- Email: [email protected]
Dimeric sialyl-Lex expression in gastric carcinoma correlates with venous invasion and poor outcome
GASTROENTEROLOGY 1998;114:462–470
Dimeric Sialyl-Lex Expression in Gastric Carcinoma Correlates With Venous Invasion and Poor Outcome MARGARIDA AMADO,* FA´TIMA CARNEIRO,* MA´RIO SEIXAS,* HENRIK CLAUSEN,‡ ˜ ES* and MANUEL SOBRINHO–SIMO *Institute of Molecular Pathology and Immunology of University of Porto, IPATIMUP, Medical Faculty of Porto, Hospital S. Joa ˜o, Porto, Portugal; and ‡Department of Oral Diagnostics, School of Dentistry, Copenhagen, Denmark
Background & Aims: High expression of sialyl-Lex in tumors of different organs correlates with hematogenous metastasis and adverse outcome. Dimeric sialylLex expression in gastric carcinoma was evaluated, and its prognostic significance within this setting was determined. Methods: Dimeric sialyl-Lex immunohistochemical expression in 97 gastric carcinomas was analyzed using the FH6 monoclonal antibody. Scoring was based on the percentage of immunoreactive cells: negative, low expression (I25%), and high expression (G25%). Results: Immunoreactivity was observed in 45 cases (46.4%), encompassing 27 and 18 cases with low and high expression, respectively. Significant relationships were found between dimeric sialyl-Lex expression and venous invasion (P 5 0.0025) and histological classification (P 5 0.05). No correlation was observed with other clinicopathologic features. Patients with tumors showing high expression of dimeric sialyl-Lex had a significantly shorter survival time than those with low or no expression (P 5 0.03). By multivariate analysis, pathological TNM (pTNM) staging and venous invasion emerged as independent prognostic factors in the whole series. Within the group of patients with tumors in pTNM stages II and III, dimeric sialyl-Lex was the only independent prognostic factor. Conclusions: High expression of dimeric sialyl-Lex correlates with venous invasion and poor outcome in gastric carcinoma.
he development and progression of carcinomas is often associated with dramatic alterations in cell surface carbohydrates.1,2 One of the most frequent glycosylation changes in gastric carcinoma cells is the increased expression of sialylated antigens (sialyl-Tn, sialyl-T, sialyl-Lex [S-Lex], sialyl-Lea [S-Lea], and CDw75), which has been shown to correlate with features of biological aggressiveness, namely, invasiveness and/or metastatic potential (for review, see Carneiro et al.3). S-Lex in both monomeric and dimeric forms as well as S-Lea are, among sialylated carbohydrates, particularly relevant from a biological standpoint. These antigens are ligands of E- and P-selectins in the vascular endothe-
T
lium,4 and dimeric S-Lex was shown to be a high-affinity ligand for E-selectin.5 It has been hypothesized that the selectin/S-Lex interaction mediates the sequence of adhesion of tumor cells to endothelia and their subsequent extravasation in a process that mimics the interaction between leukocytes and endothelial cells, a key step preceding diapedesis through vascular wall.6,7 Recently, it was shown that S-Lex analogues as well as antibodies anti–S-Lex inhibit the formation of new vessels both in in vitro and in vivo assays.8 The involvement of S-Lex in angiogenesis supports the notion that it may be related to hematogenous metastasis.9 Studies showing an association between the increased expression of S-Lex and/or S-Lea in tumor cells and blood-borne metastases10–12 give further support for the putative role of these antigens in the processes of angiogenesis and hematogenous metastasis. Moreover, evidence collected in several clinicopathologic studies showed that S-Lex and S-Lea antigens are potential prognostic markers in patients with tumors of different organs.13–18 In the present study, we analyzed the expression of dimeric S-Lex in a series of 97 gastric carcinomas. The relationship between the expression level of dimeric S-Lex and the clinicopathologic features of the cases, as well as the significance of dimeric S-Lex as a prognostic factor, were evaluated. The results indicate that high expression of dimeric S-Lex in gastric carcinomas is associated with venous invasion and may be useful as a potential prognostic factor in patients with intermediate pathological (pTNM) stages of gastric carcinoma.
Materials and Methods Surgical specimens from 97 gastric carcinomas resected consecutively at Hospital S. Joa˜o-Medical Faculty of Porto were included in the study. All tissue samples were collected after informed consent. Tissue fragments were fixed in 10% formalAbbreviations used in this paper: S-Lea, sialyl-Lea; S-Lex, sialyl-Lex; TBS, Tris-buffered saline. r 1998 by the American Gastroenterological Association 0016-5085/98/$3.00
March 1998
dehyde and embedded in paraffin. Sections of 4 µm were cut from each representative paraffin block and stained with H&E to evaluate and classify the lesions. Serial sections of each case were used for orcein staining (detection of vascular invasion) and immunohistochemistry.
