Purpose: To examine the in vitro regulation of hepatocyte growth factor activator inhibitor type 2 (HAI-2) in breast cancer cells and the in vivo predictive role for the efficacy of chemoendocrine primary therapy in patients with breast cancer.

Materials and Methods: HAI-2 regulation was studied in a panel of breast cancer cell lines comparing normoxia to hypoxia. The effect of HIF-1α RNAi on HAI-2 expression was evaluated in these cells. HAI-2 was examined in breast cancer using in situ hybridization and immunohistochemistry. The HAI-2 predictive role was assessed in T2-4 N0-1 breast cancers (n = 177) enrolled in a neoadjuvant randomized trial comparing epirubicin versus epirubicin + tamoxifen.

Results: HAI-2 mRNA and protein were regulated by hypoxia in the c-erbB2–positive cell lines, SKBR3 and BT474, and controlled by HIF-1α in these cells. Immunohistochemistry confirmed this profile with high expression of HAI-2 in c-erbB2–positive breast cancer. HAI-2 was correlated with T status (P < 0.004), node involvement (P = 0.01), and c-erbB2 expression (P = 0.05). HAI-2 also correlated with hypoxia markers such as carbonic anhydrase IX expression (P = 0.01) and HIF-1α. Additionally, high levels of HAI-2 were a significant predictor for poor clinical complete response to preoperative epirubicin in univariate (P = 0.01) and multivariate analyses (P = 0.016). No correlation with disease-free survival and survival was observed.

Conclusion: HAI-2 expression in breast cancer correlated with tumor aggressiveness in vivo. It is a HIF target in c-erbB2–positive cells and it is an independent negative predictive factor of efficacy of anthracycline therapy. The interaction of HAI-2 with the hepatocyte growth factor activation pathway may be a useful site for therapeutic intervention.

Hepatocyte growth factor (HGF) is a mitogen, morphogen, and survival factor for cells expressing Met, the HGF high-affinity tyrosine kinase receptor (1, 2). Both HGF and c-Met are expressed by neoplastic cells enabling autocrine and paracrine regulation of these processes (3). In human breast cancer, HGF plays a role in tumor metastasis and is an independent predictor of disease progression and survival (4).

HGF is secreted as an inactive pro-peptide which is cleaved by HGF activator into its active form (5). HGF activator is regulated by two inhibitors, HGF activator inhibitor type 1 (HAI-1) and type 2 (HAI-2; refs. 6, 7), both transmembrane proteins (8) that are secreted as proteolytically truncated forms (7). Both HGF (9) and c-Met are induced by hypoxia and have a critical role in cell migration under low oxygen tension (9).

Although HAIs are likely to be important in HGF activation in the invasive growth of tumors (10), little is known of their role in human tumors. Kang et al. showed HAI-1 to be an independent predictor of poor outcome in node-negative breast carcinomas (11) and when coexpressed with matriptase, patients have an elevated relative risk of recurrence compared with expression of HAI-1 alone. Although some data are available on HAI-1 (12, 13), there is only one report of HAI-2 in human normal and neoplastic tissues, showing that high expression of HAI-2 protein in breast cancer was inversely related to nodal involvement and tumor size (13).

Although investigating the role of hypoxic regulation of HGF via hypoxia-inducible factor 1α, we hypothesized that there may be reciprocal regulation of its inhibitor, HAI-2. However, surprisingly, we observed that HAI-2 was also up-regulated by hypoxia, being specifically induced in c-erbB2–positive breast cancer cell lines. To further study the clinical implications of this, we evaluated HAI-2 expression in a series of breast cancer patients enrolled in a prospective randomized trial comparing single agent epirubicin versus epirubicin plus tamoxifen. The aims of the study were to (a) investigate the regulation of HAI-2 inhibitor by hypoxia, (b) correlate HAI-2 expression with clinical and biological variables, (c) to evaluate the predictive role of this marker for treatment activity and efficacy, and (d) to assess the effect of treatment on HAI-2 expression.

Patients. Between January 1997 and January 2002, 211 patients bearing T2-4 N0-1 breast cancer, consecutively observed at the Breast Unit of Cremona, were enrolled in a randomized trial comparing single agent epirubicin (EPI arm) versus epirubicin plus tamoxifen (EPI-Tam arm) as primary systemic treatment. One hundred and five patients were randomized to receive epirubicin alone and 106 patients were randomized to receive epirubicin plus tamoxifen. On first presentation, an incision biopsy was done on each patient and a small tissue sample (0.5-0.8 cm) was removed. Chemotherapy was started within days of diagnosis. Patients in the EPI arm received 60 mg/m2 of epirubicin (Farmorubicina, Pharmacia, Milan, Italy) by slow i.v. push on days 1 and 2, whereas patients on the EPI-TAM arm received 60 mg/m2 of epirubicin by slow i.v. push on days 1 and 2, and 30 mg of tamoxifen (Kessar, Pharmacia, Milan, Italy) daily. Epirubicin infusions in both arms were repeated every 21 days for three or four cycles before definitive surgery, whereas tamoxifen was given continuously until the end of the cytotoxic treatment. All patients postoperatively received four cycles of the CMF regimen, which was given on days 1 and 8, every 28 days. The dose of cyclophosphamide and 5-fluorouracil was 600 mg/m2 body surface area, and the dose of methotrexate was 40 mg/m2 (14). Patients with estrogen receptor–positive primary tumor in both arms received tamoxifen (20 mg) up to progression or for a maximum of 5 years after CMF.

Treatment evaluation. Each month, the size of the primary tumor and the size of the axillary lymph nodes, when appreciable, were measured with a caliper by the same clinician. Clinical response was assessed according to the WHO criteria by the measurement of the changes in the product of the two largest diameters recorded in two successive evaluations. Pathologically complete response was defined as the absence of neoplastic cells either in the breast or in the axillary lymph nodes. Surgery was planned after full clinical reassessment. Quadrantectomy or modified radical mastectomy was done when indicated in association with full axillary node dissection. All patients subjected to quadrantectomy underwent irradiation of the residual breast (60 Gy delivered over 6 weeks).

