1026
Urokinase System Expression in Gastric Carcinoma Prognostic Impact in an Independent Patient Series and First Evidence of Predictive Value in Preoperative Biopsy and Intestinal Metaplasia Specimens
Bianca C. M. Beyer, M.D.1 Markus Maria Heiss, M.D.2 Erich H. Simon, M.D.3 Klaus-Uwe Gruetzner, M.D.1 Rudolf Babic, M.D.4 Karl-Walter Jauch, M.D.1 Friedrich-Wilhelm Schildberg, Heike Allgayer, M.D., Ph.D.3,5
M.D.
1
1
Department of Surgery Grosshadern, LudwigMaximilians-University, Munich, Germany.
2
Department of Surgery, Cologne-Merheim, University Witten, Cologne-Merheim, Germany.
3
Department of Experimental Surgery/Surgical Oncology Mannheim, University Heidelberg, Mannheim, Germany.
4
Pathology/Cytology Institute, Deggendorf, Germany.
5
Department of Molecular Oncology, German Cancer Research Center (DKFZ), Heidelberg, Germany. This article contains parts of the dissertations of Bianca C. M. Beyer and Erich H. Simon, which were presented in fulfillment of “Dr. Med” for the Medical Faculty, Ludwig-Maximilians-University, Munich. The authors thank Dr. Ronald Kates, Otterfing, for critical appraisal, and Erika Hillerich for excellent help with the article. The first two authors contributed equally to this article. Address for reprints: Heike Allgayer, M.D., Ph.D., Department of Experimental Surgery, Joint Unit of Molecular Oncology of Solid Tumors—DKFZ (German Cancer Research Center), Heidelberg, Klinikum Mannheim, Ruprecht-Karls-University Heidelberg, Theodor-Kutzer-Ufer 1-3, 68135 Mannheim, Germany; Fax: (011) 49 6213833809; E-mail:
[email protected] Heike Allgayer was supported by Wilhelm-SanderStiftung (Munich), Auguste-Schaedel-Dantscher-Stiftung (Garmisch), Alfried Krupp von Bohlen und Halbach-Stiftung (Essen) and the National Cancer Institute (Bethesda, MD). Received June 29, 2005; revision received August 22, 2005; accepted September 20, 2005.
BACKGROUND. The prognostic relevance of urokinase-type plasminogen activator (u-PA), u-PA receptor (u-PAR), and plasminogen activator inhibitor 1 (PAI-1) in gastric carcinoma was demonstrated in an independent patient series. To the authors’ knowledge,the roles of these activators as predictors of aggressive phenotypes in preoperative biopsies, Helicobacter pylori infection, and intestinal metaplasia have to date not been investigated simultaneously in resected tumors. The objectives of the current study were 1) to demonstrate the prognostic relevance of u-PA, u-PAR, and PAI-1 in an independent series; 2) to evaluate u-PA system expression in preoperative biopsy specimens compared with resected tumors; and 3) to evaluate u-PA system expression in intestinal metaplasias and samples with H. pylori infection. METHODS. In 104 patients with gastric carcinoma (median follow-up, 68 mos), u-PA, u-PAR, and PAI-1 in tumors and metaplasias were evaluated immunohistochemically. Preoperative biopsies were evaluated in a subset of patients. Patients were screened for H. pylori (urease) and tumor cells in bone marrow (u-PAR/ CK18). RESULTS. u-PA and PAI-1 were confirmed as independent prognostic parameters, and u-PAR was associated with a trend toward a poor prognosis. u-PA system tumor expression was found to be correlated significantly with u-PAR in disseminated tumor cells and H. pylori-infected tumors, implicating a role of H. pylori in protease induction. There was a significant correlation noted between u-PA system staining between preoperative biopsies and the results in resected tumors. The expression of u-PAR and PAI-1 in intestinal metaplasias was found to be associated significantly with advanced tumor stage (depth of invasion; pathologic tumor status) and lymph node involvement (pathologic lymph node status) and was correlated significantly with u-PA system expression in tumors. CONCLUSIONS. To the author’s know the current study is the first to date to demonstrate that u-PA system expression may serve as a predictor of risk in intestinal metaplasias and preoperative biopsies, implicating consequences for neoadjuvant therapy. The independent impact on recurrence and survival and a correlation with u-PAR-expression of minimal residual disease were identified in this independent series. Cancer 2006;106:1026 –35. © 2006 American Cancer Society.
KEYWORDS: urokinase-type plasminogen activator system, gastric carcinoma, intestinal metaplasia, Helicobacter pylori, preoperative biopsies, minimal residual disease, prognosis.
T
he urokinase-type plasminogen activator (u-PA) system promotes invasion and metastasis in many carcinomas.l,2 Through plasminogen activation, u-PA, a 55-kilodalton (kD) serine protease, cleaves extracellular matrix components, including fibronectin, laminin, and Type IV collagen.1,2 The proteolytic efficacy of u-PA relies on interac-
© 2006 American Cancer Society DOI 10.1002/cncr.21682 Published online 24 January 2006 in Wiley InterScience (www.interscience.wiley.com).
u-PA System in Gastric Carcinoma/Beyer et al.
