ANTICANCER RESEARCH 27: 3029-3034 (2007)
Clinical Significance of nm23 Gene Expression in Gastric Cancer STEFAN P. MÖNIG1, BRIT NOLDEN2, THOMAS LÜBKE1, ALEXANDRA POHL1, GUIDO GRASS3, PAUL M. SCHNEIDER1, HANS P. DIENES2, ARNULF H. HÖLSCHER1 and STEPHAN E. BALDUS4 1Department
of Visceral and Vascular Surgery, 2Institute of Pathology and Center for Clinical Trials, University of Cologne, Cologne; 4Institute of Pathology, University of Düsseldorf, Düsseldorf, Germany
3Coordinating
Abstract. Background: The expression of the nm23 gene has been associated with the development of metastasis. Numerous studies have shown down-regulation of nm23 expression in metastatic breast and colon cancer. The expression of the putative metastasis-suppressor gene nm23 in gastric carcinoma is controversial. The aim of this study was the analysis of nm23 expression in a large series of gastric cancer patients. Patients and Methods: In a retrospective immunohistochemical study specimens obtained from 116 gastric cancer patients (mean age 64 years; range: 33-85) who had undergone gastrectomy with extended lymphadenectomy were analyzed. Nm23 expression in the tumor epithelium was studied by immunohistochemistry followed by a semi-quantitative (score 0-3) evaluation. Statistical analysis including Chi-square test, uni- and multivariate survival analyses were performed. Results: The nm23 staining pattern was positive (score 2-3) in 100 (86.2%) specimens and negative (score 0-1) in 16 (13.8%) samples. Lymph node metastasis was found in 65% of the patients. No significant correlations could be determined between nm23 expression and other variables such as gender, age, tumor differentiation, WHO-, Laurén-, Goseki-, or Ming-classification. The intensity of nm23 staining in the tumor cells was not significantly correlated with depth of tumor infiltration (T-stage), lymph node metastasis (N-stage), distant metastasis (M-stage), UICC-stage, or prognosis. Conclusion: Our series did not show a correlation of nm23 expression in terms of lymph node and distant metastasis or prognosis in gastric cancer patients. The preoperative evaluation of lymph node status is one of the key factors in the pretherapeutic staging and therapeutic
Correspondence to: Stefan P. Mönig, MD, Department of Visceral and Vascular Surgery, University of Cologne, Kerpener Str. 62, 50937 Cologne, Germany. Tel: +49 221 478 4803, Fax: +49 221 478 6258, e-mail:
[email protected] Key Words: nm23, immunohistochemistry, lymph node metastasis, gastric cancer.
0250-7005/2007 $2.00+.40
decision making in gastric cancer patients (1). In early gastric cancer the presence or absence of lymph-node metastases is a critical determinant of whether less invasive treatment, such as endoscopic mucosal resection, can be performed (2, 3). Even in advanced carcinoma nodal status is of high interest, since extensive lymphadenectomy may increase postoperative morbidity and mortality (4). It has been shown that determination of lymph node size by imaging techniques is not a reliable indicator for detecting metastasis (5, 6). To date no preoperative staging parameter is able to predict lymph node metastasis in gastric cancer. Metastasis is a cascade of linked sequential steps involving multiple host-tumor interactions (7). The mechanism by which malignant tumors give rise to distant metastases is still largely unexplained because of its complexity. The nm23 gene, located on chromosome 17q, region 21.3-22, has been evaluated as a putative metastasis suppressor gene by differential hybridization between low and high metastatic murine melanoma cell lines in which reduced nm23 mRNA levels were correlated with increased metastatic potential (8, 9). Two nm23 genes, nm23-H1 (NME1) and nm23-H2 (NME2), encoding distinct nm23/NDP (nucleoside diphosphate) kinase proteins with molecular weights of 18.5kDa and 17-kDa, respectively, have been reported on chromosome 17 (10, 11). Both proteins exhibit 88% amino acid sequence identity (11) and are involved in cell proliferation (12), differentiation (13), and motility (14), as well as basement membrane formation and growth arrest (15). In various malignant tumors, reduced expression of nm23 has been utilized as an indicator closely related to metastatic potential (16). However contradictory reports concerning nm23 expression have also been found in some tumors. Thus the role of nm23 expression has not yet been clarified, in spite of extensive studies. Furthermore, only a few studies in a small number of patients of the clinical and clinicopathological roles of nm23 expression have been reported in gastric cancer. Therefore, the expression of nm23 in gastric carcinoma was investigated to determine its importance as a predictor of disease progression and prognosis.