Histopathologic grade and immunohistochemistry. All assessments in breast cancer specimens were carried out blinded to patient outcome and whether the samples they examined were obtained from incisional biopsy or definitive surgery. Viable tumor tissues obtained from the incisional biopsy and from the definitive surgery were used to generate the prechemotherapy and postchemotherapy tissue microarrays. Tumor grade was evaluated using the modified Bloom and Richardson system (15). The immunohistochemical methodology used for routine markers as ER, PgR, c-erbB2, Ki67, p53, and bcl-2 is fully described elsewhere (16). HIF-1α protein was detected using the ESEE 122 (IgG1 monoclonal antibody; dilution, 1:40) monoclonal antibody and carbonic anhydrase IX with murine monoclonal antibody M75 (a kind gift from S. Pastorekova, Centre of Molecular Medicine, Institute of Virology, Slovak Academy of Sciences, Bratislava, Slovak Republic) at a dilution of 1:50 for 30 min (17). The immunostaining for HIF-1α and carbonic anhydrase IX were quantified in carcinoma cells by semiquantitative scoring as previously described (18, 19). Immunohistochemical staining for HAI-2 was done with the murine monoclonal antibody (R&D Systems Ltd., Abingdon, United Kingdom), after antigen retrieval (autoclaving for 5 min in 10 nmol/L of citrate buffer, pH 6.0), at the concentration of 10 μg/mL diluted in TBS containing 1% bovine serum albumin for 16 h at 4° C. Negative controls consisted of (a) omission of the primary antibody, (b) primary antibody on MDA MB-435 cell line, (c) IgG1 isotype matched negative control antibody (Dako Ltd., Ely, United Kingdom). Positive controls consisted of staining with primary antibody–embedded paraffin section of MDA MB 231. For the secondary antibody, a polymer from the Envision horseradish peroxidase kit (Dako) was used. Immunostaining for HAI-2 was quantified in carcinoma cells by semiquantitative scoring. In brief, for HAI-2, a score for the intensity of 0 to 2 (0, no staining; 1, weak staining; 2, strong staining) was given. For all comparisons with survival and response, any staining was counted as positive.

In vitro experiments. A panel of human breast cancer cell lines (MDA MB 231, MDA MB 468, MDA MB 435, SKBR3, MCF7, T47D, ZR75, and BT474) were studied. Additional human renal cell lines expressing VHL (RCC4/VHL, 786-0/VHL, and Caki-1/VHL) or empty vector (RCC4/EV, 786-0/EV, and Caki-2/EV) and 293T were also analyzed. Cell lines were obtained from the Cancer Research UK cell service, and were grown in DMEM supplemented with 10% FCS (Globepharm, Esher, United Kingdom) and l-glutamine (2 μmol/L; Invitrogen, Paisley, United Kingdom), except for BT474, which were cultured in RPMI 1640 with identical supplementation. Experiments were done in normoxia (humidified air with 5% CO2) or hypoxia [Napco 7001 incubator (Winchester, VA) with 0.1% O2, 5% CO2, and balanced N2].

The protein and the mRNA used for the analysis of HAI-2 expression were extracted from the same experiments. Protein was detected using monoclonal antibody to HAI-2 according to the manufacturer's instructions (R&D Systems). As a loading control, tubulin was assessed with a mouse monoclonal antibody to β-tubulin (Sigma-Aldrich Company Ltd., Gillingham, United Kingdom) at 1:1,000. Overnight primary antibody incubation was followed by incubation with goat anti-mouse horseradish peroxidase (Dako). Blots showed two distinct bands at 35 and 23 kDa, as previously reported (20).

The methods we used for the extraction, quantification, and evaluation of the quality of RNA were as previously described (21). cDNA was synthesized by reverse transcribing total RNA using the High Capacity cDNA archive kit (Applied Biosystems Warrington, United Kingdom).

RNAi treatment of cells and transfection procedures. HIF-1α or inverted control duplex was diluted to give a final concentration of 20 nmol/L in Opti-Mem I (Invitrogen). The HIF-1α sense- and antisense-stranded small interfering RNA (siRNA) oligonucleotides were designed to specifically target the HIF-1α mRNA, whereas the inverted control did not target any known gene. Details of the oligonucleotides (Eurogentec Ltd., Hythe, Southhampton, United Kingdom) were as follows: HIF-1α antisense, 5′-UCAAGUUGCUGGUCAUCAGdTdT-3′; HIF-1α sense, 5′-CUGAUGACCAGCAACUUGAdTdT-3′; control antisense, 5′-GACUACUGGUCGUUGAACUdTdT-3′ and control sense, 5′-AGUUCAACGACCAGUAGUCdTdT-3′. Transfection of siRNA duplexes was done with cells at 30% to 40% confluency using 24 μL of OligofectAMINE transfection reagent (Invitrogen). Protein assay and gene expression analyses were done 24 h after siRNA transfection. Normoxic and hypoxic pressure were applied.

RNase protection assay. The RNase protection assay was used as a means of assessing patterns of mRNA expression for HAI-2. The full-length cDNA sequence of HAI-2 (331 bp) was cloned into pCRII-TOPO (Invitrogen). U6 small nRNA (accession no. X0136) was used as a loading control gene. The RNase protection assay protocol and generation of 32P-labeled RNA probes to HAI-2 and U6 small nRNA have been processed and described by Peterson et al. (22). Protected fragments were resolved on an 8% polyacrylamide gel and analyzed on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

Real-time quantitative PCR. Real-time quantitative PCR assay and relative quantitation of gene expression were done as previously described (21). Eukaryotic 18S rRNA was used as a reference gene to normalize for differences in the amount of total RNA in each sample. The following primer/probe kits were purchased as Assays-on-Demand from Applied Biosystems: HAI-2 (Hs00173936_m1) and 18S (Hs99999901_s1). The reaction efficiency for each gene was calculated after obtaining standard curves for each PCR reaction by making 2-fold serial dilutions covering the range equivalent to 20 to 0.625 ng of RNA (5-0.125 ng for 18S rRNA). All calculations were based on the mean value of PCR reactions done in triplicate. The comparator for the cell line was the median breast cancer cell in normoxia.

In situ hybridization. Radioactive in situ hybdrization on paraffin-embedded sections was done following the method described by Poulsom et al. (23). In brief, 4 μmol/L of formalin-fixed paraffin sections were dewaxed, rehydrated, and permeabilized with proteinase K (20 μg/mL). To these sections, 106 cpm of 35S-labeled single stranded RNA riboprobe was added. The probe used in the in situ hybridization for HAI-2 was a 331 bp fragment. The signal was visualized using dark-field microscopy and compared with Giemsa-counterstained light-field sections.

Statistics. χ2 test, χ2 test for trend, and Fisher exact test were used when indicated to perform comparisons of proportions. Non–paired Student's t test and Kruskal-Wallis ANOVA were done when indicated to compare continuous variables. Multivariate logistic regression was used to identify covariates independently associated to clinically complete response. All variables included in multivariate analyses were dichotomized variables with the exception of Ki67. This latter variable had a left-skewed distribution and was modeled using log transformation. Disease-free survival was calculated from randomization to the occurrence of disease relapse. Patients were censored if they were free from recurrence at the last follow-up. The disease-free survival curve was estimated using the Kaplan-Meier method. Unadjusted differences were assessed with the log-rank test. All P values reported were two-sided; values <0.05 were considered statistically significant.

