm, K., Kates, R., Ho¨fler, 34. Grambsch, P., and Therneau, T. Proportional hazard tests and H., Ja¨nicke, F., Graeff, H., and Schmitt, M. Invasion marker PAI-1 diagnostics based on weighted residuals. Biometrika, 81: 515–526, remains a strong prognostic factor after long-term follow-up both for 1994. primary breast cancer and following first relapse. Breast Cancer Res. 35. Hastie, T., and Tibshirani, R. Varying coefficient models. J. Am. Treat., 54: 147–157, 1999. Stat. Assoc., 90: 78–94, 1993.

36. Sweep, C. G. J., Geurts-Moespot, J., Grebenschikov, N., de Witte, 40. EBCTCG. Tamoxifen for early breast cancer: an overview of the J. H., Heuvel, J. J., Schmitt, M., Duffy, M. J., Ja¨nicke, F., Kramer, randomised trials. Lancet, 351: 1451–1467, 1998. M. D., Foekens, J. A., Bru¨nner, N., Brugal, G., Pedersen, A. N., and 41. EBCTCG. Polychemotherapy for early breast cancer: an overview Benraad, T. J. External quality assessment of trans-European multi- of the randomised trials. Lancet, 352: 930–942, 1998. center antigen determinations (enzyme-linked immunosorbent assay) of 42. Pierga, J. Y., Laine-Bidron, C., Beuzeboc, P., DeCremoux, P., urokinase-type plasminogen activator (uPA) and its type-1 inhibitor Pouillart, P., and Magdelenat, H. Plasminogen activator inhibitor-1 (PAI-1) in human breast cancer tissue extracts. Br. J. Cancer, 78: (PAI-1) is not related to response to neoadjuvant chemotherapy in breast 1434–1441, 1998. cancer. Br. J. Cancer, 76: 537–540, 1997. 37. Benraad, T. J., Geurts-Moespot, A., Sala, M., Piffanelli, A., Ross, 43. Ja¨nicke, F., Thomssen, C., Pache, L., Schmitt, M., and Graeff, H. A., and Foekens, J. A. Quality control of cathepsin-D measurement by Urokinase (uPA) and PAI-1 as selection criteria for adjuvant chemo- the EORTC receptor study group. Eur. J. Cancer, 28: 72–75, 1992. therapy in axillary node-negative breast cancer patients. In: M. Schmitt, 38. Lah, T. T., Cercek, M., Blejec, A., Kos, J., Gorodetsky, E., Somers, H. Graeff, and F. Ja¨nicke (eds.), Prospects in Diagnosis and Treatment of R., and Daskal, I. Cathepsin B, a prognostic indicator in lymph node- Cancer, pp. 207–218. Amsterdam: Elsevier Science Publishers, B. V., 1994. negative breast carcinoma patients: comparison with cathepsin D, ca- 44. Foekens, J. A., Look, M. P., Peters, H. A., van Putten, W. L. J., thepsin L, and other clinical indicators. Clin. Cancer Res., 6: 578–584, Portengen, H., and Klijn, J. G. M. Urokinase-type plasminogen activator 2000. and its inhibitor PAI-1: predictors of poor response to tamoxifen therapy in 39. Foekens, J. A., Peters, H. A., Look, M. P., Portengen, H., Schmitt, recurrent breast cancer. J. Natl. Cancer Inst. (Bethesda), 87: 751–756, 1995. M., Kramer, M. D., Bru¨nner, N., Ja¨nicke, F., Meijer-van Gelder, M. E., 45. Billgren, A. M., Rutqvist, L. E., Johansson, H., Hagerstrom, T., and Henzen-Logmans, S. C., van Putten, W. L. J., and Klijn, J. G. M. The Skoog, L. The role of cathepsin D and PAI-1 in primary invasive breast urokinase system of plasminogen activation and prognosis in 2780 cancer as prognosticators and predictors of treatment benefit with ad- breast cancer patients. Cancer Res., 60: 636–643, 2000. juvant tamoxifen. Eur. J. Cancer, 36: 1374–1380, 2000.

Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2001 American Association for Cancer Research. Prognostic Impact of Proteolytic Factors (Urokinase-Type Plasminogen Activator, Plasminogen Activator Inhibitor 1, and Cathepsins B, D, and L) in Primary Breast Cancer Reflects Effects of Adjuvant Systemic Therapy

Nadia Harbeck, Uta Alt, Ursula Berger, et al.

Clin Cancer Res 2001;7:2757-2764.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/7/9/2757

Cited articles This article cites 39 articles, 11 of which you can access for free at: http://clincancerres.aacrjournals.org/content/7/9/2757.full#ref-list-1

Citing articles This article has been cited by 10 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/7/9/2757.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/7/9/2757. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.