Significant correlation between expression of interleukin-1α and liver metastasis in gastric carcinoma

Authors


Abstract

BACKGROUND

Interleukin (IL)-1 has been reported to augment the hematogeneous metastasis of some cancers by inducing the expression of adhesion molecules on vascular endothelial cells and also increasing the expression of proteases from tumor cells in vitro. The purpose of this study was to determine the clinical significance of the IL-1α expression in primary tumors of gastric carcinoma.

METHODS

Paraffin embedded sections of the tumors obtained from 109 patients who underwent a gastrectomy for gastric carcinoma with subserosal invasion were stained for IL-1α by using the streptavidin biotin method. Staining was considered positive when 10% or more of the malignant cells displayed cytoplasmic staining.

RESULTS

Sixty (55.0%) tumors expressed IL-1α. Positive staining for IL-1α was more likely in patients with differentiated tumors than in those with undifferentiated tumors. The expression of IL-1α showed a significant correlation with liver metastasis. Of the 84 patients receiving a curative resection, those with tumors expressing IL-1α had a significantly higher incidence of liver recurrence than those without. A logistic regression analysis revealed positive staining for IL-1α to be the best predictive factor of patients who develop liver recurrence.

CONCLUSIONS

The authors' results suggest that the immunohistochemical expression of IL-1α is a useful predictor of liver recurrence in patients undergoing a curative resection for gastric carcinoma with subserosal invasion. Cancer 2001;91:1272–6. © 2001 American Cancer Society.

Interleukin (IL)-1α, one of the important inflammatory cytokines, has been reported to be produced by cancer cell lines derived from carcinomas of the pancreas, lung, ovary, and stomach.1–4 Using a melanoma model, Anasagasti et al. reported that host- and/or melanoma-derived IL-1 promoted metastasis in such organs as the bone marrow, spleen, liver, lung, and adrenal gland.5 Takeda et al. found two lymphoma cell lines producing IL-1α, which metastasize spontaneously to the liver.6 Moreover, IL-1α also has been reported to induce the expression of adhesion molecules on vascular endothelial cells7, 8 and proteases on cancer cells in vitro.9, 10 Although these reports suggest that IL-1α produced by carcinoma cells can augment metastasis, so far few studies have been made on the clinical significance of the IL-1α expression in carcinoma cells of primary tumors.

We previously reported that higher concentrations of IL-1α in primary tumor tissue significantly correlated with a higher incidence of liver metastasis in patients with gastric carcinoma.11 In the current study, we investigated the relation between the immunohistologic expression of IL-1α in the primary tumor and liver metastasis in gastric carcinoma.

PATIENTS AND METHODS

Patients

Among the 681 patients who underwent a gastrectomy for gastric carcinoma at the First Department of Surgery, National Defense Medical College Hospital between 1983 and 1989, 109 consecutive patients with tumors invading the subserosa but not the serosa were enrolled in this study. Liver metastasis was found in 9 of the 109 patients at operation. A curative resection with a regional lymphadenectomy was performed in 84 patients. The recurrence after a curative gastrectomy was assessed by computed tomography, ultrasonography, and endoscopy once a year for more than 5 years after surgery. The median follow-up period was 125 months (range, 76–162 months).

Immunohistochemistry

Paraffin embedded sections were cut 4 μm thick, deparaffinized in xylene and ethanol, and then heated in a plastic jar filled with phosphate-buffered saline in an autoclave at 121 °C for 8 minutes. After being cooled at room temperature, the sections then were immersed in 3% hydrogen peroxide and normal goat serum to inhibit endogenous peroxidase and nonspecific binding. Immunohistochemical studies were performed using the streptavidin-biotin-peroxidase complex method (Histofine SAB-PO(R) kit; Nichirei, Tokyo, Japan). Briefly, all sections were incubated for 120 minutes with a 1:50 dilution of an anti–IL-1α rabbit polyclonal antibody (80-3054-01; Genzyme, Cambridge, MA). They then were washed and sequentially treated with a biotinylated goat anti-rabbit antibody and a streptavidin horseradish peroxidase-conjugated reagent at room temperature. The bound peroxidase was visualized using a 0.025% solution of diaminobenzidine. The sections were counterstained with hematoxylin and examined by two independent investigators who were blinded to the clinical data.