The statistical analysis for the biological and clinical part was done using the GraphPad Prism version 3.0 and Statistica for Windows (Tulsa, OK) packages, respectively.

Regulation of HAI-2 expression in cancer cell lines by hypoxia. To investigate the potential reciprocal regulation of HAI-2, we compared renal cancer cell lines with VHL mutation and paired controls transfected with empty vector or wild-type VHL. ER-positive and -negative breast cancer cell lines were studied. In this panel of cell lines, we observed no change of mRNA or protein expression under 0.1% hypoxia (Fig. 1), with the exception of SKBR3 which is characterized by amplified c-erbB2. Significant hypoxic induction of HAI-2 mRNA was not only confirmed by real-time quantitative PCR in SKBR3 (P = 0.001) but also in another c-erb-B2–amplified cell line, BT474, when compared with normoxia (P = 0.007). There was no significant up-regulation of HAI-2 in the c-erbB2–negative breast cancer cell lines, MCF-7 or MDA MB 231 (Fig. 2A). The protein levels mirrored the results obtained by real-time quantitative PCR, with an increase of HAI-2 protein expression under hypoxia in both c-erbB2–positive cell lines (Fig. 2B).

Fig. 1.

RNase protection assay and Western blot for HAI-2 in renal cancer cell lines (A and C, respectively) and in breast cancer cell lines (B and D, respectively). Only in the c-erbB2–positive cell lines such as SKBR3 was there up-regulation of HAI-2 mRNA and protein under hypoxic conditions.

Fig. 1.

RNase protection assay and Western blot for HAI-2 in renal cancer cell lines (A and C, respectively) and in breast cancer cell lines (B and D, respectively). Only in the c-erbB2–positive cell lines such as SKBR3 was there up-regulation of HAI-2 mRNA and protein under hypoxic conditions.

Close modal
Fig. 2.

A, up-regulation of HAI-2 mRNA in c-erbB2–positive cell lines as SKBR3 (ER−/c-erbB2+) and BT474 (ER−/c-erbB2+) evaluated by real-time quantitative PCR. HAI-2 mRNA is expressed in arbitrary units. B, up-regulation of HAI-2 protein levels in the same breast cancer cell lines.

Fig. 2.

A, up-regulation of HAI-2 mRNA in c-erbB2–positive cell lines as SKBR3 (ER−/c-erbB2+) and BT474 (ER−/c-erbB2+) evaluated by real-time quantitative PCR. HAI-2 mRNA is expressed in arbitrary units. B, up-regulation of HAI-2 protein levels in the same breast cancer cell lines.

Close modal

Laughner et al. (24) reported that HER2 overexpression was associated with an increased expression of HIF-1α protein, due to either increased synthesis or decreased degradation. To determine whether HAI-2 expression was mediated by a HIF-1α–dependent mechanism, SKBR3 cells were transfected with siRNA targeting HIF-1α. Quantitative real-time PCR showed a significant reduction in HAI-2 expression between treated samples and controls (scramble and mock) both in normoxia (P = 0.009 and P = 0.003, respectively) and in hypoxia (P = 0.002 and P = 0.003, respectively; Fig. 3A). Western blotting showed a down-regulation of HAI-2 protein level expression in SKBR3 cells treated with HIF-1α siRNA in both conditions. In the scramble and mock controls, HAI-2 expression was not affected and the protein level was higher in hypoxia compared with normoxia (Fig. 3B). Considering that HAI-2 was not overexpressed in renal cancer cells with VHL mutation (which constitutively expresses HIF-1α) and that HIF-1 α is relevant for HAI-2 expression, these results suggested that HAI-2 induction was also mediated by complementary, cell type–specific pathways such as c-erbB2. To investigate the clinical significance of this observation of HAI-2–selective regulation by hypoxia in erbB2-positive cells, we carried out the following studies.

Fig. 3.

A, down-regulation of HAI-2 mRNA in SKBR3 c-erbB2–positive cell lines treated with HIF-1α siRNA. HAI-2 mRNA is expressed in arbitrary units. B, the expression of HAI-2 protein in SKBR3-treated HIF-1α siRNA was reduced by either normoxia or hypoxia compared with controls.

Fig. 3.

A, down-regulation of HAI-2 mRNA in SKBR3 c-erbB2–positive cell lines treated with HIF-1α siRNA. HAI-2 mRNA is expressed in arbitrary units. B, the expression of HAI-2 protein in SKBR3-treated HIF-1α siRNA was reduced by either normoxia or hypoxia compared with controls.

Close modal

HAI-2 in situ hybridization and immunohistochemistry of breast tissues.In situ hybridization for HAI-2 (Fig. 4) was done on five invasive breast cancers, two in situ ductal carcinomas and two normal tissues in order to correlate mRNA expression with detection of protein by immunohistochemistry, and to validate the HAI-2 antibody for semiquantitative analysis of tissue sections. HAI-2 was homogeneously present in all tumor cells with no expression in adjacent normal tissue. Among the immunohistochemistry-positive tumors, three showed strong reactivity with HAI-2–specific RNA probe after in situ hybridization, and two showed moderate reactivity. For normal breast tissues, the absence of immunohistochemical staining was mirrored by in situ hybridization with no signal detected. The MDA MB 231 cell line (which stained positive) and MDA MB-435 (which was negative), served as further controls, matching the Western blot findings.

Fig. 4.

Expression of HAI-2 in invasive breast cancer tissues; in situ hybridization (ISH) of HAI-2. Dark-field (A and D) and light-field (B and E). Immunohistochemistry (IHC) for HAI-2 in parallel sections to the in situ hybridization shows positive invasive ductal carcinoma of no special type (C) and negative staining in normal breast. HAI-2 expression in breast cancer cells (G), MDA MB 231 was used as a positive control (1) and MDA MB 435 as negative controls (3). Also, an IgG1 isotype–matched negative control antibody was used (2). HAI-2 had different expression profiles in invasive breast cancer tumors according to c-erbB2 status. Invasive ductal breast cancer of no special type, HAI-2 expression in c-erbB2–negative invasive ductal carcinoma of no special type (H). Invasive ductal breast cancer of no special type, HAI-2 expression in c-erbB2–positive tumor (I).

Fig. 4.