tions with other factors, among them, plasminogen activator inhibitor 1 (PAI-1) and the cell-surface receptor u-PA (u-PAR). u-PA binds to its 55– 60 kD, glycosylated u-PAR, activating plasminogen much more efficiently than the fluid-phase enzyme.3 Receptorbound u-PA is inactivated by PAI-1; the u-PAR/u-PA/ PAI complex is internalized into the cell with ␣-2 macroglobulin receptor and its ligand,4 and u-PAR is recycled to the cell surface. The interaction of u-PA system parameters is involved in several cellular functions, such as migration, chemotaxis, invasion, intravasation, metastasis, and angiogenesis. A strong association between u-PA system expression and the invasive phenotype has been demonstrated in a number of experimental studies.2,5,6 Correspondingly, numerous studies have demonstrated the overexpression of u-PA, u-PAR, and PAI-1 in human malignant tumors and/or surrounding stromal cells.6 –9 Prospective studies, including our own, have demonstrated a correlation between high u-PAR, u-PA, and/or PAI-1 expression levels with short survival and advanced tumor stages in malignancies such as breast carcinoma,10,11 lung carcinoma,12 colon carcinoma,13,14 and gastric carcinoma.9,15,16 Furthermore, our own data17–19 proposed u-PAR as one characteristic of metastatic phenotypes of disseminated tumor cells in bone marrow. Nonetheless, to our knowledge, few if any attempts have been made to demonstrate the prognostic impact of the u-PA system in independent, prospective patient series by the same research groups, although this is a prerequisite for identifying new factors as statistically powerful clinical markers. To date, unfortunately, continual improvements in the treatment of patients with gastric carcinoma in recent years have not led to a significant gain in overall prognosis. There remains an enormous difference between the 5-year survival rate among patients with early-stage gastric carcinoma (approximately 90%) and patients with all other forms (approximately 20%).20 Unfortunately, established prognostic factors (such as pathologic tumor [pT], pathologic lymph node [pN], and metastasis [M] classifications) have neither contributed to new therapeutic options nor provided an adequate basis for distinguishing risk subgroups with an especially aggressive kind of phenotype. This lack of risk-group discrimination may explain the unsatisfactory impact of (neo-)adjuvant therapy in the treatment of patients with gastric carcinoma (chemotherapy and/or radiotherapy) to date. For a more precise definition of prognostic and therapeutic subgroups, a refinement of morphologic risk classifications by molecular parameters would be of great benefit. In particular, markers are needed that
1027
are able to predict individual tumor biology very early in the diagnosis of gastric carcinoma—ideally, in preoperative biopsy specimens. In addition, to prevent gastric carcinoma, there is a diagnostic need to identify markers that are capable of predicting whether conditions, such as Helicobacter pylori infection, or potential precursor lesions, such as intestinal metaplasias, present a high risk of transforming into carcinoma. In our previous work on gastric carcinoma, we implicated the u-PA system as a predictor of poor prognosis9 and defined what to our knowledge is the first biologic high-risk groups by using PAI-1 and minimal residual disease as additional markers.16 The objectives of the current study were 1) to define the prognostic relevance of the u-PA system in gastric carcinoma and validate it for the first time in a prospective patient series, independent of previous series; 2) to determine associations of u-PA system expression with additional biologic parameters, such as evidence of H. pylori infection; 3) to study the correlation of u-PA system immunohistochemistry between resected gastric carcinomas and corresponding preoperative biopsies, determining its potential use in preoperative diagnosis; and, finally 4) to study u-PA system expression in intestinal metaplasia and determine its potential impact as a predictor of high-risk precursor lesions. To our knowledge, this is the first comprehensive study on the u-PA system in gastric carcinoma to provide a synthesis of these different aspects of diagnosis, prediction, and prognosis.
MATERIALS AND METHODS Patients and Tumors A prospective series of 104 patients underwent surgery for gastric carcinoma between December 1994 and December 1997. The mean patient age was 64 years (range, 25– 89 yrs), and the male:female ratio was 58: 46. Eighty-three patients (80%) underwent curative resection (R0) with radical lymph node dissection (compartments I and II), and 20% underwent palliative resection, including 17 patients who had microscopic residual disease and 2 patients who had macroscopic residual disease. According to International Union Against Cancer (UICC) criteria, 8.1% of patients had Stage IA disease, 16.3% of patients had Stage IB disease, 22.1% of patients had Stage II disease, 23.1% of patients had Stage IIIA disease, 13.5% of patients had Stage IIIB disease, and 13.5% of patients had Stage IV disease. According to the Lauren classification,21 49 patients (47.1%) had intestinal carcinoma, 42 patients (40.4%) had diffuse carcinoma, and 9 patients (8.7%) had mixed-type carcinoma. According to histopathological grading (G),
1028
CANCER March 1, 2006 / Volume 106 / Number 5
one tumor was well differentiated (Grade 1), 29 tumors were Grade 2, and 74 tumors were poorly differentiated (Grade 3). Lymphangiosis carcinomatosa (lymphangioinvasion) was present in 89 patients. According to the Borrmann classification,22 there were 5 Class I tumors, 13 Class II tumors, 75 Class III tumors, and 8 Class IV tumors. In 34 patients, tumors were localized to the cardia or fundus; and, in 74 patients, tumors were localized to the corpus or antrum. Sixteen patients received intraoperative radiation therapy (28 grays), and they all underwent curative resection. Four patients received neoadjuvant chemotherapy, and 18 patients received adjuvant chemotherapy. One patient received chemotherapy after undergoing noncurative resection. Prospective follow-up was performed at 6-month intervals (physical examination, ultrasound, gastroscopy, chest X-ray, hematology, and blood chemistry and the tumor markers carcinoembryonic antigen, CA 19-9, and CA 72-4). Tumor recurrence was diagnosed by biopsy, explorative surgery, or imaging.