We considered the staining to be positive when 10% or more of the cancer cells showed cytoplasmic staining. The primary antibody was replaced with normal rabbit immunoglobulin G as a negative control.

Statistical Analysis

We tested the frequency distributions by the chi-square test. The differences of the means of the continuous variables between the two groups were tested by Student t test. The survival rates were calculated according to the Kaplan–Meier method, and the significance of differences in the survival rate was determined by the generalized Wilcoxon test. P values of less than 0.05 were considered to be statistically significant.

To clarify the predictive factors for liver recurrence after a curative resection, we performed a logistic regression analysis using the SPSS software package for the Macintosh, Release 6.1J (SPSS Inc., Tokyo, Japan). For this analysis, the expression of IL-1α as well as six common clinicopathologic features were chosen as predictive variables (Table 1).

Table 1. Variables Used in the Logistic Regression Analysis
VariableQuantification
  • a

    Histologic types are classified into two types: differentiated tumors (papillary and tubullar adenocarcinoma) and undifferentiated tumors (poorly differentiated adenocarcinoma, mucinous carcinoma, and signet ring cell carcinoma).

  • b

    Lymphnode metastasis is described according to the TNM classification of gastric carcinoma.

Tumor size (mm)
Macroscopic type
 Superficial1
 Borrmann type I or II2
 Borrmann type III or IV3
Histologic typea
 Differentiated1
 Undifferentiated2
Lymphatic invasion
 None1
 Mild2
 Moderate3
 Severe4
Venous invasion
 None1
 Mild2
 Moderate3
 Severe4
Lymphnode metastasisb (no. of lymph nodes)
 01
 1–62
 7–153
 > 154
Expression of IL-1α
 Negative1
 Positive2

The pathologic types were described according to the Japanese Classification of Gastric Carcinoma.12 Tumor histology was classified into two types, including differentiated and undifferentiated. The stroma in tumors were classified quantitatively into three types, including medullary, intermediate, and scirrhous. The International Union Against Cancer classification was used for tumor staging.13

RESULTS

Immunohistochemical Staining for IL-1α

The cytoplasm of cancer cells were diffusely stained for IL-1α (Fig. 1). The apical cells of the normal glands adjacent to the tumor and macrophage-like cells also showed faint staining for IL-1α. Among the 109 tumors included in this study, 60 (55.0%) showed positive staining for IL-1α.

Figure 1.

Immunohistochemical staining for IL-1α in gastric carcinoma (original magnification ×100)

Relation between the IL-1α Expression and Clinicopathologic Features

A higher incidence of the tumors with positive staining for IL-1α was observed in the patients with liver metastasis than in those without. The incidence of IL-1α positive staining was significantly higher in both the Borrmann type I or type II tumors than in those of other types, and also in differentiated tumors than in those of undifferentiated tumors. In addition, positive staining for IL-1α was significantly more likely in tumors showing a medullary stromal type than in those with a scirrhous or intermediate stromal type. No correlation was found between the IL-1α expression and either the tumor size, peritoneal metastasis, or lymph node metastasis. (Table 2).

Table 2. Relation between IL-1α Staining and the Clinicopathologic Features
FeatureIL-1α(−) (n = 49)IL-1α(+) (n = 60)Positive rate (%)P value
  • IL: interleukin; SD: standard deviation; NS: not significant.

  • a

    Borrmann type I or II vs. other types.

  • b

    Histologic type and stromal type are described according to the Japanese classification of gastric carcinoma.

  • c

    Medullary type vs. other types.

Tumor size (mm, mean ± SD)61 ± 3764 ± 25NS
Liver metastasis< 0.05
 Positive1888.9
 Negative485252.0
Peritoneal metastasisNS
 Positive2360.0
 Negative475755.3
Macroscopic type< 0.001a
 Superficial19726.9
 Borrmann type I or II62379.3
 Borrmann type III or IV243055.6
Histologic typeb< 0.001
 Differentiated93278.0
 Undifferentiated402841.2
Stromal typeb< 0.001c
 Medullary21688.9
 Intermediate163870.4
 Scirrhous31616.2
Lymph node metastasisNS
 Positive355058.8
 Negative141041.7

Impact of the IL-1α Expression on the Survival Rate and Recurrence

When 84 patients who underwent a curative resection were analyzed, the 5-year survival rates for the 44 patients with positive staining for IL-1α and the 40 patients with negative staining were 47.6% and 66.2%, respectively (P = 0.125; Fig. 2). A significantly higher incidence of liver recurrence was observed in the patients with positive staining for IL-1α than in those with negative staining for IL-1α (22.7% vs. 2.5%; P < 0.01; Table 3).