Expression of HAI-2 in invasive breast cancer tissues; in situ hybridization (ISH) of HAI-2. Dark-field (A and D) and light-field (B and E). Immunohistochemistry (IHC) for HAI-2 in parallel sections to the in situ hybridization shows positive invasive ductal carcinoma of no special type (C) and negative staining in normal breast. HAI-2 expression in breast cancer cells (G), MDA MB 231 was used as a positive control (1) and MDA MB 435 as negative controls (3). Also, an IgG1 isotype–matched negative control antibody was used (2). HAI-2 had different expression profiles in invasive breast cancer tumors according to c-erbB2 status. Invasive ductal breast cancer of no special type, HAI-2 expression in c-erbB2–negative invasive ductal carcinoma of no special type (H). Invasive ductal breast cancer of no special type, HAI-2 expression in c-erbB2–positive tumor (I).

Close modal

Relationship between HAI-2 expression and clinical and immunohistochemical prognostic variables. HAI-2 was assessed in 191 of the 211 patients prospectively enrolled in the trial (90.0%) by immunohistochemistry. For the remaining 20 patients, the blocks had insufficient material or could not be located. Patient characteristics are shown in Table 1. Ninety-two patients were randomized in the EPI arm, whereas 99 patients were randomized in the EPI-TAM arm. One hundred and seventy-seven patients had HAI-2 evaluated at baseline, 144 patients had HAI-2 assessed at residual tumor, whereas 130 patients had HAI-2 assessed both before and after treatment. HAI-2 immunostaining was detected in 60 tumor samples collected before treatment (33.9%) and in 49 tumor samples collected afterwards (34.0%).

Table 1.

Patient characteristics

No. randomized 191 
EPI arm 92 
EPI-TAM arm 99 
Grading  
    2 49 (26.3%) 
    3 137 (73.7%) 
T stage  
    T2 147 (77.0%) 
    T3 44 (23.0%) 
N status  
    N0 108 (56.5%) 
    N1 83 (43.5%) 
    ER− 40 (21.0%) 
    ER+ 150 (79.0%) 
    PgR− 98 (51.6%) 
    PgR+ 92 (48.4%) 
    p53− 95 (50.0%) 
    p53+ 95 (50.0%) 
    c-erbB2− 141 (73.8%) 
    c-erbB2+ 50 (26.2%) 
    bcl2− 52 (27.4%) 
    bcl2+ 138 (72.6%) 
Response to treatment (clinical)  
    Complete response 35 (18.4%) 
    Partial response 113 (59.5%) 
    No response 42 (22.1%) 
    Not evaluable 
Pathologically complete response 7 (3.6%) 
Baseline, HAI-2 (177)  
    0 117 (66.1%) 
    1 31 (17.5%) 
    2 29 (16.9%) 
After treatment, HAI-2 (144)  
    0 95 (66.0%) 
    1 25 (17.4%) 
    2 24 (16.6%) 
No. randomized 191 
EPI arm 92 
EPI-TAM arm 99 
Grading  
    2 49 (26.3%) 
    3 137 (73.7%) 
T stage  
    T2 147 (77.0%) 
    T3 44 (23.0%) 
N status  
    N0 108 (56.5%) 
    N1 83 (43.5%) 
    ER− 40 (21.0%) 
    ER+ 150 (79.0%) 
    PgR− 98 (51.6%) 
    PgR+ 92 (48.4%) 
    p53− 95 (50.0%) 
    p53+ 95 (50.0%) 
    c-erbB2− 141 (73.8%) 
    c-erbB2+ 50 (26.2%) 
    bcl2− 52 (27.4%) 
    bcl2+ 138 (72.6%) 
Response to treatment (clinical)  
    Complete response 35 (18.4%) 
    Partial response 113 (59.5%) 
    No response 42 (22.1%) 
    Not evaluable 
Pathologically complete response 7 (3.6%) 
Baseline, HAI-2 (177)  
    0 117 (66.1%) 
    1 31 (17.5%) 
    2 29 (16.9%) 
After treatment, HAI-2 (144)  
    0 95 (66.0%) 
    1 25 (17.4%) 
    2 24 (16.6%) 

In a univariate analysis, HAI-2 expression at baseline was significantly positively associated with T status, N status, and c-erbB2 expression (P < 0.004, P < 0.01, and P < 0.05, respectively) but was not correlated with tumor grade, p53, Ki67, bcl2, and steroid hormone receptor status (all P > 0.05; Table 2). HAI-2 was significantly higher in c-erbB2 positive than negative tumors. There was a significant positive relationship between HAI-2 and carbonic anhydrase IX expression (P = 0.01; a HIF-1α regulated protein; ref. 25) and a similar trend with HIF-1α (P = 0.059; χ2 for proportion of positive staining), although the latter just failed to attain significance which may reflect the shorter half-life of HIF-1α protein (in minutes) compared with carbonic anhydrase IX (38 h).

Table 2.

Relationship between HAI-2 expression and clinical and immunohistochemical variables (univariate analysis)

HAI-2 Intensity
P
012
Grading     
    2 30/113 (26.5%) 9/30 (30.0%) 7/29 (24.1%) 0.90* 
    3 83/113 (73.5%) 21/30 (70.0%) 22/29 (75.9%)  
p53 54/116 (46.5%) 17/31 (54.8%) 16/29 (55.2%) 0.97* 
c-erbB2 45/117 (38.5%) 16/31 (51.6%) 17/29 (58.6%) 0.05* 
bcl2 90/116 (77.6%) 20/31 (64.5%) 21/29 (72.4%) 0.37* 
ER 93/116 (80.2%) 25/31 (80.6%) 24/29 (82.7%) 0.76* 
PgR 59/116 (50.1%) 16/31 (51.6%) 11/29 (37.9%) 0.28* 
Ki67 mean (95% confidence interval) 22.5 (18.8-26.2) 19.0 (15.7-22.3) 23.8 (15.2-32.3) 0.80 
HIF-1α 84/110 (76.4%) 29/31 (93.5%) 24/29 (82.7%) 0.18* 
CAIX 20/106 (18.9%) 9/31 (29.0%) 12/29 (41.4%) 0.01* 
T2 97/117 (82.9%) 22/31 (71.0%) 17/29 (58.6%) <0.004* 
T3-4 20/117 (17.1%) 9/31 (29.0%) 12/29 (41.4%)  
N+ 44/117 (37.6%) 16/31 (51.6%) 18/29 (62.1%) 0.01* 
HAI-2 Intensity
P
012
Grading     
    2 30/113 (26.5%) 9/30 (30.0%) 7/29 (24.1%) 0.90* 
    3 83/113 (73.5%) 21/30 (70.0%) 22/29 (75.9%)  
p53 54/116 (46.5%) 17/31 (54.8%) 16/29 (55.2%) 0.97* 
c-erbB2 45/117 (38.5%) 16/31 (51.6%) 17/29 (58.6%) 0.05* 
bcl2 90/116 (77.6%) 20/31 (64.5%) 21/29 (72.4%) 0.37* 
ER 93/116 (80.2%) 25/31 (80.6%) 24/29 (82.7%) 0.76* 
PgR 59/116 (50.1%) 16/31 (51.6%) 11/29 (37.9%) 0.28* 
Ki67 mean (95% confidence interval) 22.5 (18.8-26.2) 19.0 (15.7-22.3) 23.8 (15.2-32.3) 0.80 
HIF-1α 84/110 (76.4%) 29/31 (93.5%) 24/29 (82.7%) 0.18* 
CAIX 20/106 (18.9%) 9/31 (29.0%) 12/29 (41.4%) 0.01* 
T2 97/117 (82.9%) 22/31 (71.0%) 17/29 (58.6%) <0.004* 
T3-4 20/117 (17.1%) 9/31 (29.0%) 12/29 (41.4%)  
N+ 44/117 (37.6%) 16/31 (51.6%) 18/29 (62.1%) 0.01* 

NOTE: Most of the factors assessed were not statistically relevant except T and N status, CAIX, and c-erbB2 expression.

