| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Clinical Investigations |
Department of Obstetrics and Gynecology [N. H., R. E. K., M. K., M. S.] and Department of Experimental Oncology [A. K.], Technical University, D-81675 Munich, Germany; Department of Medical Oncology, Rotterdam Cancer Institute (Daniel den Hoed Kliniek) and University Hospital Rotterdam, 3075 Rotterdam, The Netherlands [M. P. L., M. E. M-v. G., J. G. M. K., J. A. F.]; and Department of Obstetrics and Gynecology, University of Hamburg, D-20246 Hamburg, Germany [F. J.]
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
There is already abundant experimental evidence that the plasminogen activator system plays a key role in tumor invasion and metastasis (1 , 2) . A critical balance of uPA, its cell surface receptor uPA-R (CD 87), and its inhibitor PAI-1 is the prerequisite for efficient focal proteolysis, adhesion, and migration, and hence, subsequent tumor cell invasion and metastasis.
In primary breast cancer, uPA and PAI-1 are the first novel tumor biological markers for which level-I evidence for their prognostic impact according to the recently proposed criteria of Hayes et al. (3) has been provided (4 , 5) . A strong prognostic impact of uPA and PAI-1 as determined by biochemical assays in primary tumor tissue has been reported by numerous investigators under a variety of demographic conditions (reviewed in Refs. 2 , 6 , and 7 ). Patients with high tumor antigen levels of either factor have a significantly worse survival than patients with low levels.
The particular combination of the factors uPA/PAI-1 (both factors low versus either or both high) is superior to either of these factors taken alone as well as to established prognostic factors tumor size, grade, hormone receptor or menopausal status with regard to selection of low-risk patients (5
, 8)
. It also retains its significance in multivariate analysis including novel tumor biological factors such as HER2 protein overexpression, cathepsin D, p53, S phase, MIB1, or DNA ploidy (9)
. Risk assessment on the basis of uPA/PAI-1 provides clinically important information, not only for the low-risk group but also for the high-risk group, particularly in node-negative breast cancer. Node-negative patients with low uPA/PAI-1 have an excellent prognosis with a 5-year DFS exceeding 90%, even without adjuvant systemic therapy; they are candidates for being spared the burden of adjuvant CT (4
, 9) . The low-risk group identified by uPA/PAI-1, comprising about 50% of node-negative patients, is substantially larger than that characterized by the St. Gallen criteria (10)
, and much closer to the 
70% of node-negative patients cured by locoregional treatment alone. In contrast, node-negative patients with high uPA/PAI-1 have a rather high risk for relapse, comparable with that associated with more than three involved lymph nodes; for these patients, adjuvant systemic treatment would be strongly recommended (8)
.
To optimize adjuvant therapy for high-risk patients as classified by uPA/PAI-1, there is an urgent clinical need to know whether these patients do indeed benefit from adjuvant systemic therapy, and if so, which therapy would be most beneficial. Unfortunately, information on the predictive value of uPA and/or PAI-1 with regard to therapy response is still scarce. Results from local response to neoadjuvant CT (11) or response to palliative systemic therapy in advanced or metastatic breast cancer (12 , 13) cannot be readily transferred to the adjuvant setting. Recent reports show that the prognostic impact of uPA/PAI-1 is lost in patients who received adjuvant chemo- or endocrine therapy, thus suggesting a benefit from adjuvant therapy in high uPA/PAI-1 patients (8 , 14) . Consistent with this, the first interim analysis of a prospective randomized multicenter therapy trial ("Chemo N0"), in which patients were stratified according to their uPA and PAI-1 levels, indicated that high-risk patients benefit from adjuvant CMF-CT (4) .
The present study addresses for the first time the issue of a predictive impact of uPA/PAI-1 in breast cancer (n = 3424) with regard to adjuvant chemo- and/or endocrine therapy.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3
. Treatment decisions with regard to primary surgery and adjuvant systemic therapy were based primarily on consensus recommendations at the time. Treatment strategies differed between Munich and Rotterdam, particularly regarding the administration of adjuvant systemic therapy to node-positive patients (see Table 2
; Refs. 8
, 15
). Of those patients treated by adjuvant CT, the majority received CMF (Rotterdam, 76%, Munich, 65%); about one-fifth in both collectives received anthracycline-containing regimens (Rotterdam, 24%, Munich, 21%); and the rest of the Munich patients were treated by other CT regimens. The endocrine treatment used in both collectives was tamoxifen.