3029
ANTICANCER RESEARCH 27: 3029-3034 (2007) Patients and Methods Patients. Specimens from 116 primary gastric adenocarcinomas collected between May 1996 and August 2000 were obtained from the files of the Institute of Pathology, University of Cologne, Germany. The mean age of the patients was 64 years (range: 33-85), 69 patients were male and 47 were female (ratio 1.47:1). One hundred and four patients (89.7%) had undergone total gastrectomy and 12 patients (10.3%) subtotal gastrectomy with extended lymphadenectomy (compartment I and II). An average of 39 lymph nodes had been resected and were analyzed for each patient. Neoadjuvant chemotherapy with cisplatin, 5-FU, and leucovorin (PLF) had been applied in 15 cases (12.9%) with locally advanced tumor. None of the patients had received adjuvant (postoperative) chemotherapy. In cases of tumor recurrence palliative chemotherapy with 5-FU, leucovorin, and etoposide (ELF) had been administered. The mean follow-up was 42 months (SE±3). All the specimens were categorized according to tumor differentiation, UICC-, WHO-, Laurén-, Goseki-, and Ming-classification. Immunohistochemical analysis. The tumor samples had been routinely fixed in 5% phospate-buffered formalin and embedded in paraffin. Tissue specimens (sections 5mm thick) were deparaffinized according to standard histological technique. Endogenous peroxidase activity was blocked by 3% H2O2/methanol for 30 min. at room temperature (RT). Non-specific binding sites were blocked by normal swine serum X 901 (Dako, Copenhagen, Denmark), diluted 1:20 (v/v) in Tris-buffered saline pH 7.2 (TBS) for 30 min. at RT. The primary monoclonal anti-human nm23-H1 antibody (Novocastra, Newcastle-upon-tyne, UK, clone 37.6) was applied, which is specific for nm-23-H1 protein, but exhibits a low crossreactivity with the nm23-H2 homologue according to the manufacturer’s information. It was incubated overnight at 4ÆC (dilution in TBS 1:100 v/v containing 2.5% bovine serum albumin). Between each of the following steps, the specimens were washed three times in TBS. After the incubation with the primary antibody biotinylated rabbit anti-mouse immunoglobulin E413 (Dako), diluted 1:400 (v/v) in TBS/2.5% bovine serum albumin (BSA), was added for 30 min. at RT. All sections were incubated for 30 min at RT with streptavidin-peroxidase conjugate P397 (Dako), diluted 1:400 (v/v) in TBS/2.5% BSA. The reaction was visualized by 200 mg/ml 3amino-9-ethyl-carbazol (Sigma, Munich, Germany) in 50 mM sodium acetate buffer containing 5% dimethyl-formamide and 0.01% H2O2 for 30 min at RT. The slides were counterstained with hematoxylin for 2 min and mounted in glycerol jelly. Non-neoplastic tumor-associated gastric mucosa was used as an internal positive control. As the negative controls the primary antibody was substituted by normal mouse serum. Semiquantitative analysis. All the slides were evaluated without knowledge of patient and clinical status by two independent pathologists. Scoring was exclusively restricted to tumor cell staining, and stromal staining was not considered. We evaluated at least 50 high-power fields of the tumor of each patient. The degree of expression of nm23 was estimated by semi-quantitative evaluation into four groups according to the number of positively-stained tumor cells: score 0=negative; score 1=up to 30% positive tumor cells; score 2=30-60% positive cells; score 3=equal or more than 60% positive tumor cells. Samples in which 30% or more of cancer cells were stained (score 2+3) were counted as positive, and samples in
3030
Table I. nm23 tissue status and clinico-pathological parameters of 116 patients. Total Gender Men Women Borrmann Early cancer I II III IV WHO papilary tubular signet-ring cell mucinous unclassified Differentiation Well Moderate Poor Laurén Intestinal Diffuse Mixed Goseki I II III IV Ming Expanding Infiltrative Depth of invasion T1 T2 T3 T4 Nodal status N0 N1 N2 N3 Metastasis M0 M+ UICC Ia Ib II IIIa IIIb IV