*

χ2 for trend.

Kruskall-Wallis ANOVA.

HAI-2 expression and response to treatment. Among the 177 patients with HAI-2 assessed at baseline, one patient refused to continue the treatment after the first cycle and was not assessable for response. One hundred and thirty-eight out of 176 assessable cases (78.4%) attained a clinical response (complete + partial), 33 patients (18.7%) attained a complete clinical, 105 patients (59.7%) attained a partial response, and 6 patients (3.4%) attained a pathologically complete response. Table 3 shows the distribution of disease response according to HAI-2 immunostaining. The overall clinical response (complete + partial response) was significantly inversely correlated with the intensity of HAI-2 expression (P = 0.03), mainly due to the greater complete clinical response rate in HAI-2–negative compared with HAI-2–positive tumors (P = 0.01).

Table 3.

Relationship between HAI-2 expression and response to treatment

HAI-2 Intensity
P
012
Clinical     
    Complete response 28/116 (24.1%) 3/31 (9.7%) 2/29 (6.9%) 0.01* 
    Partial response 69/116 (59.5%) 18/31 (58.1%) 18/29 (62.1%)  
    No response 19/116 (16.4%) 10/31 (32.2%) 9/29 (31.0%) 0.03 
Pathological     
    Complete response 6/116 (5.2%) 0/31 0/29 <0.08 
HAI-2 Intensity
P
012
Clinical     
    Complete response 28/116 (24.1%) 3/31 (9.7%) 2/29 (6.9%) 0.01* 
    Partial response 69/116 (59.5%) 18/31 (58.1%) 18/29 (62.1%)  
    No response 19/116 (16.4%) 10/31 (32.2%) 9/29 (31.0%) 0.03 
Pathological     
    Complete response 6/116 (5.2%) 0/31 0/29 <0.08 
*

χ2 for trend for complete response rate.

χ2 for trend for overall (complete + partial) response rate.

Fisher exact test of HAI-2–positive versus –negative tumors.

The predictive role of HAI-2 expression for clinically complete response was confirmed after adjusting for tumor grade, T, N, bcl2, p53, steroid hormone receptors, c-erbB2, Ki67, and treatment administered in multivariate analysis (odds ratio, 0.4; 95% confidence interval, 02-0.8; P = 0.016; Table 4). After a median follow-up of 53 months, 45 of 183 patients relapsed (24.6%) and 22 (12.0%) died of disease, but HAI-2 was not related to relapse-free or overall survival.

Table 4.

Multivariate logistic analysis of independent factors predictive for clinical complete response

Odds ratio (95% confidence intervals)P
Variables in the model   
    Grading 2.9 (0.9-8.7) 0.058 
    HAI-2 0.4 (0.2-0.8) 0.016 
Variables which failed to enter the model   
    p53 1.1 (0.4-2.8) 0.587 
    Bcl2 1.2 (0.3-4.9) 0.983 
    ER 0.7 (0.2-2.0) 0.750 
    PgR 1.4 (0.5-4.0) 0.485 
    N status 0.7 (0.3-1.6) 0.353 
    Treat 1.7 (0.7-3.8) 0.2221 
    Log Ki67 2.8 (0.7-11.2) 0.275 
    c-erbB2 0.6 (0.3-1.4) 0.259 
    T 0.4 (0.1-1.4) 0.148 
Odds ratio (95% confidence intervals)P
Variables in the model   
    Grading 2.9 (0.9-8.7) 0.058 
    HAI-2 0.4 (0.2-0.8) 0.016 
Variables which failed to enter the model   
    p53 1.1 (0.4-2.8) 0.587 
    Bcl2 1.2 (0.3-4.9) 0.983 
    ER 0.7 (0.2-2.0) 0.750 
    PgR 1.4 (0.5-4.0) 0.485 
    N status 0.7 (0.3-1.6) 0.353 
    Treat 1.7 (0.7-3.8) 0.2221 
    Log Ki67 2.8 (0.7-11.2) 0.275 
    c-erbB2 0.6 (0.3-1.4) 0.259 
    T 0.4 (0.1-1.4) 0.148 

Effect of treatment on HAI-2 immunostaining. In the 130 matched patients with HAI-2 assessed at baseline and post-chemotherapy residual disease, HAI-2 showed a modest but significant change on therapy. HAI-2 positivity was present in 55 baseline tumor samples (42.3%) and in 49 residual tumor samples after chemotherapy (37.7%; P = 0.02, Mc Nemar χ2). On an individual basis, however, the changes were less consistent because 29 HAI-2–positive tumors become negative after treatment whereas the opposite (negative to positive) was observed in 23 patients.

HGF in association with its receptor (c-Met) plays an important role in the invasive growth of tumor cells (26). The induction of biologically active HGF is processed by a member of the kringle-serine proteinase superfamily, the HGF activator, which is modulated by HAIs. Because the levels of other proteinase inhibitors such as plasminogen activator inhibitor (PAI no. 0) or tissue inhibitor of metalloproteinase are mediated by hypoxia and HIF-1α (27, 28), we examined HAI-2 expression in human renal and breast cancer cell lines under both normoxia and hypoxia. Only cancer cells overexpressing c-erbB2 (SKBR3 and BT474) showed an increase of HAI-2 mRNA and protein levels under hypoxic conditions. At the molecular level, Laughner et al. (24) showed that HER2 signaling increases HIF-1α protein synthesis, such that the combination of HER2 overexpression and hypoxia had a synergistic effect on HIF-1α protein concentration. We and others have previously observed a direct association of c-erbB2 with HIF-1α (29, 30). Our findings that HAI-2 expression is up-regulated under hypoxic conditions in the presence of c-erbB2 expression may result from an analogous HIF-1α induction and synergy increasing HAI-2 levels. The knockdown by RNAi against HIF-1α confirms the role of this pathway in c-erbB2–positive cells. Our results suggest that c-erbB2 expression may regulate the hypoxia response and this should be further investigated.