Immunohistochemical Staining Preoperative biopsies and perioperatively resected material Four to seven preoperative biopsies were obtained from each patient. Tissues were snap-frozen in liquid nitrogen immediately after fixation on cork. After samples were cut into 4-m serial sections, slides were subjected to immunohistochemistry or were refrozen. Staining was performed as described previously.9,16 Briefly, slides were fixed in acetone, inactivated for endogenous peroxidase, and rehydrated. To prevent unspecific Fc-reactions, a preincubation was performed with 150 L horse serum in 10 mL Tris-buffered saline and 150 L of avidin/biotin serum for 30 minutes each. To block endogenous avidin/ biotin activity, we applied the avidin/biotin blocking kit (Vectastain; Vector Laboratories, Burlingame, CA). The primary antibody (mouse u-PA no. 394 [dilution, 1:1000]; u-PAR no. 3937 [dilution, 1:1000l]; PAI-1 no. 3783 [dilution, 1:800]) was incubated for 40 minutes. Vectastain avidin/biotin elite complex was applied for 40 minutes. Aminoethylcabazole (28 mg in 105 L of 0.5% hydrogen peroxidase) and 98 mL of acetate buffer (11 g 0.05 mol/L acetic acid plus 16.4 g sodium acetate [pH 5.9]) was added for 30 minutes. Slides were counterstained with hematoxylin, coated in glycerol, and dried. One section from each tumor that was treated with MLG/7S (Nordic, Tilburg, The Netherlands) against murine immunoglobulin G instead of the primary antibody in equimolar protein concentration served as a negative control; cell lines HT 29/SW 403, which have known strong expression, were used as positive controls.
All slides were coded and evaluated by two independent investigators. Analysis of staining was restricted to the reactions observed in tumor cells. Staining of normal stromal cells was not considered. Experience from our extensive previous analyses of the u-PA system revealed that staining could be quantified reproducibly and yielded a significant correlation of our scoring system with staining intensity.9,16 The staining results were classified semiquantitatively into 4 groups, as described previously,9,16 according to the number of positively stained tumor cells, as follows: score 0, negative; score 1, ⱕ 30% positive tumor cells; score 2, 30 –70% positive cells; score 3, ⱖ 70% positive tumor cells. Scores were evaluated regardless of the staining patterns observed in individual tumor cells.
Diagnosis of h. pylori-positive gastritis H. pylori-positive gastritis was diagnosed by combining the histomorphologic diagnosis (genta staining: hematoxylin and eosin, silver, Alcian blue) of gastritis with a commercial rapid-urease test according to the manufacturer’s instructions (HUT-test; Astra Zeneca, Wilmington, DE).
Bone marrow aspiration biopsies Bone marrow was obtained perioperatively from both iliac crests at a mean volume of 5 mL each into a heparinized syringe, and mononuclear cells were isolated by using Ficoll-Hypaque as described in our previous work.19
Immunocytochemical staining for cytokeratin 18-positive or cytokeratin 18/u-par positive cells in bone marrow Bone marrow cells (⫻ 106) from each aspirate were stained with the single alkaline phosphatase antialkaline phosphatase (APAAP) method for disseminated tumor cells by using cytokeratin 18 (CK18) as a marker, as described previously17–19 Staining was performed according to the established CK18-APAAP protocol, which also was described previously.17–19 If disseminated tumor cells were detected, then another 106 cells were subjected to double-staining for u-PAR/ CK18. Our immunogold/alkaline phosphatase method17–19 was performed as described previously using the same positive, negative, and cross-reaction controls. All slides were coded and evaluated by two investigators.
Statistical Analysis Chi-square analysis (Bonferroni-corrected) was performed to determine correlations between expected and detected frequencies. The parameters were coded as follows. Expression of u-PA, u-PAR, and PAI-1 was
u-PA System in Gastric Carcinoma/Beyer et al.
scored from 0 to 3; Lauren classification was categorized as intestinal versus diffuse/mixed; lymphangioinvasion/vessel infiltration was categorized as present versus absent; intestinal metaplasia was categorized as present versus absent; and pT, pN, and M status as well as UICC disease stage, tumor grade, and Borrmann classification were categorized as established previously. Pearson correlations between mean staining scores in primary tumors and biopsies were computed. Group-oriented life-table curves were calculated using the Kaplan–Meier method and were compared by using the Mantel–Cox log-rank test. To augment the prognostic impact of univariate factors, multivariate survival analysis was performed by using a Cox proportional hazards model that included the established risk factors for gastric carcinoma. Parameters for the multivariate analysis were coded the same as in the chi-square analysis with the addition of the dichotomized variables surgical intent (curative/ not curative) and surgical procedure (extended/not extended). Tumor localization was considered as cardia/fundus versus corpus/antrum. Parameters were entered into multivariate analysis after a significant univariate P value was determined. Statistical estimation and testing (2-sided significance level, P ⫽ 0.05) were performed using SPSS statistical software for Windows (version 12.0; SPSS Inc., Chicago, IL).
RESULTS Patient Results Two of 104 patients died in hospital, and 102 patients were followed every 6 months postoperatively for a median of 58 months (range, 7– 68 mos). Eighty-three of those patients had undergone resection with curative intent. Sixty-two patients died, including 61 deaths from malignant disease. Among the patients who underwent curative resection, 42 patients developed recurrences, including 9 peritoneal carcinoses, 20 locoregional recurrences, and 13 distant metastases (9 to the liver, 2 to bone, and 1 generalized metastasis). Immunohistochemistry in 102 perioperatively resected primary tumors showed that staining for u-PA, u-PAR, and PAI-1 was mostly membranous staining or cytoplasmic staining with a clear, membranous component, as observed previously.9 Nine of 102 tumors that were stained for u-PA, 12 of 102 tumors that were stained for u-PAR, and 10 of 102 tumors that were stained for PAI-1 were negative for those parameters in tumor cells (score 0). Twenty-nine u-PA-stained tumors, 46 u-PAR-stained tumors, and 36 PAI-1stained tumors had a score of 1; 41 u-PA-stained tumors, 36 u-PAR-stained tumors, and 36 PAI-1-stained tumors had a score of 2; and 23 u-PA-stained tumors,
1029
FIGURE 1. This photomicrograph shows a sample of intestinal metaplasia that had a staining score of 3 (ⱖ 70% positive tumor cells) for urokinase-type plasminogen activator receptor (original magnification, ⫻ 400).