|
|
|
Laboratory Assays
In the Munich cohort, uPA and PAI-1 antigen were prospectively measured by ELISA (uPA: Imubind 894; PAI-1: Imubind 821; both from American Diagnostica Inc., Greenwich, CT) in detergent extracts of the primary tumor tissue as reported in Jänicke et al. (16)
. In the Rotterdam cohort, uPA and PAI-1 antigen were measured by ELISA, using the same antibodies as above, in cytosol preparations of the primary tumor as described by Foekens et al. (15)
.
Statistical Methods
Variable Definition and Recoding for the Combined Analysis.
Because of differences in measurement techniques between the data sets, data recoding was required. For the laboratory measurements of uPA, PAI-1, ER, and PgR, fractional ranks were computed with respect to each distribution. Fractional ranks also kept the variables on a convenient scale from zero to one, thus facilitating comparison of the ß coefficients of different factors. In particular, ranked ER and PgR measurements were able to be included as continuous variables in the statistical models, even though biochemical and immunohistochemical assays were used. The weak correlations found between these laboratory measurements and the remaining "classical" staging factors support the inference that similar ranks imply similar biological characteristics even across data sets.
A binary variable for uPA/PAI-1 was defined as 0 for uPA and PAI-1 both below their respective cutoffs and as 1 otherwise (i.e., either or both above the respective cutoff; Ref. 17
). Previously determined and validated univariate cutoff values by the Munich group (4
, 17)
were applied to the Rotterdam data by transforming the Munich cutoffs to fractional ranks and applying these to the Rotterdam data, resulting in almost exactly the same percentage of uPA/PAI-1 "high" versus "low" in each cohort (see Table 1
).
The pT stage (18) was coded using two auxiliary binary variables: (a) pT1 (coded 0) versus all others (coded 1); and (b) pT1 and pT2 (coded 0) versus pT3 and pT4 (coded 1). Fractional ranks were assigned separately within the two data sets for the number of affected lymph nodes (variable denoted "lymph nodes"). Equal numbers of nodes correspond to equal fractional ranks across data sets, to within a few percentage points. For patient age, fractional ranks were first computed for the Rotterdam data set, and the Munich ages were then transformed to this scale. To model the nonlinear dependence of HR on age as closely as possible, both the fractional rank itself as well as its square is included in the models. The three binary variables for adjuvant therapy (RT, HT, and CT) were coded such that the value 1 represented "known to have been treated by the respective kind of adjuvant therapy." A binary variable "data set" was introduced and used to stratify the analysis as discussed below. This variable accounts for systematic differences in demographic influences, unobserved factors contributing to the stage of the disease, or adjuvant systemic therapy strategies.
Survival Analysis.
The Cox proportional hazards model was used with continuous ranked variables and binary variables as described above. All of the tests were performed at a significance level of 
= 0.05 with a 95% CI. Variables were included according to likelihood ratios in a stepwise forward fashion using the SPSS software package (SPSS Inc., Chicago, IL). Unless otherwise stated, main (i.e., linear) effects were always included as a first block, whereas interactions were included as a second block in the analysis. This method implies that a main effect that is significant in the first block will be retained in the model, even if an interaction in the second block is so strong as to reduce the main effect coefficient below the level of significance. All of the models were stratified by data set. The SPSS software package was also used to compute fractional ranks, correlation coefficients, associations, and other statistical properties.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
DFS Including uPA/PAI-1 and Their Interactions with Therapy in All Patients.
The 5-year relapse rates associated with low and high uPA/PAI-1 were 28 and 46%, respectively. The probabilities of being treated by CT or HT in subgroups defined by uPA/PAI-1 are depicted in Table 3
.