We analyzed the expression and clinical association of c-erbB2 in a series of patients treated in a trial of neoadjuvant therapy to relate HAI-2 expression with response to treatment and survival. We observed a positive correlation between the expression of HAI-2 and T status, N status, and c-erbB2 expression—confirming that the in vitro observation could be a model for clinical findings.

Nevertheless, these findings contrast with those of Parr et al. who reported HAI-2 to be inversely correlated with tumor grade and node status (13). This may be due to the different populations studied (our 211 consecutive patients compared with their 100 patients) and the mRNA methodology as opposed to immunohistochemistry used to measure HAI-2 levels. Parr et al. also showed that HAI-1 and HAI-2 showed more intense staining in the normal tissues than in breast cancer tissues, whereas we observed the opposite, a pattern matched by in situ hybridization. A possible explanation for this discrepancy might be the antibody used for immunohistochemistry. Nagaike et al. reported different results on the same sections using three kinds of anti–HAI-1 antibodies. Two antibodies recognized the extracellular domain and one antibody recognized the intracytoplasmic domain of HAI-1. The latter showed the similar staining pattern obtained in in situ hybridization (31). The antibody we used recognizes the intracellular domain, avoiding the potential of additionally identifying cleaved extracellular HAI-2 (31).

Recently, Parr et al. described the important roles of HAI-1 and HAI-2 in cancer metastasis (32). They showed that overexpressing HAI-1 or HAI-2 in fibroblasts decreased the production of bioactive HGF and suppressed the effect on in vitro invasion and in vivo growth in xenografts. Also, the inhibition of production of HAI-1 or HAI-2 in breast cells increased growth. However, others have found the activating role of HAI-1 in a different model. The expression of the membrane form of HAI-1 correlated significantly with enhanced in vivo tumorigenicity of glioblastoma cell lines (33). In colon cancer cells, an up-regulation (at the invasion front) of membrane form HAI-1 immunoreactivity was detected, suggesting that the membrane form HAI-1 modulates tumor-host interaction in favor of the cancer cells in the invasion field (31). The membrane form of HAI-1 was also found to be up-regulated in injured and regenerating epithelial tissues (34). These observations are likely to be related to another role of HAI-1 in directly activating the protease matriptase. Although not investigated for HAI-2, HAI-1 regulates the activation and expression of matriptase even though it can inhibit the matriptase activity. HAI-1 inhibits matriptase, but paradoxically, is also required for its activation and this depends on the level of HAI. HAI was needed for the trafficking of matriptase to the cell surface. Interestingly, the role of HAI-1 was not required if matriptase was mutated, hence, it had no protease activity, implying that HAI-1 had to prevent inappropriate activity during proteolytic systems and trafficking (35, 36). Clearly, the HGF/HGF activator/HAI pathways are complex and the in vivo analysis of human cancer and its stroma are warranted to better understand these mechanisms.

Our results raise the possibility that c-erbB2 may partly alter tumor progression through HAI-2. There is ample precedent that up-regulated secretion of proteases is accompanied by the up-regulation of their cognate inhibitors, and that the latter are essential for the effectiveness of the proteases, often through additional properties. Previously, one mechanism that has been proposed by which c-erb-B2 regulates tumor proteolysis is through urokinase-type plasminogen activator regulation (37), in which PAI-1 plays a critical role (38). The parallels with PAI-1 are strong, with high levels being associated with the development of a more aggressive tumor phenotype (39). In cancer, the plasminogen activator system with its key components urokinase-type plasminogen activator, its cell surface receptor urokinase-type plasminogen activator regulation or CD87, and its inhibitor PAI-1 plays a major role in tumor invasion and metastasis. Elevated levels of these factors in tumor tissue are associated with aggressiveness and poor patient outcome (38). Regarding PAI-1, it is the primary regulator of plasminogen activation and therefore essential factor regulating physiologic thrombotic/fibrinolytic balance in the vasculature. However, PAI-1 also has a number of unusual properties for a protease inhibitor (40). In tissues, via interactions with integrins and/or extracellular matrix components, it orchestrates cell adhesion and migration (41), and therefore, PAI-1 could be considered a regulator of tumor invasion and metastasis as well as cancer-related angiogenesis.

Similar observations regarding the apparently paradoxical inhibitory and stimulatory effects of tissue inhibitors of metalloproteinases in tumorigenesis were described by Jiang et al. (42), and were related to the antiapoptotic role of tissue inhibitors of metalloproteinases and the requirement of tissue inhibitors of metalloproteinases for proteolytic activation of certain matrix metalloproteinases.

Showing that c-Met is inducible by hypoxia and enhances movement and migration under hypoxia (9) suggests another possible role for HAI-2. That is, to prevent excessive activation of its ligand locally because this would act to compete for a concentration gradient of HGF, attracting the tumor cells away from the hypoxic area. A complex balance of activation would be required to maintain a HGF gradient with perhaps a small increase in autocrine stimulation, but not enough to overcome a gradient attracting cells away from hypoxia.

Primary chemotherapy administered to patients with breast cancer is a useful model to identify baseline features that are able to predict which patients are most likely to benefit or not from cytotoxic treatment. In addition, tumor biopsy specimens obtained in matched pair cases at diagnosis and definitive surgery provide information on the interaction between biological markers and treatment. In this study, the maximum standard dose of epirubicin was given as preoperative chemotherapy. At baseline, there was a direct relationship between HAI-2 expression and T status, N status, and c-erbB2 expression. HAI-2 was also significantly correlated with the hypoxia marker, carbonic anhydrase IX (43), and also showed a positive relationship with HIF-1α. These data further support the in vitro findings of the association of HAI-2 protein with c-erbB2, especially under hypoxic conditions.

Our study is the first to show a relationship between HAI-2 expression and treatment outcome. High expression of HAI-2 was a significant predictor for poor clinical complete response rate, not only in univariate analysis but also in multivariate analysis after adjusting for standard prognostic and predictive clinical and biological variables, in spite of the association of HAI-2 with other poor prognosis features. Pathologically complete response was observed in only six cases, but of note, they were all HAI-2–negative. Thus, HAI-2 expression may be a key regulator of the overall activity of the HGF pathway, which has been shown to strongly influence drug resistance to Adriamycin and camptothecin (44, 45). HGF seems to be a potential mediator of apoptosis and cell survival caused by chemotherapeutic agents in tumor cells and some epithelial cells (46). HGF-induced drug resistance may result not only from the activation of an antiapoptosis pathway, but also from an increase in the repair of DNA strand breaks (44, 47). In spite of the negative relationship with disease response, HAI-2 was not related to disease-free survival or survival. The power of the analysis is limited due to the low percentage of events, in addition, all patients received adjuvant treatments, thus, introducing a confounding factor. Another possibility which should be noted, however, is that predictive factors for disease response are not necessarily prognostic factors.