Histological Classification Carcinomas were classified according to Laure´n19 as intestinal (n 5 51), diffuse (n 5 29), and unclassified carcinomas (n 5 17) and according to the classification proposed by Carneiro et al.20 as glandular (n 5 44), isolated cell (n 5 9), solid (n 5 9), and mixed carcinomas (n 5 35). The growth pattern was classified according to the criteria of Ming.21 The pathological staging was achieved using the unified 1987 TNM system for gastric carcinoma. Venous invasion was detected in 57 of the 97 cases. Penetration of the gastric wall and lymph node metastases were recorded in every case.
Immunohistochemistry The monoclonal antibody FH6 generated against dimeric S-Lex epitope22 was used to study the 97 cases and the normal-appearing gastric mucosa, at distance from the tumors, in 30 of these cases. The series of normal-appearing mucosa was not expanded further because the results were consistently similar in the 30 cases. A modification of the avidin-biotinperoxidase complex method23 was applied. The paraffin sections (4 µm thick) were dewaxed and treated with 0.3% hydrogen peroxide in methanol for 30 minutes to block the endogenous peroxidase, washed in Tris-buffered saline (TBS), and then incubated for 20 minutes with normal rabbit serum at a dilution of 1:5 in TBS containing 25% bovine serum albumin. Excess normal serum was removed and replaced by the monoclonal antibody FH6 diluted at 1:5. After overnight incubation (<18 hours) at 4°C, slides were washed with TBS, and the sections were incubated with a 1:200 dilution of biotin-labeled secondary antibody for 30 minutes. After washing with TBS, sections were incubated with avidin-biotinperoxidase complex (10 mg/mL of biotin-labeled peroxidase) for 60 minutes. This was followed by staining the sections for 5 minutes with 0.05% 3,3-diaminobenzidine tetrahydrochloride, freshly prepared in 0.05 mol/L Tris buffer, pH 7.6, containing 0.01% hydrogen peroxide. Finally, sections were counterstained with hematoxylin, dehydrated, and mounted. Dilution of primary antibodies, biotin-labeled secondary antibodies, and avidin-biotin-peroxidase complex were made with TBS containing 12.5% bovine serum albumin. All series included as a positive control a colon adenocarcinoma previously shown to display prominent expression of dimeric S-Lex. Negative controls were performed by substitution of the primary antibody with immunoglobulins of the same subclass and concentration as the monoclonal antibody.
Scoring A semiquantitative approach was used to score the staining reactions. Expression of dimeric S-Lex in carcinomas was classified in the following manner: negative, no immunore-
DIMERIC S–LEx IN GASTRIC CARCINOMA
463
activity or immunostaining in very rare cells; low expression, #25% of immunoreactive cells; and high expression, .25% immunoreactive cells. The cutoff value of 25% positive cells to distinguish between low and high expression was chosen taking into account the tendency for a bimodal distribution of the immunoreactivity scores in the whole series, evaluated in 5% intervals, showing an intermediate lowest level at 25%. Scoring of immunoreactivity was evaluated irrespectively of localization of positive cells and intensity of the staining. Evaluations were performed independently by two observers (M.A. and F.C.) blinded to the clinicopathologic features and follow-up data. In the event of disagreement, slides were reviewed by the observers, and a consensus was obtained.
Statistical Analysis Statistical analysis was performed using a x2 test employing Statview 4.01 software (Abacus Concepts, Inc., Berkeley, CA). Differences were considered statistically significant at P values of ,0.05. The relationship between the expression of dimeric S-Lex and survival rate of patients was assessed using both univariate and multivariate analysis (BMDP Statistical Software, Cork, Ireland). The following parameters were taken into consideration in these analyses: age and sex of the patients, tumor site, gross appearance of the tumors, histological classifications according to Laure´n19 and Carneiro et al.,20 the classification of Ming,21 amount of lymphoid infiltration, degree of desmoplastic response, depth of wall penetration, venous and lymphatic invasion, pathological staging (pTNM), and expression of dimeric S-Lex. Any death occurring within the first month of surgery was considered a postoperative death. Survival curves were calculated using the Berkson’s actuarial method in a series of patients from which the six postoperative deaths were excluded (n 5 91) and in the series of patients with tumors in pTNM stages II and III (n 5 60). Survival curves were compared statistically using the generalized Savage (Mantel– Cox) test. The evaluation of the prognostic significance of several factors was performed by multivariate regression techniques. The Cox’s proportional hazards model24 was used to evaluate the influence on survival of the aforementioned parameters. The maximum partial likelihood ratio method was used to perform a linear stepwise multivariate regression to find out the factors that are significantly or suggestively related to the survival rate.