In Table 4
, we report the results of a proportional hazards analysis for DFS in all of the patients, stratified by data set. The first stage included established prognostic factors (ER, PgR, age, lymph nodes, pT stage, all coded as described above under "Statistical Methods") as well as uPA/PAI-1, CT, HT, and RT. The second stage included the interactions CT and HT with uPA/PAI-1, and lymph nodes with uPA/PAI-1, as well as the "treatment interaction," i.e., CT with HT.
|
were significant except RT and ER. The HR of uPA/PAI-1 was 2.0 (CI, 1.8–2.3; P < 0.001). The overall effect of age, including squared fractional rank, in the model was a gradual, almost linear, drop of the HR from the youngest patients (defined to have HR = 1), leveling off to about HR = 0.5 by about age 60 and apparently rising slightly above 0.5 at about age 65. The interaction between lymph nodes and uPA/PAI-1 was not significant in the analysis of all of the patients, nor was CT x HT, implying no evidence against an additive effect of this treatment combination. The key result is the significant (negative) interaction between CT and the variable uPA/PAI-1. This interaction implies that the higher HR of relapse (2.01) associated with high uPA/PAI-1 (compared with low uPA/PAI-1) is significantly reduced (0.68 x 2.01 = 1.36) in patients who receive adjuvant CT. This benefit occurrs in addition to the independent overall risk reductions of about one-third attributable to CT (HR = 0.69), or HT (HR = 0.68). No significant interaction was found between HT and uPA/PAI-1 (95% CI for this HR, 0.66–1.28). Hence, the benefits of both therapies were significant, but only for CT was an additional (enhanced) benefit seen among high uPA/PAI-1 patients.
Fig. 1
illustrates the HRs of CT and HT taking into account significant interactions with uPA/PAI-1 according to Tables 4
5
6
(for discussion of Tables 5
and 6
see the sections that follow). For all of the patients, the significant interaction CT x uPA/PAI-1 is seen in the upper panel of Fig. 1
as a hazard reduction, attributable to CT, that is strongly affected by uPA/PAI-1. The lack of a significant interaction HT x uPA/PAI-1 manifests itself in the figure in that the hazard reduction attributable to HT is not affected by uPA/PAI-1.
|
|
|
shows the proportional hazards analysis for DFS, which was stratified by data set as in Table 4
and which includes the same factors. In this subgroup, ranked ER is significant, higher values being associated with higher HR, whereas ranked PgR is associated with lower HR. For tumor stage, only the distinction pT1 versus all others is significant, but not pT1/pT2 versus the rest. The HRs for age imply that young age is an even more strongly unfavorable factor in this subgroup than in patients on the whole.
In patients with 0–3 affected lymph nodes, the effects involving therapy and uPA/PAI-1 are qualitatively and even quantitatively very close to those seen in the analysis of all patients: The hazard associated with high uPA/PAI-1 is slightly greater; adjuvant endocrine therapy has about the same benefit as in all patients; and the interaction between adjuvant CT and uPA/PAI-1 is similar (see center panel of Fig. 1
).
DFS Including uPA/PAI-1 and Their Interactions in Patients with Four or More Involved Nodes.
In patients with four or more involved axillary lymph nodes, the adjuvant therapy percentages were as follows: with low uPA/PAI-1 (n = 398; 5-year relapse rate, 56%), 27% were treated by adjuvant endocrine therapy, and 36% by CT. With high uPA/PAI-1 (n = 388; 5-year relapse rate, 72%), these percentages are slightly lower at 26% and 29%, respectively. The results of a Cox analysis performed for this subgroup are reported in Table 6
. The factors were included as in the previous models; the analysis was again stratified by data set.
In these patients, it is noteworthy that uPA/PAI-1 had an enormous impact (ß = log HR = 3.02), but there was also a large negative interaction of uPA/PAI-1 with lymph nodes (ß = log HR = -2.79). There was also an (apparently) very high hazard (ß = log HR = 5.36) associated with lymph nodes within this subgroup of patients with 4 or more affected nodes, but this number was partly an artifact of the representation in fractional ranks. To facilitate interpretation of this HR, as well as the interaction of lymph nodes with uPA/PAI-1, we compared the hazard for 10 versus 4 affected nodes: For patients with low uPA/PAI-1, the HR of patients with 10 affected nodes was about twice as high as for 4 affected nodes, as seen in Table 6
. (If this mere doubling of risk going from 4 to 10 nodes seems too moderate in view of ß = 5.36, it must be kept in mind that the fractional rank for lymph nodes for a patient with 4 nodes is already quite high, about 0.78.) In contrast, for patients with high uPA/PAI-1, the interaction means that the HR of patients with 10 affected nodes was only about 1.5 times that of patients with 4 affected nodes. Summarizing, the results (including interaction of lymph nodes with uPA/PAI-1) imply that the number of affected nodes even above 4 is important, but more so for low uPA/PAI-1 than for high uPA/PAI-1. In patients with >4 affected lymph nodes, the benefits of CT and HT and their relation to uPA/PAI-1 were again qualitatively and even quantitatively very close to those seen in the analysis of all patients and in the group with 0–3 affected nodes (see bottom panel of Fig. 1
).