Although, in cancer, the exact role of HAI-2 is still controversial, it is reasonable to postulate that HAI-2 may have an important role in the regulation of HGF activity and of other proteinase activities in breast cancer. In keeping with this notion, HAI-2 up-regulation in breast cancer was compared with nonneoplastic breast tissues, data mirroring the increase in PAIs and tissue inhibitors of metalloproteinases in cancer that are likely to protect tumor cells from excessive activated proteolysis. Indeed, our demonstration of hypoxic induction of HAI-2 further emphasizes the parallels between HAI-2 and the urokinase-type plasminogen activator system (48).

In conclusion, HAI-2 might become clinically relevant in breast cancers as a potential target for directed therapy (e.g., antibody or low molecular weight inhibitor approaches). However, because this is the first study analyzing the interaction with chemotherapy, further prospective studies are needed to validate this study and to elucidate the mechanisms of HAI-2 interaction with HGF regulation.

Grant support: In part by the Associazione Patologia Oncologica Mammaria, Cremona, Italy; the Association of Amici dell'Ospedale di Cremona; the Consiglio Nazionale Ricerche, Rome, Italy; Cancer Research UK (CRU no. 0); and the Victorian Breast Cancer Research Consortium, Australia.

The costs of publication of this article were defrayed in part by the 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
Naldini L, Weidner KM, Vigna E, et al. Scatter factor and hepatocyte growth factor are indistinguishable ligands for the MET receptor.
EMBO J
1991
;
10
:
2867
–78.
2
Matsumoto K, Nakamura T. Emerging multipotent aspects of hepatocyte growth factor.
J Biochem (Tokyo)
1996
;
119
:
591
–600.
3
Danilkovitch-Miagkova A, Zbar B. Dysregulation of Met receptor tyrosine kinase activity in invasive tumors.
J Clin Invest
2002
;
109
:
863
–7.
4
Yamashita J, Ogawa M, Yamashita S, et al. Immunoreactive hepatocyte growth factor is a strong and independent predictor of recurrence and survival in human breast cancer.
Cancer Res
1994
;
54
:
1630
–3.
5
Miyazawa K, Shimomura T, Kitamura A, Kondo J, Morimoto Y, Kitamura N. Molecular cloning and sequence analysis of the cDNA for a human serine protease reponsible for activation of hepatocyte growth factor. Structural similarity of the protease precursor to blood coagulation factor XII.
J Biol Chem
1993
;
268
:
10024
–8.
6
Shimomura T, Denda K, Kitamura A, et al. Hepatocyte growth factor activator inhibitor, a novel Kunitz-type serine protease inhibitor.
J Biol Chem
1997
;
272
:
6370
–6.
7
Kawaguchi T, Qin L, Shimomura T, et al. Purification and cloning of hepatocyte growth factor activator inhibitor type 2, a Kunitz-type serine protease inhibitor.
J Biol Chem
1997
;
272
:
27558
–64.
8
Kataoka H, Itoh H, Nuki Y, et al. Mouse hepatocyte growth factor (HGF) activator inhibitor type 2 lacking the first Kunitz domain potently inhibits the HGF activator.
Biochem Biophys Res Commun
2002
;
290
:
1096
–100.
9
Pennacchietti S, Michieli P, Galluzzo M, Mazzone M, Giordano S, Comoglio PM. Hypoxia promotes invasive growth by transcriptional activation of the met protooncogene.
Cancer Cell
2003
;
3
:
347
–61.
10
Kataoka H, Itoh H, Hamasuna R, Meng JY, Koono M. Pericellular activation of hepatocyte growth factor/scatter factor (HGF/SF) in colorectal carcinomas: roles of HGF activator (HGFA) and HGFA inhibitor type 1 (HAI-1).
Hum Cell
2001
;
14
:
83
–93.
11
Kang JY, Dolled-Filhart M, Ocal IT, et al. Tissue microarray analysis of hepatocyte growth factor/Met pathway components reveals a role for Met, matriptase, and hepatocyte growth factor activator inhibitor 1 in the progression of node-negative breast cancer.
Cancer Res
2003
;
63
:
1101
–5.
12
Oberst MD, Johnson MD, Dickson RB, et al. Expression of the serine protease matriptase and its inhibitor HAI-1 in epithelial ovarian cancer: correlation with clinical outcome and tumor clinicopathological parameters.
Clin Cancer Res
2002
;
8
:
1101
–7.
13
Parr C, Watkins G, Mansel RE, Jiang WG. The hepatocyte growth factor regulatory factors in human breast cancer.
Clin Cancer Res
2004
;
10
:
202
–11.
14
Zambetti M, Giacobone A, Terenziani M, et al. Sequential adriamycin and CMF in metastatic breast cancer.
Oncologist
1997
;
2
:
223
–7.
15
Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up.
Histopathology
1991
;
19
:
403
–10.
16
Bottini A, Berruti A, Brizzi MP, et al. Cytotoxic and antiproliferative activity of the single agent epirubicin versus epirubicin plus tamoxifen as primary chemotherapy in human breast cancer: a single-institution phase III trial.
Endocr Relat Cancer
2005
;
12
:
383
–92.
17
Pastorekova S, Zavadova Z, Kostal M, Babusikova O, Zavada J. A novel quasi-viral agent, MaTu, is a two-component system.
Virology
1992
;
187
:
620
–6.
18
Colpaert CG, Vermeulen PB, Fox SB, Harris AL, Dirix LY, Van Marck EA. The presence of a fibrotic focus in invasive breast carcinoma correlates with the expression of carbonic anhydrase IX and is a marker of hypoxia and poor prognosis.
Breast Cancer Res Treat
2003
;
81
:
137
–47.
19
Chia SK, Wykoff CC, Watson PH, et al. Prognostic significance of a novel hypoxia-regulated marker, carbonic anhydrase IX, in invasive breast carcinoma.
J Clin Oncol
2001
;
19
:
3660
–8.
20
Morris MR, Gentle D, Abdulrahman M, et al. Tumor suppressor activity and epigenetic inactivation of hepatocyte growth factor activator inhibitor type 2/SPINT2 in papillary and clear cell renal cell carcinoma.
Cancer Res
2005
;
65
:
4598