nm23+
% nm23+
p
69 47
61 39
88.4 83.0
0.367
23 9 22 25 37
18 9 19 23 31
78.3 100.0 86.4 92.0 83.8
0.470
5 68 37 3 3
5 59 30 3 3
100.0 86.8 81.1 100.0 100.0
0.628
1 33 82
1 29 70
100.0 87.9 85.4
0.550
45 56 15
40 48 12
88.9 85.7 80.0
0.681
53 9 20 34
47 8 17 28
88.7 88.9 85.0 81.4
0.855
45 71
40 60
88.9 84.5
0.505
23 38 45 10
18 35 39 8
78.3 92.1 86.7 80.0
0.446
41 35 18 22
33 32 14 21
80.5 91.4 77.8 95.5
0.205
97 19
86 14
88.7 73.7
0.083
19 18 21 14 8 36
14 17 18 14 8 29
73.7 94.4 85.7 100.0 100.0 80.6
0.155
which less than 30% of tumor cells (score 0+1) were counted as negative, as suggested by Yoo et al. (17). Statistical analysis. Explorative, univariate data analysis was performed (SPSS Inc., Chicago, IL, USA). In order to evaluate correlations
Mönig et al: nm23 Expression in Gastric Cancer
Figure 1. Positive immunoreaction for nm23 in tubular gastric adenocarcinoma (original magnification x250).
between the staining results and clinico-pathological variables, the Chi-square test was applied at a significance level of 5%. Univariate survival analysis was performed according to the Kaplan-Meier approach. To analyze the predictive value of nm23 compared to other known predictors, Cox’s regression analysis was performed. The following variables were included in the conditional forward model: UICC-stage, sex, grading, WHO-type, residual tumor (R) classification, and nm23-status.
Results The classification and staging results of all 116 tumor specimens are summarized in Table I. According to UICC 19 (16.4%) patients were classified as stage Ia, 18 (15.5%) as stage Ib, 21 (18.1%) as stage II, 14 (12.1%) as stage IIIa, 8 (6.9%) as stage IIIb and 36 (31%) as stage IV. Forty one specimens (35.3%) were subtyped as pN0 and 75 (64.7%) as pN+ (showing lymph node metastasis), of the latter, 35 cases (30.2%) were classified as pN1, 18 cases (15.5%) as pN2, and 22 cases (19%) as pN3. Distant metastases (pM+) were seen in 19 patients (16.4%). Immunohistochemistry for nm23 revealed strong cytoplasmic staining in the tumor cells (Figure 1) that was generally more intense than the nm23 expression in normal gastric mucosa adjacent to the tumor in almost all specimens. No difference in expression was detcted between the tumor center and the invasion front.
Overall the nm23 staining pattern was scored positive (score 2-3) in 100 (86.2%) specimens and negative (score 01) in only 16 (13.8%) samples. Of 100 cases with nm23 positive immunoreactivity, 43 cases exhibited moderate nm23 expression (score 2) and 57 cases strong expression (score 3). Table I summarizes nm23 expression in relation to different histological subtypes and different stages of gastric cancer. The intensity of nm23 staining in tumor epithelia did not correlate significantly with depth of tumor infiltration (Tstage), lymph node metastasis (N-stage), distant metastasis (M-stage) or UICC-stage. Concerning tumor extension, advanced pT2-pT4 tumors were more often nm23 positive (88.2%) than early pT1 gastric carcinomas (78.3%). Additionally no significant correlation was seen between nm23 expression and other variables such as gender, Borrmannclassification, tumor differentiation, WHO-, Laurén-, Goseki-, or Ming-classification (Table I). There was no significant difference in median survival time between patients with nm23-positive (31 months; 95%-CI:1745) and negative tumors (36 months: 95%-CI: 13-59) (p=0.7179) (Figure 2). Overall cumulative survival at 5 years was 40% in nm23 negative tumors and 35.8% in nm23 positive tumors (p=0.718). In multivariate survival analysis nm23 was not an independent prognostic factor (p=0.411), but a high predictive value of the pTNM staging system and the Rcategory could be confirmed (p