Figure 2.

Comparison of the cumulative survival curves according to the expression of IL-1α in patients undergoing a pathologically curative resection for gastric carcinoma invading the subserosa. Solid line: 40 patients with IL-1α positive tumors; broken line: 44 patients with IL-1α negative tumors.

Table 3. Relation between the Expression of IL-1α and Liver Recurrence after a Pathologically Curative Resection
ExpressionNo. of casesLiver recurrencePercentage
  • IL: interleukin.

  • a

    P < 0.01 by the chi-square test.

  • b

    P < 0.01 by the chi-square test.

IL-1α(+)441022.7a
IL-1α(−)4012.5b

Multivariate Analysis

A multivariate analysis using the logistic regression method revealed the expression of IL-1α to be the only significant predictor for liver recurrence after a curative gastrectomy for gastric carcinoma (Table 4).

Table 4. Identification of Features Contributing to Liver Recurrence Based on a Logistic Regression Analysis
VariableCoefficientStandard errorP value
  1. IL: interleukin.

IL-1α2.661.080.01

DISCUSSION

Our study showed that the expression of IL-1α in primary gastric tumors significantly correlated with both liver metastasis at operation and with liver recurrence after a curative gastrectomy in patients with gastric carcinoma with subserosal invasion. Liver metastasis occurs in less than 10% of the patients with gastric carcinomas invading not beyond the proper muscle layer. The incidence of liver metastasis increases when tumors invade deeper into the gastric wall, whereas conversely, the major cause of death in the patients with tumors invading the serosa is peritoneal metastasis.14 As a result, we chose patients with tumors invading into but not beyond the subserosa as the subjects of this study.

Tumor cells express many cytokines including IL-1α, although the number of studies regarding their contribution to the tumor growth and metastasis is still very low.15 IL-1α plays various roles in local inflammation, immunologic reactions, and cell differentiation.16, 17 Regarding gastric carcinoma, Ito et al. reported that six of eight cell lines expressed IL-1α mRNA at various levels and also confirmed that two of these cell lines secreted IL-1α protein into the culture fluid.4 In this study, we found the cytoplasm of gastric carcinoma cells to stain diffusely for IL-1α.

Cancer cells metastasize to distant organs via such steps as basement membrane degradation, migration, adhesion to the vessel walls, and angiogenesis.18 IL-1α has been reported to increase the metastatic ability of many cancer cell lines by inducing adhesion molecules on endothelial cells.1, 7, 8, 19, 20 Recently, IL-1α has been reported to enhance the invasive ability of cancer cells by inducing a urokinase-type plasminogen activator and matrix metalloproteinases.9, 10 Moreover, the concentrations of IL-1α in the tumor have been reported to correlate with the expression of thymidine phosphorylase, one of angiogenetic factors, in both colon and gastric carcinoma.21, 22 These mechanisms may explain our results that showed a significant association between the IL-1α expression and liver metastasis in patients with gastric carcinoma.

In this study, the incidence of positive staining for IL-1α was significantly higher in differentiated tumors than in undifferentiated tumors, and in tumors with scanty stroma than in those with abundant stroma. The intracellular IL-1α precursor is reported to regulate cellular differentiation in such cells as epithelial or ectodermal cells.23 Moreover, cyclooxygenase-2, which is facilitated by IL-1α, is also known to induce prostaglandin E2 (PGE2) synthesis. In addition, PGE2 inhibits fibroblast proliferation and collagen synthesis.24–26 These factors could be associated with the finding that we herein showed that IL-1α was more likely expressed in tumors either with a differentiated type or with scanty stroma.

Although the likelihood of gastric carcinoma metastasizing to the liver has been reported to be correlated with either a differentiated histology or medullary stromal features,23, 27 it remains difficult to predict liver recurrence after a curative resection for gastric carcinoma. Our results suggest that the IL-1α expression thus can be a great help in identifying the patients who are at a high risk of developing liver metastasis. The network around IL-1α is complex and it is still not completely understood. The resolution of the mechanism enhancing hematogenous metastasis, including the correlation with proteases, angiogenetic factors, or adhesion molecules, still remains to be elucidated.

Ancillary