–606.
21
Patel NS, Li JL, Generali D, Poulsom R, Cranston DW, Harris AL. Up-regulation of δ-like 4 ligand in human tumor vasculature and the role of basal expression in endothelial cell function.
Cancer Res
2005
;
65
:
8690
–7.
22
Petersen NE, Larsen LK, Nissen H, et al. Improved RNase protection assay for quantifying LDL-receptor mRNA; estimation of analytical imprecision and biological variance in peripheral blood mononuclear cells.
Clin Chem
1995
;
41
:
1605
–13.
23
Poulsom R, Longcroft JM, Jeffery RE, Rogers LA, Steel JH. A robust method for isotopic riboprobe in situ hybridisation to localise mRNAs in routine pathology specimens.
Eur J Histochem
1998
;
42
:
121
–32.
24
Laughner E, Taghavi P, Chiles K, Mahon PC, Semenza GL. HER2 (neu) signaling increases the rate of hypoxia-inducible factor 1α (HIF-1α) synthesis: novel mechanism for HIF-1-mediated vascular endothelial growth factor expression.
Mol Cell Biol
2001
;
21
:
3995
–4004.
25
Wykoff CC, Beasley NJ, Watson PH, et al. Hypoxia-inducible expression of tumor-associated carbonic anhydrases.
Cancer Res
2000
;
60
:
7075
–83.
26
Trusolino L, Comoglio PM. Scatter-factor and semaphorin receptors: cell signalling for invasive growth.
Nat Rev Cancer
2002
;
2
:
289
–300.
27
Uchiyama T, Kurabayashi M, Ohyama Y, et al. Hypoxia induces transcription of the plasminogen activator inhibitor-1 gene through genistein-sensitive tyrosine kinase pathways in vascular endothelial cells.
Arterioscler Thromb Vasc Biol
2000
;
20
:
1155
–61.
28
Norman JT, Clark IM, Garcia PL. Hypoxia promotes fibrogenesis in human renal fibroblasts.
Kidney Int
2000
;
58
:
2351
–66.
29
Giatromanolaki A, Koukourakis MI, Sivridis E, et al. Relation of hypoxia inducible factor 1α and 2α in operable non-small cell lung cancer to angiogenic/molecular profile of tumours and survival.
Br J Cancer
2001
;
85
:
881
–90.
30
Bos R, Zhong H, Hanrahan CF, et al. Levels of hypoxia-inducible factor-1α during breast carcinogenesis.
J Natl Cancer Inst
2001
;
93
:
309
–14.
31
Nagaike K, Kohama K, Uchiyama S, et al. Paradoxically enhanced immunoreactivity of hepatocyte growth factor activator inhibitor type 1 (HAI-1) in cancer cells at the invasion front.
Cancer Sci
2004
;
95
:
728
–35.
32
Parr C, Jiang WG. Hepatocyte growth factor activation inhibitors (HAI-1 and HAI-2) regulate HGF-induced invasion of human breast cancer cells.
Int J Cancer
2006
;
119
:
1176
–83.
33
Miyata S, Uchinokura S, Fukushima T, et al. Diverse roles of hepatocyte growth factor activator inhibitor type 1 (HAI-1) in the growth of glioblastoma cells in vivo.
Cancer Lett
2005
;
227
:
83
–93.
34
Itoh H, Yamauchi M, Kataoka H, Hamasuna R, Kitamura N, Koono M. Genomic structure and chromosomal localization of the human hepatocyte growth factor activator inhibitor type 1 and 2 genes.
Eur J Biochem
2000
;
267
:
3351
–9.
35
Oberst MD, Chen LY, Kiyomiya K, et al. HAI-1 regulates activation and expression of matriptase, a membrane-bound serine protease.
Am J Physiol Cell Physiol
2005
;
289
:
C462
–70.
36
Lee MS, Kiyomiya K, Benaud C, Dickson RB, Lin CY. Simultaneous activation and hepatocyte growth factor activator inhibitor 1-mediated inhibition of matriptase induced at activation foci in human mammary epithelial cells.
Am J Physiol Cell Physiol
2005
;
288
:
C932
–41.
37
Allgayer H, Babic R, Gruetzner KU, Tarabichi A, Schildberg FW, Heiss MM. c-erbB-2 is of independent prognostic relevance in gastric cancer and is associated with the expression of tumor-associated protease systems.
J Clin Oncol
2000
;
18
:
2201
–9.
38
Duffy MJ, Duggan C. The urokinase plasminogen activator system: a rich source of tumour markers for the individualised management of patients with cancer.
Clin Biochem
2004
;
37
:
541
–8.
39
Dano K, Andreasen PA, Grondahl-Hansen J, Kristensen P, Nielsen LS, Skriver L. Plasminogen activators, tissue degradation, and cancer.
Adv Cancer Res
1985
;
44
:
139
–266.
40
Stefansson S, McMahon GA, Petitclerc E, Lawrence DA. Plasminogen activator inhibitor-1 in tumor growth, angiogenesis and vascular remodeling.
Curr Pharm Des
2003
;
9
:
1545
–64.
41
Chorostowska-Wynimko J, Skrzypczak-Jankun E, Jankun J. Plasminogen activator inhibitor type-1: its structure, biological activity and role in tumorigenesis (Review).
Int J Mol Med
2004
;
13
:
759
–66.
42
Jiang Y, Goldberg ID, Shi YE. Complex roles of tissue inhibitors of metalloproteinases in cancer.
Oncogene
2002
;
21
:
2245
–52.
43
Potter C, Harris AL. Hypoxia inducible carbonic anhydrase IX, marker of tumour hypoxia, survival pathway and therapy target.
Cell Cycle
2004
;
3
:
164
–7.
44
Meng Q, Mason JM, Porti D, Goldberg ID, Rosen EM, Fan S. Hepatocyte growth factor decreases sensitivity to chemotherapeutic agents and stimulates cell adhesion, invasion, and migration.
Biochem Biophys Res Commun
2000
;
274
:
772
–9.
45
Holt RU, Baykov V, Ro TB, et al. Human myeloma cells adhere to fibronectin in response to hepatocyte growth factor.
Haematologica
2005
;
90
:
479
–88.
46
Fan S, Wang JA, Yuan RQ, et al. Scatter factor protects epithelial and carcinoma cells against apoptosis induced by DNA-damaging agents.
Oncogene
1998
;
17
:
131
–41.
47
Fan S, Ma YX, Wang JA, et al. The cytokine hepatocyte growth factor/scatter factor inhibits apoptosis and enhances DNA repair by a common mechanism involving signaling through phosphatidyl inositol 3′ kinase.
Oncogene
2000
;
19
:
2212
–23.
48
Dimova EY, Samoylenko A, Kietzmann T. Oxidative stress and hypoxia: implications for plasminogen activator inhibitor-1 expression.
Antioxid Redox Signal
2004
;
6
:
777
–91.