8 u-PAR-stained tumors, and 20 PAI-1-stained tumors had a score of 3. Stromal cells/macrophages and few normal epithelial cells adjacent to the tumor tissue exhibited staining for u-PA system parameters. Similar to the findings from our first series,9 staining of normal and/or stromal cells in resected tumor tissue was not included in scoring. In 98 of 102 patients, we were able to perform an additional screening for intestinal metaplasia in corresponding normal tissues. Staining for u-PA, u-PAR, and PAI-1 was observed as cytoplasmic staining with an additional membranous component (Fig. 1). H. pylori gastritis was diagnosed in 36 of 102 patients. Disseminated (CK18-positive) tumor cells in bone marrow were noted in 64 of 102 patients, and 33 of those patients had additional evidence of u-PAR expression in tumor cells.
Correlation of u-PA System Staining in Resected Tumors with Established Tumor Characteristics, H. pylori Infection, Intestinal Metaplasia, and Disseminated Tumor Cells There was a significant, positive correlation (chisquare test) noted between PAI-1 staining (score, 0 –3) in resected tumor tissues from all 102 patients and pT classification (P ⫽ 0.034), and a significant association was established with UICC classification (P ⫽ 0.001) (Table 1). For u-PAR, we observed a significant, positive association with tumor grade (P ⫽ 0.022) and metastatic status (P ⫽ 0.020). There was a strong, mutual, statistical association between u-PA, u-PAR, and PAI-1 in primary tumors (P ⬍ 0.001 each).
1030
CANCER March 1, 2006 / Volume 106 / Number 5
TABLE 1 Correlations of Urokinase-Type Plasminogen Activator System Immunohistochemistry in Primary Tumors P value (chi-square) Established tumor characteristics
u-PA in tumor tissue
u-PAR in tumor tissue
PAI-1 in tumor tissue
pT pN M G R UICC stage Lauren classification Borrmann classification Presence of Helicobacter pylori Intestinal metaplasia CK18-positive cells, bone marrow u-PAR on CK18-positive cells, bone marrow
NS NS NS NS NS NS NS NS
NS NS 0.020 0.022 NS NS NS NS
0.034 NS NS NS NS 0.001 NS NS
0.238 0.013
0.014 0.006
⬍ 0.001 0.030
NS
NS
NS
0.009
0.014
NS
u-PA: urokinase-type plasminogen activator; u-PAR: urokinase-type plasminogen activator receptor; PAI–1: plasminogen activator inhibitor 1; pT: depth of invasion; NS: nonsignificant; pN: lymph node metastasis; M: distant metastasis; G: histopathologic grading; R: surgical curability; UICC: International Union Against Cancer.
There was a significant correlation noted between the presence of H. pylori infection and high u-PAR staining (P ⫽ 0.014; chi-square test) and PAI-1 staining (P ⬍ 0.001; chi-square test) in resected tumors (Table 1). High staining for u-PA (P ⫽ 0.013; chi-square test), u-PAR (P ⫽ 0.006; chi-square test), and PAI-1 (P ⫽ 0.030; chi-square test) in tumor tissue also was associated significantly with the detection of intestinal metaplasia (Table 1). No significant correlation was observed between u-PA system staining in resected tumors and the presence of disseminated tumor cells in bone marrow. However, a significant association between strong uPA staining (P ⫽ 0.009) and u-PAR staining (P ⫽ 0.014) was observed with the additional evidence of u-PAR expression on disseminated tumor cells in bone marrow (Table 1).
FIGURE 2. (A) This chart illustrates the results from a Kaplan–Meier analysis of
Analysis of the Univariate Prognostic Impact of u-PA System Immunohistochemistry in Resected Tumor Tissues
overall survival (all 102 patients) according to immunohistochemical staining for urokinase-type plasminogen activator (UPA). There was a significant difference between the groups (P ⫽ 0.002; Mantel–Cox log-rank test). Score 0: 9 patients, 0 events (mean survival time [MST], 49 mos; standard deviation [SD], 4 mos); score 1: 29 patients, 16 events (MST, 35 mos; SD, 9 mos); score 2: 42 patients, 26 events (MST, 22 mos; SD, 5 mos); score 3: 23 patients, 20 events (MST, 11 mos; SD, 2 mos). (B) This chart illustrates the results from a Kaplan–Meier analysis of overall survival (all 102 patients) according to immunohistochemical staining for plasminogen activator inhibitor 1 (PAI-1). There was a significant difference between the groups (P ⬍ 0.001; Mantel–Cox log-rank test). Score 0: 10 patients, 3 events (MST, 32 mos; SD, 12 mos); score 1: 36 patients, 18 events (MST, 35 mos; SD, 2 mos); score 2: 36 patients, 30 events (MST, 19 mos; SD, 2 mos); score 3: 20 patients, 17 events (MST, 13 mos; SD, 1 mo).
In Kaplan–Meier analysis (Mantel–Cox log-rank test), a significant association between increasing u-PA levels (score, 0 –3) and poorer disease-free survival was detected in 83 patients who underwent curative resection (P ⬍ 0.001). The overall survival of R0 patients (P ⬍ 0.001) and of all patients (P ⫽ 0.002) (Fig. 2A) was
reduced significantly in patients who had tumors with high u-PA staining. For resected tumor tissues that demonstrated high u-PAR staining, a trend was observed toward an association with shorter disease-free and overall sur-
u-PA System in Gastric Carcinoma/Beyer et al.