Benefits of CT and HT for DFS in Subgroups according to uPA/PAI-1.
The effect of uPA/PAI-1 on response to therapy is also seen by constructing separate Cox models for high and low uPA/PAI-1 (again stratified by data set). In all of the patients, the HR was 0.68 for CT and 0.74 for HT, according to a multivariate model for the low-uPA/PAI-1 subgroup. Within the multivariate model for the high-uPA/PAI-1 subgroup, the corresponding HRs were 0.49 for CT and 0.63 for HT. (In terms of log HR, the difference in HR for CT between low and high uPA/PAI-1 was about three SEs, whereas for HT the corresponding difference was only one-third of a SE.) Hence, these models were consistent with the tendency for more relative benefit caused by CT in patients with high uPA/PAI-1 than in patients with low uPA/PAI-1, after controlling for other factors.
Cox models were also performed separately for high and low uPA/PAI-1 patients in the subgroup of patients with 0–3 affected lymph nodes. In the low-uPA/PAI-1 group (n = 1418; 5-year relapse rate 20%; 9% receiving HT, 17% receiving CT), it turned out that neither of the adjuvant therapy forms were significant: the 95% CI for the HR of CT was 0.60–1.22, and for HT it was 0.59–1.44. Because of statistical uncertainty, in this subgroup, a low-to-moderate benefit of either therapy is not ruled out. In contrast, in the high-uPA/PAI-1 subgroup (n = 1174; 5-year relapse rate 38%; 10% receiving HT, 19% receiving CT), both adjuvant therapy forms are significant and strong: with a HR of 0.51 (0.33–0.78), HT approximately halves the hazard; the benefit of CT appears to be even stronger with a HR of 0.43 (95% CI, 0.31–0.59). Comparing the result in these subgroups with the interaction analysis for 0–3 nodes reported above, the detection of a significant interaction CT x uPA/PAI-1 manifests itself in the uPA/PAI-1 subgroups as distinctly different HRs with nonintersecting CIs. In the case of HT, the 95% CI for the HRs in the two risk groups overlap substantially, and this is consistent with the lack of a significant interaction.
In patients with four or more affected nodes, a separate Cox regression (two-stage model) for the low-uPA/PAI-1 subgroup showed that these patients benefited significantly from adjuvant HT with a HR of 0.62 (95% CI, 0.43–0.90) and apparently also from CT with a HR of 0.70 (95% CI, 0.49–0.98). Lymph nodes were a very strong factor in this group (log HR = 5.68). The benefit from CT derived from the regression model for the group with four or more nodes and high uPA/PAI-1 corresponded to HR 0.60 (95% CI, 0.44–0.81). For HT, the HR was 0.68 (95% CI, 0.49–0.95). Lymph nodes were weaker in this model (log HR = 2.43). These results taken together are consistent with the full model with interactions for patients with 4 or more nodes reported above.
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Ideally, the gold standard for determining predictive information is a properly designed prospective study. However, for ethical reasons, new studies cannot include control groups of patients without adjuvant systemic therapy. On the other hand, large retrospective data sets containing substantial patient numbers with and without adjuvant systemic therapy are available for analysis. In general, it is quite difficult to ascertain predictive information from retrospective data in breast cancer, because adjuvant treatment decisions were made on the basis of guidelines in force at that time taking into account prognostic factors. Such factors thus act as confounding variables for retrospectively analyzing efficacy of adjuvant treatment. Moreover, different adjuvant treatment policies were used in different centers.