1031
TABLE 2 Multivariate Analysis of Recurrence-Free Survival in Patients who Underwent Curative Resection and Overall Survival in for All Patients Multivariate analysis RFS: R0 patients (n ⴝ 83)
OS: all patients (n ⴝ 102)
Characteristic
P value
HR
95% CI
P value
HR
95% CI
pT pN M R u-PA u-PAR PAI-1
NS
— — — — 1.301 — 1.1812
— — — — 0.900–1.882 — 1.257–2.612
NS NS NS 0.006 NS NS 0.003
— — — 1.249 1.192 — 1.619
— — — 0.853–1.798 0.903–1.571 — 1.184–2.213
— — 0.050 — 0.001
RFS: recurrence-free survival; R0: curative resection; OS: overall survival; 95% CI: 95% confidence interval; HR: hazards ratio; pT: depth of invasion; NS: nonsignificant; pN: lymph node metastasis; M: distant metastasis; R: surgical curability; u-PA: urokinase-type plasminogen activator; u-PAR: urokinase-type plasminogen activator receptor; PAI-1: plasminogen activator inhibitor 1.
vival (P ⬎ 0.05). For PAI-1 staining, a highly significant association was observed for disease-free survival (P ⬍ 0.001) and overall survival (P ⬍ 0.001) among R0 patients and for overall survival in all 102 patients (P ⬍ 0.001) (Fig. 2B).
Analysis of the Multivariate Prognostic Impact of u-PA System Immunohistochemistry in Resected Tumor Tissues In multivariate (Cox proportional hazards) analysis, the established risk factors in gastric carcinoma (pT status, pN status, M status, tumor grade, surgical curability, necessity of extended resections, Lauren classification, Borrmann classification, and UICC stage) and u-PA system parameters (u-PA, u-PAR, and PAI-1) were considered if they were found to be significant ion the univariate analyses. For disease-free survival in R0 patients (n ⫽ 83 patients), u-PA and PAI-1 were the dominant independent parameters (u-PA: P ⫽ 0.050; hazard ratio, 1.301[ 95% confidence interval (95% CI)], 0.900 –1.882]; PAI-1: P ⫽ 0.001; hazard ratio, 1.812 [95% CI, 1.257–2.612]) (Table 2). For overall survival in R0 patients (n ⫽ 83) and for overall survival of all patients. including surgical curability (R) as the prominent risk factor, PAI-1 was confirmed as an independent predictor of poor prognosis (P ⫽ 0.003; hazard ratio, 1.62 [95%CI, 1.18 –2.21]) (Table 2).
Correlation between Immunohistochemistry for u-PA System Parameters in Preoperative Biopsy Specimens and Resected Tumor Tissues In a subset of 12 patients, a mean of 5.5 preoperative biopsies (range, 4 –7 biopsies) per patient had been taken during routine gastroduodenoscopy and were stained and scored for u-PA, u-PAR and PAI-1 according to the methods described above. In those 12 patients
and in a mean of 5.5 biopsies per patient, a highly significant (Pearson) correlation between mean staining scores in biopsies and in primary tumor tissues was observed for u-PA (P ⫽ 0.019), u-PAR (P ⫽ 0.043), and PAI-1 (P ⫽ 0.022). These results indicate that comparable results are obtained when using preoperative biopsy material instead of intraoperatively resected tumor tissue.
Intestinal Metaplasias in Intraoperatively Resected Normal Tissues In 98 of 102 patients who underwent tumor resection, corresponding resected normal tissue could be evaluated for the presence of additional intestinal metaplasias. In 32 of those 98 patients, metaplasias were observed in normal gastric tissue. In all 32 patients, an immunohistochemical analysis for u-PA system parameters in intestinal metaplasias could be performed (Fig. 1).
Correlation of Intestinal Metaplasias with Established Tumor Characteristics The presence of intestinal metaplasia was found to be correlated significantly with an adenocarcinoma phenotype according to histopathology (P ⫽ 0.003; chisquare) and according to the Lauren intestinal type (P ⬍ 0.001; chi-square test). We observed a significant association between metaplasia detection with advanced pT status (P ⫽ 0.026, chi-square) and reduced surgical curability (patients who could not undergo curative resection; P ⫽ 0.046; chi-square test).
Correlation of Intestinal Metaplasia with u-PA System Parameters in Resected Tumor Tissue, H. pylori Gastritis, and Disseminated Tumor Cells in Bone Marrow The presence of intestinal metaplasia was associated significantly with a high staining score for u-PA (P
1032
CANCER March 1, 2006 / Volume 106 / Number 5
⫽ 0.013; chi-square test), u-PAR (P ⫽ 0.006; chi-square test), and PAI-1 (P ⫽ 0.030; chi-square test) in corresponding resected primary tumor tissue. In addition, there was a significant association between intestinal metaplasia and the detection of H. pylori infection (P ⬍ 0.001; chi-square test). There was no association noted between metaplasia and disseminated tumor cells in bone marrow or u-PAR expression in disseminated tumor cells.
The Prognostic Relevance of the Detection of Intestinal Metaplasia or H. pylori Infection A Kaplan–Meier analysis (Mantel–Cox log-rank test) showed that the presence of neither intestinal metaplasia nor H. pylori-positive gastritis was associated significantly with overall or recurrence-free survival (P ⬎ 0.05).
The Correlative and Prognostic Relevance of u-PA System Immunohistochemistry in Intestinal Metaplasia We also investigated whether the u-PA system parameters that were analyzed immunohistochemically in intestinal metaplastic lesions were correlated with each other, with established tumor characteristics, with u-PA system immunohistochemistry in corresponding primary tumors, with H. pylori status, or with disseminated tumor cell status in bone marrow. u-PA staining in metaplasias (in the 32 patients who had metaplasia) exhibited a significant, positive correlation with strong staining for u-PAR (P ⫽ 0.034; chi-square test) but not with PAI-1 staining (P ⬎ 0.05) in metaplasias. In addition, high u-PAR staining was found to be correlated significantly with high PAI-1 staining in metaplasias (P ⫽ 0.048; chi-square test). There were significant associations between high u-PAR scores in metaplasias with advanced pN status (P ⫽ 0.030; chi-square test) and pT status (P ⫽ 0.046; chi-square test). For PAI-1 in metaplasia, we found a significant, positive association with pT status (P ⫽ 0.025, chi-square test) (Table 3). Staining for u-PAR in intestinal metaplasia was correlated significantly with evidence of H. pylori (Table 3). No significant correlations were observed for u-PA staining in metaplasias. Finally, a prognostic analysis was performed for u-PA system immunohistochemistry in intestinal metaplasias (Kaplan–Meier analysis; Mantel–Cox logrank test; n ⫽ 32 patients). There was no significant association noted between immunohistochemistry for u-PA, u-PAR, or PAI-1 in metaplasias and disease-free or overall survival in the 32 patients investigated.
TABLE 3 Overview of Correlations of Urokinase-Type Plasminogen Activator System Immunohistochemistry in Intestinal Metaplasias P value (chi-square)
Characteristic u-PA in metaplasias u-PAR in metaplasias PAI-1 in metaplasias pT pN M G Lauren classification Borrmann classification UICC stage Helicobacter pylori Presence of disseminated tumor cells u-PAR on disseminated tumor cells
u-PA in metaplasias
u-PAR in metaplasias
PAI-1 in metaplasias
— 0.034 NS NS NS NS NS NS NS NS NS
0.034 — 0.048 0.046 0.030 NS NS NS NS NS 0.020
NS 0.048 — 0.025 NS NS NS NS NS NS NS
NS
NS
NS
NS
NS
NS
u-PA: urokinase-type plasminogen activator; u-PAR: urokinase-type plasminogen activator receptor; PAI-1: plasminogen activator inhibitor 1; NS: nonsignificant; pT: depth of invasion; pN: lymph node metastasis; M: distant metastasis; G: histopathologic grading; UICC: International Union Against Cancer.
DISCUSSION In the current study, the prognostic impact of u-PA and PAI-1 in gastric carcinoma was demonstrated in a patient series independent of our previously reported9 series. These results suggest that a biologic prediction of risk through the u-PA system is possible in preoperative biopsy specimens, because there is a significant correlation between staining results in biopsies and tumors that are resected later in the same patients. The current study also demonstrated a significant association between u-PA system expression in gastric carcinomas and H. pylori infection, and we suggest a possible role of this bacterium in tumorassociated proteolysis. Finally, to our knowledge, the current study is the first to demonstrate the up-regulation of all three u-PA system parameters in intestinal metaplasias from patients with gastric carcinoma and a significant correlation between u-PA system expression in metaplasias and advanced pN or pT status. These findings suggest a biologic hypotheses concerning carcinogenesis processes, particularly an association between u-PA system expression in metaplasias and disease progression. The prognostic impact of the u-PA system has been reported in diverse malignacies10 –14 and in gastric carcinoma by both colleagues (Nekarda et al.)15 and by own group.9 However, to our knowledge, few groups to date have attempted to validate their own
u-PA System in Gastric Carcinoma/Beyer et al.
results in an independent prospective series of patients. Therefore, we believe the results of our study confirm an association between u-PA and PAI-1 in a second, independent, and prospective series of patients with gastric carcinoma and an estimated impact that was exceeded only by surgical curability. The functional relevance of u-PA system expression for metastatic spread is reflected in a significant correlation between u-PA, u-PAR, and PAI-1 in the primary tumor and the detection of u-PAR on disseminated tumor cells in bone marrow from the same patients, indicating that tumors with high u-PA system expression are associated with the spread of invasive cell phenotypes. Surprisingly, u-PAR demonstrated a trend toward an association with poor clinical prognosis, although that trend was not found to be significant in the current series. This finding is in line with results from other reports in this15 and other tumor entities, such as breast carcinoma, in which u-PA and PAI-1, rather than u-PAR, most often were the strongest prognostic factors.10,11,23 This phenomenon of less prognostic significance for u-PAR than for u-PA or PAI-1 may be because of the different molecular forms of u-PAR (such as cleaved variants) that appear on tissues or because of additional biologic functions of u-PAR, for example, providing a switch between cell proliferation and dormancy.5,24 However, in our previous study and in the current study, because a strong correlation between u-PAR and both u-PA and PAI-1 in primary tumor tissues was observed, there is a clear indication that all 3 factors are up-regulated in parallel in gastric carcinoma, corroborating the functional importance of an interaction of all 3. Nevertheless, the prognostic impact of the u-PA system, independent of established risk factors such as pT or pN status, implicates this system strongly as a marker for identifying gastric carcinoma in patients who present with especially aggressive tumors, regardless of their clinical tumor stage. This biologic information would be extremely useful for preoperative endoscopic diagnosis. To our knowledge to date, not many studies have investigated the u-PA system in preoperative biopsies. De Bruin et al.25 were among the first to study endoscopic biopsies in gastric carcinoma for the expression of u-PA and tissue-type plasminogen activator (t-PA), but not for u-PAR or PAI-1. Similarly, Sier et al.7 measured u-PA and t-PA in biopsy specimens, which demonstrated higher expression levels in carcinomas than in normal mucosa; however, those authors did not compare their results with results from tissues that were resected later. In the current study, we observed a significant association between biopsy staining scores for u-PA, u-PAR,