Nonetheless, for factors that do not strongly correlate with treatment decisions, the problem of confounding can be reduced by various methods, in particular, by appropriate use of multivariate analysis and stratification. Because (in contrast to ER and PgR) these requirements are satisfied rather well by uPA and PAI-1, the results presented in this paper should indeed reflect the predictive properties of uPA/PAI-1. An important step in "de-convoluting" the confounding factors in retrospective data is to introduce a multivariate statistical scoring model using as much of the information as possible in the other variables. A good scoring model will reduce the unexplained variation in the data and improve the chances of seeing interactions if they are present. Consequently, in this paper, a strategy of avoiding the use of cutoffs whenever possible, i.e., by representing most of the measurements as continuous variables, was applied. The only exception to the strategy of continuous variables were the factors uPA and PAI-1 themselves, for which previously optimized cutoffs, validated in a prospective multicenter trial (4 , 17) , were applied.
The present paper confirms that uPA/PAI-1 have a significant impact on patient outcome but also provides additional evidence supporting their use in the clinic by demonstrating how effects of adjuvant systemic therapy differ in patients classified according to uPA/PAI-1. As illustrated in Fig. 1
, primary breast cancer patients with low uPA/PAI-1 generally benefit from adjuvant endocrine and CT. However, the benefits of CT (but not endocrine therapy) are strongly enhanced in patients with high uPA/PAI-1. This finding is in accordance with the benefit from CMF observed in high-risk patients in the Chemo N0 trial (4)
. It is important to note that patients with high uPA/PAI-1 also benefit from adjuvant endocrine therapy, even though adjuvant CT has a greater beneficial impact on their DFS.
Node-negative patients with low uPA/PAI-1 have a very low risk of relapse per se. The benefit of endocrine therapy found in low uPA/PAI-1 patients even with zero to three affected nodes supports the clinical conclusions drawn from the Chemo N0 trial (4) as well as from retrospective data (8) : Node-negative patients with low uPA/PAI-1 (but not those with high uPA/PAI-1) may be candidates for being spared the burden of adjuvant CT but still could benefit from endocrine therapy if indicated, taking into account the known side effects of CT and the preventive benefits of endocrine therapy.
Retrospective analyses in advanced and metastatic breast cancer have shown decreased response to palliative endocrine therapy in patients with high uPA or PAI-1 levels in primary tumor tissue compared with patients with low levels (12 , 13) . This should not be regarded as a contradiction to our results obtained in the adjuvant setting, but can be understood taking the underlying tumor biology into account. High levels of uPA and PAI-1 do reflect an aggressive phenotype that may be overcome or suppressed by early systemic therapy as in the adjuvant setting but may be far too advanced for response to palliative therapy at a later stage.
Our results also provide insight into the interaction between uPA/PAI-1 and lymph nodes (5) . In patients with 4 or more nodes, we saw basically an "either-or" relationship relating the hazard to increases in number of affected lymph nodes and to uPA/PAI-1; this relationship suggests a picture in which multiple disease processes putting these patients in grave danger of relapse are active in parallel (e.g., lymphatic and hematogenous tumor cell dissemination). [An analogous statistical relationship also exists in the interaction between uPA and PAI-1 as separate factors (8) .] Our results thus support the following interpretation of the interaction between nodal status and uPA/PAI-1 found in Look et al. (5) . In that analysis, both uPA and PAI-1 had somewhat weaker impact in the node-positive subgroup when added to a base model in which nodal category is already included. This is just the trend that would be expected from the either-or relationship between uPA/PAI-1 and the number of nodes found in the present paper. The interaction between lymph nodes and uPA/PAI-1, thus, appears to be relevant to phenomena that are entirely different from the predictive role of uPA/PAI-1 for the response to adjuvant therapy found here. In our analysis, the "lymph nodes-uPA/PAI-1" interaction is relevant only for the group with 4 or more nodes, whereas the predictive interaction "CT-uPA/PAI-1" is also present in the zero-to-three-node subgroup.