1033
and PAI-1 and scores observed in corresponding resected tumors. This means that it is possible to define patients who are at risk for metastasis and recurrence prior to tumor surgery. The clinical consequences may be drawn especially for the application of neoadjuvant (radio-)chemotherapy in such patients, particularly because there are data suggesting that u-PA and PAI-1 are markers that predict response to chemotherapy.23 It is known that H. pylori is associated with chronic atrophic gastritis and intestinal metaplasia,26 a correlation that also was observed statistically in the current study. In contrast to mucosa-associated lymphoid tissue lymphoma,26 data suggesting a causative or prognostic role for H. pylori in gastric carcinoma still are less clear.26 This also was observed in the current study, in which H. pylori infection was not prognostic. A possible explanation may be that H. pylori has a functional role in carcinogenesis but is less important when the tumor already is established.26 A most noteworthy observation in the current study was a correlation between H. pylori infection and high staining scores for the u-PA system (u-PAR and PAI-1) in primary tumors. In addition, a significant correlation was observed between high u-PAR expression in intestinal metaplasias and H. pylori infection. This finding is in line with other studies, which found higher levels of u-PA, PAI-1, and cathepsins in H. pylori-positive patients with duodenal ulcers27 and elevated u-PA levels and activity in patients with H. pylori-associated pangastritis.28 Some studies have suggested that binding of plasminogen to the bacterial wall supports u-PA-mediated plasminogen activation29 as a potential mechanism employed by H. pylori to induce u-PA-mediated proteolysis. Thus, H. pylori may be able to induce gastritis or ulcers in part through the induction of u-PA-associated parameters. Our current results indicated an association between H. pylori and elevated u-PA system expression in gastric carcinoma. It is interesting to speculate that H. pylori may have an indirect biologic role in carcinogenesis or even in the progression of gastric carcinoma by inducing u-PA-associated proteolysis. The presence of intestinal metaplasia was not indicative of a poor prognosis in the current study. One explanation for this finding may be that we did not differentiate metaplasias into grades of dysplasia and severity similar to what is done, for example, in the Sydney classification.30 It has been speculated that Grade 3 dysplastic metaplasias may be associated with an increased risk of developing gastric carcinoma30; however, other studies concluded that grading of metaplasias was not helpful in differentiating the risk of gastric carcinoma.31
1034
CANCER March 1, 2006 / Volume 106 / Number 5
Although we did not observe an association between intestinal metaplasia and prognosis, it is worth noting that there was a significant association between prognosis and metaplasias with high levels of u-PA, u-PAR, and PAI-1 staining in primary tumors. However, the most interesting were our results for u-PA system staining in intestinal metaplasias themselves. First, there was a significant correlation between u-PA and u-PAR and between u-PAR and PAI-1 in metaplasias, implicating a concerted up-regulation of u-PA system parameters not only in primary carcinomas but also in associated metaplasias. Second, there was a significant correlation between u-PA staining in metaplasias and u-PA staining in corresponding primary tumors. This implies that, similar to what has been observed previously in endoscopic biopsies, uPA system staining in premalignant metaplasias may be predictive of u-PA system expression in the corresponding tumor. A more speculative conclusion would be that the metaplasia and the corresponding tumor may have evolved through common molecular pathways, leading to overexpression of the u-PA system, or that the tumor even may have evolved from the metaplasia, because similar molecular changes, at least regarding expression of the u-PA system, are noted in both. Third, although there was no significant prognostic impact of u-PA system staining in metaplasias (possibly attributable to the small patient numbers; n ⫽ 32 patients), a significant association was observed between u-PAR and PAI-1 staining in intestinal metaplasias and advanced pT and/or pN status. These data suggest that u-PA system expression in associated metaplasias may predict a more progressive tumor phenotype. It is interesting to speculate that, in a patient who has not yet developed gastric carcinoma, the development of an aggressive tumor type in the future may be predicted by assessing the u-PA system in an intestinal metaplastic lesion. Taken together, the current results demonstrate the prognostic impact of u-PA and PAI-1 in gastric carcinoma in an independent patient series and the association between u-PA system expression and H. pylori infection. The possibility of predicting aggressive tumor biology in preoperative biopsy specimens and through u-PA system expression in associated intestinal metaplasias implies the clinical consequences of using the u-PA system as a preoperative indicator for neoadjuvant chemotherapy and as a predictor of tumor biology in associated premalignant lesions.
REFERENCES 1.
Blasi F. Urokinase and urokinase receptor: a paracrine/ autocrine system regulating cell migration and invasiveness. Bioessays. 1993;15:105–111.
2.
3. 4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 15.
16.
17.
18.
Kim J, Yu W, Kovalski K, Ossowski L. Requirement for specific proteases in cancer cell intravasation as revealed by a novel semiquantitative PCR-based assay. Cell. 1998;94:353– 362. Ellis V, Behrendt N, Dano K. Plasminogen activation by receptor-bound urokinase. J Biol Chem. 1991;266:12752–12758. Conese M, Olson D, Blasi FAD. Protease nexin-1-urokinase complexes are internalized through a mechanism that requires both urokinase-receptor and ␣2-macroglobulin-receptor. J Biol Chem. 1994;269:17886 –17892. Bohuslav J, Horejsi V, Hansmann C, et al. Urokinase-plasminogen activator-receptor, 2-integrins, and Src-kinases within a single receptor complex of human monocytes. J Exp Med. 1995;181:1381–1390. Liu G, Shuman MA, Cohen RL. Co-expression of urokinase, urokinase receptor and PAI-1 is necessary for optimum invasiveness of cultured lung cancer cells. Int J Cancer. 1995; 60:501–506. Sier CFM, Verspaget HW, Griffioen G, Ganesh S, Vloedgraven JH, Lamers CB. Plasminogen activators in normal tissue and carcinomas of the human oesophagus and stomach. Gut. 1993; 34:80 – 85. Pyke C, Kristensen P, Ralfkiaer E, Eriksen J, Dano K. The plasminogen activation system in human colon cancer: messenger RNA for the inhibitor PAI-1 is located in endothelial cells in the tumor stroma. Cancer Res. 1991;51:4067– 4071. Heiss MM, Babic R, Allgayer H, et al. Tumor-associated proteolysis and prognosis: new functional risk factors in gastric cancer defined by the urokinase-type plasminogen activator system. J Clin Oncol. 1995;13:2084 –2093. Duffy MJ, Reilly D, O’Sullivan C, O’Higgins N, Fennelly JJ, Andreasen P. Urokinase plasminogen activator, a new and independent prognostic marker in breast cancer. Cancer Res. 1990;50:6827– 6829. Jaenicke F, Schmitt M, Pache L, et al. Urokinase (uPA) and its inhibitor PAI-1 are strong and independent prognostic factors in node negative breast cancer. Breast Cancer Res Treat. 1993;24:195–208. Pedersen H, Grondahl-Hansen J, Francis D, et al. Urokinase and plasminogen activator inhibitor type 1 in pulmonary adenocarcinoma. Cancer Res. 1994;54:120 –123. Mulcahy HE, Duffy MJ, Gibbons D, et al. Urokinase-type plasminogen activator and outcome in Dukes’ B colorectal cancer. Lancet. 1994;344:583–584. Ganesh S, Sier CFM, Heerding MM, et al. Urokinase receptor and colorectal cancer survival. Lancet. 1994;344:401– 402. Nekarda H, Schmitt M, Ulm K, et al. Prognostic impact of urokinase-type plasminogen activator and its inhibitor PAI-1 in completely resected gastric cancer. Cancer Res. 1994;54:2900 –2907. Heiss MM, Allgayer H, Gruetzner KU, et al. Clinical value of an extended biological staging by bone marrow micrometastases and tumor-associated proteases in gastric cancer. Ann Surg.1997;226:736 –745. Allgayer H, Heiss MM, Riesenberg R, Babic R, Jauch KW, Schildberg FW. Urokinase plasminogen activator receptor (uPA-R)—a potential characteristic of metastatic phenotypes in minimal residual tumor disease. Cancer Res. 1997; 57:1394 –1399. Heiss MM, Allgayer H, Gruetzner KU, et al. Individual development and uPA-receptor-expression of disseminated tumour cells in bone marrow: a reference to early systemic disease in solid cancer. Nat Med. 1995;1:1035–1039.
u-PA System in Gastric Carcinoma/Beyer et al. 19. Heiss MM, Simon EH, Beyer BCM, et al. Minimal residual disease in gastric cancer: first evidence of an independent prognostic relevance of urokinase receptor gene expression by disseminated tumor cells in a large series of patients. J Clin Oncol. 2002;20:2005–2016. 20. Hermanek P, Maruyama K, Sobin LH. Stomach carcinoma. In Hermanek P, Gospodarowicz MK, Henson DE, editors. Prognostic factors in cancer. Berlin: Springer, 1995:47– 64. 21. Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histo-clinical classification. Acta Pathol Microbiol Scand. 1965;64:31– 49. 22. Borrmann R. Geschwuelste des Magens. In: Henke FU, Lubarsch O, editors. Handbuch der speziellen pathologischen anatomie und histologie, vol VI/1. Berlin: Springer, 1926: 161–173. 23. Harbeck N, Kates R, Look MP, et al. Enhanced benefit from adjuvant chemotherapy in breast cancer patients classified high-risk according to uPA and PAI-1 (n ⫽ 3424). Cancer Res. 2002;62:4617–1622. 24. Aguirre-Ghiso JA, Liu D, Mignatti A, Kovalski K, Ossowski L. Urokinase receptor and fibronectin regulate the ERKMAPK to p38MAPK activity ratios that determine carcinoma cell proliferation or dormancy in vivo. Mol Biol Cell. 2001;12:863– 879.
1035
25. De Bruin PA, Verspaget HW, Griffioen G, Verheijen JH, Dooijewaar Lamers CB. Plasminogen activators in endoscopic biopsies as indicators of gastrointestinal cancer: comparison with resection specimens. Br J Cancer. 1989;60:397– 400. 26. Genta RM. The gastritis connection: prevention and early detection of gastric neoplasms. J Clin Gastroenterol. 2003; 36:S44 –S49. 27. Herszenyi L, Plebani M, Carraro P, et al. Impaired fibrinolysis and increased protease levels in gastric duodenal mucosa of patients with active duodenal ulcer. Am J Gastroenterol. 1997;92:843– 847. 28. Gotz JM, Vergouwe Y, Verspaget HW, et al. Gastric mucosal plasminogen activators in Helicobacter pylori infection. Dig Dis Sci. 1996;41:1577–1582. 29. Ljungh A. Helicobacter pylori interactions with plasminogen. Methods. 2000;21:151–157. 30. Leung WK, Lin SR, Ching JY, et al. Factors predicting progression of gastric intestinal metaplasia: results of a randomised trial on Helicobacter pylori eradication. Gut. 2004; 53:1217–1219. 31. Conchillo JM, Houben G, de Bruine A, Stockbrugger R. Is type III intestinal metaplasia an obligatory precancerous lesion in intestinal-type gastric carcinoma? Eur J Cancer Prev. 2001;10:307–312.