The intended clinical application of the present work is to exploit fully the risk assessment information provided by uPA/PAI-1 in the context of clinical decision-making in primary breast cancer. The rationale is not limited to that of finding a very low-risk group who could be spared systemic treatment altogether, but rather to understand on the basis of the currently available evidence, including uPA and PAI-1 measurements, which treatment options are benefiting which patients. The "take-home message," although based on rather complex models and computations, is actually quite easy to interpret. Breast cancer patients with high uPA/PAI-1 have a more aggressive disease stage than conventional factors would otherwise lead the physician to believe. However, our results provide evidence that the DFS disadvantage attributable to this more aggressive disease phenotype can be largely counteracted by adjuvant endocrine therapy and in particular by adjuvant CT as illustrated in Fig. 1
. We certainly do not claim that uPA/PAI-1 are the only variables that should be taken into account for therapy decisions. Nonetheless, our results suggest that a significant and substantial improvement in decision support will be achievable by testing breast cancer patients for uPA and PAI-1. We do not yet have a basis for claiming that high levels of uPA and PAI-1 are predictive for targeted therapy benefit in the sense that HER2 overexpression is predictive of response to Herceptin therapy (19)
. Yet, it is quite likely that the potential improvement in decision support that is already present for conventional therapy approaches will be enhanced when novel, promising therapeutics targeting the plasminogen activation system (20, 21, 22)
become available for clinical application. Hence, considering the underlying tumor biology and our finding that high-risk patients according to uPA/PAI-1 benefit from conventional adjuvant systemic therapy, particularly from CT, we think that it is all the more promising to combine novel therapeutics targeting the plasminogen activation system with such conventional systemic therapy.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported in part by the State of Bavaria (KKF Project 8756159), the Deutsche Forschungsgemeinschaft (DFG SFB 469, B13), the Wilhelm-Sander-Stiftung (Grant 1996.066.2), and the Dutch Cancer Society (Grant DDHK 2000-2256), Amsterdam, the Netherlands. 
2 To whom requests for reprints should be addressed, at Frauenklinik, Klinikum rechts der Isar, Technische Universitaet Muenchen, Ismaninger Strasse 22, D-81675 Munich, Germany. Phone: 49-89-4140-5419; Fax: 49-89-4140-4846; E-mail: nadia.harbeck{at}lrz.tum.de
3 The abbreviations used are: uPA, urokinase-type plasminogen activator; CI, confidence interval; CT, chemotherapy; DFS, disease-free survival; CMF, cyclophosphamide, methotrexate,5-fluorouracil; ER, estrogen receptor; HR, hazard ratio; HT, hormone therapy; PAI-1, plasminogen activator inhibitor type 1; PgR, progesterone receptor; RT, radiotherapy.
Received 3/19/02. Accepted 6/11/02.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Harris, H. Fritsche, R. Mennel, L. Norton, P. Ravdin, S. Taube, M. R. Somerfield, D. F. Hayes, and R. C. Bast Jr American Society of Clinical Oncology 2007 Update of Recommendations for the Use of Tumor Markers in Breast Cancer J. Clin. Oncol., November 20, 2007; 25(33): 5287 - 5312. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Dorn, M. Schmitt, R. Kates, B. Schmalfeldt, M. Kiechle, A. Scorilas, E. P. Diamandis, and N. Harbeck Primary Tumor Levels of Human Tissue Kallikreins Affect Surgical Success and Survival in Ovarian Cancer Patients Clin. Cancer Res., March 15, 2007; 13(6): 1742 - 1748. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A.-S. Schrohl, M. E. Meijer-van Gelder, M. N. Holten-Andersen, I. J. Christensen, M. P. Look, H. T. Mouridsen, N. Brunner, and J. A. Foekens Primary Tumor Levels of Tissue Inhibitor of Metalloproteinases-1 Are Predictive of Resistance to Chemotherapy in Patients with Metastatic Breast Cancer Clin. Cancer Res., December 1, 2006; 12(23): 7054 - 7058. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. D. Sternlicht, A. M. Dunning, D. H. Moore, P. D.P. Pharoah, D. G. Ginzinger, K. Chin, J. W. Gray, F. M. Waldman, B. A.J. Ponder, and Z. Werb Prognostic Value of PAI1 in Invasive Breast Cancer: Evidence That Tumor-Specific Factors Are More Important Than Genetic Variation in Regulating PAI1 Expression. Cancer Epidemiol. Biomarkers Prev., November 1, 2006; 15(11): 2107 - 2114. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Zhang, X.-O. Shu, Q. Cai, Z. Ruan, Y.-T. Gao, and W. Zheng Functional plasminogen activator inhibitor-1 gene variants and breast cancer survival. Clin. Cancer Res., October 15, 2006; 12(20): 6037 - 6042. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Urban, V. Vuaroqueaux, M. Labuhn, M. Delorenzi, P. Wirapati, E. Wight, H.-J. Senn, C. Benz, U. Eppenberger, and S. Eppenberger-Castori Increased Expression of Urokinase-Type Plasminogen Activator mRNA Determines Adverse Prognosis in ErbB2-Positive Primary Breast Cancer J. Clin. Oncol., September 10, 2006; 24(26): 4245 - 4253. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Alfano, I. Iaccarino, and M. P. Stoppelli Urokinase Signaling through Its Receptor Protects against Anoikis by Increasing BCL-xL Expression Levels J. Biol. Chem., June 30, 2006; 281(26): 17758 - 17767. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. S. Dua, C. M. Isacke, and G. P.H. Gui The Intraductal Approach to Breast Cancer Biomarker Discovery J. Clin. Oncol., March 1, 2006; 24(7): 1209 - 1216. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Duffy Predictive Markers in Breast and Other Cancers: A Review Clin. Chem., March 1, 2005; 51(3): 494 - 503. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. E. M.-v. Gelder, M. P. Look, H. A. Peters, M. Schmitt, N. Brunner, N. Harbeck, J. G. M. Klijn, and J. A. Foekens Urokinase-Type Plasminogen Activator System in Breast Cancer: Association with Tamoxifen Therapy in Recurrent Disease Cancer Res., July 1, 2004; 64(13): 4563 - 4568. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. N. Span, E. Waanders, P. Manders, J. J.T.M. Heuvel, J. A. Foekens, M. A. Watson, L. V.A.M. Beex, and F. C.G.J. Sweep Mammaglobin Is Associated With Low-Grade, Steroid Receptor-Positive Breast Tumors From Postmenopausal Patients, and Has Independent Prognostic Value for Relapse-Free Survival Time J. Clin. Oncol., February 15, 2004; 22(4): 691 - 698. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Manders, V. C. G. Tjan-Heijnen, P. N. Span, N. Grebenchtchikov, J. A. Foekens, L. V. A. M. Beex, and C. G. J. Sweep Predictive Impact of Urokinase-Type Plasminogen Activator: Plasminogen Activator Inhibitor Type-1 Complex on the Efficacy of Adjuvant Systemic Therapy in Primary Breast Cancer Cancer Res., January 15, 2004; 64(2): 659 - 664. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Manders, F. C. G. J. Sweep, V. C. G. Tjan-Heijnen, A. Geurts-Moespot, D. T. H. van Tienoven, J. A. Foekens, P. N. Span, J. Bussink, and L. V. A. M. Beex Vascular Endothelial Growth Factor Independently Predicts the Efficacy of Postoperative Radiotherapy in Node-Negative Breast Cancer Patients Clin. Cancer Res., December 15, 2003; 9(17): 6363 - 6370. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Qin, W. Zhu, C. Wagner-Mann, and E. R. Sauter Nipple Aspirate Fluid Expression of Urokinase-Type Plasminogen Activator, Plasminogen Activator Inhibitor-1, and Urokinase-Type Plasminogen Activator Receptor Predicts Breast Cancer Diagnosis and Advanced Disease Ann. Surg. Oncol., October 1, 2003; 10(8): 948 - 953. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. M. Matrisian, G. W. Sledge Jr., and S. Mohla Extracellular Proteolysis and Cancer: Meeting Summary and Future Directions Cancer Res., October 1, 2003; 63(19): 6105 - 6109. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Goldhirsch, W. C. Wood, R. D. Gelber, A. S. Coates, B. Thurlimann, and H.-J. Senn Meeting Highlights: Updated International Expert Consensus on the Primary Therapy of Early Breast Cancer J. Clin. Oncol., September 1, 2003; 21(17): 3357 - 3365. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A.-S. Schrohl, M. Holten-Andersen, F. Sweep, M. Schmitt, N. Harbeck, J. Foekens, and N. Brunner Tumor Markers: From Laboratory To Clinical Utility Mol. Cell. Proteomics, June 1, 2003; 2(6): 378 - 387. [Full Text] [PDF]
|
|
|
|
|
|
I. Zemzoum, R. E. Kates, J. S. Ross, P. Dettmar, M. Dutta, C. Henrichs, S. Yurdseven, H. Hofler, M. Kiechle, M. Schmitt, et al. Invasion Factors uPA/PAI-1 and HER2 Status Provide Independent and Complementary Information on Patient Outcome in Node-Negative Breast Cancer J. Clin. Oncol., March 15, 2003; 21(6): 1022 - 1028. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