Results Relationship Between Dimeric S-Lex Expression and Clinicopathologic Features Immunoreactivity was either absent or restricted to mucous neck cells in normal-appearing gastric mucosa distant from tumors. In the gastric mucosa at the immediate periphery of tumors, weak expression of dimeric S-Lex antigen was observed in 78% of the cases, usually in the cytoplasm of neck and foveolar/superficial cells. Dimeric S-Lex was expressed in 45 of 97 gastric
464
AMADO ET AL.
carcinoma cases (46.4%). Immunoreactivity was observed at the cell membrane (Figure 1) or in the cytoplasm (Figure 2) and varied greatly from case to case with either diffuse or focal distribution (Figure 2). No apparent regional differences were found in the expression of dimeric S-Lex between the superficial and invasive edge of the tumors. Table 1 summarizes the clinicopathologic features of the 97 cases and the corresponding dimeric S-Lex expression. Eighteen of 97 tumors (18.6%) showed high dimeric S-Lex expression (.25% immunoreactive cells). These cases encompassed, according to the classification proposed by Carneiro et al.,20 1 of 9 isolated cell carcinomas (11.1%), 4 of 44 glandular carcinomas (9.1%), 1 of 9 solid carcinomas (11.1%), and 12 of 35 mixed carcinomas (34.3%). This difference was statistically significant (P 5 0.047). Within the group of mixed carcinomas, no significant difference was found between the expression of S-Lex observed in cases predominantly composed of isolated cells (high expression in 20.0% of the cases) and those predominantly composed of glandular structures (high expression in 46.7% of the cases) (P 5 0.21). No significant relationship was found between dimeric S-Lex expression and the classification of Laure´n (Table 1). In the whole series, we found a significant relationship between high expression of dimeric S-Lex and the presence of venous invasion (P 5 0.0025) (Table 1). Neoplastic cells within blood vessels showed a pattern of dimeric S-Lex expression similar to that observed in the primary tumors (Figure 3A). The same holds true regarding the comparison between the three cases of liver metastases with the respective primaries (Figure 3B). A suggestive association was found between high
Figure 1. A case of glandular (intestinal) carcinoma showing S-Lex expression at the cell membrane (arrow ) and, to a lesser degree, in the cytoplasm of neoplastic cells (arrowhead ) (original magnification 6203).
GASTROENTEROLOGY Vol. 114, No. 3
Figure 2. In this case of isolated cell (diffuse) carcinoma, one can see a focus of neoplastic cells expressing S-Lex in the cytoplasm, adjacent to a focus of nonimmunoreactive cells (original magnification 3103).
expression of dimeric S-Lex and infiltrative macroscopic appearance (P 5 0.06). No significant association was found between dimeric S-Lex expression and the other clinicopathologic features (Table 1). Relationship Between Dimeric S-Lex Expression and Survival (Excluding Six Postoperative Deaths) Regarding the survival time, the patients were divided into two groups: short-term survivors (surviving time of ,3 years) and long-term survivors (surviving time of $3 years). The percentage of cases with high dimeric S-Lex expression was significantly greater in short-term survivors (28.6%) than in long-term survivors (2.9%) (P 5 0.009). In univariate analysis, patients with tumors showing high dimeric S-Lex expression had a significantly shorter survival time than those with low or no expression (P 5 0.03) (Figure 4). Survival rates were evaluated both at 3 and 5 years. The 3-year survival rate of patients with carcinomas showing high dimeric S-Lex expression (11.0%) was much lower than those of patients with carcinomas showing low expression (56.1%) and no expression (67.0%) (Table 2). The same trend was observed in the 5-year survival rates: 0%, 43.8%, and 46.4% for patients with carcinomas with high, low, and no expression, respectively. When only the patients with intermediate pTNM stages II and III were studied, those with carcinomas showing high dimeric S-Lex expression had a significantly shorter survival time than those with carcinomas showing low or no expression (P 5 0.02) (Figure 5). Similar results, although not reaching the threshold of statistical significance (P 5 0.2), were obtained when the study was restricted to patients with carcinomas showing venous invasion: the 5-year survival
DIMERIC S–LEx IN GASTRIC CARCINOMA
March 1998
465
Table 1. Summary of the Data on the Expression of S-Lex and the Clinicopathologic Features of the Cases Category Age (yr ) ,40 40–65 .65 Sex Men Women Site Antrum Body Fundus Cardia Gastric stump Macroscopy Fungat./ulcerofungat. Ulcerating/ulceroinfilt. Infiltrative Laure´n’s classification Intestinal Diffuse Unclassified Carneiro et al. classification Isolated cell carcinoma Glandular carcinoma Solid carcinoma Mixed carcinoma Ming’s classification Expanding Infiltrative Lymphoid infiltration Minimal/absent Abundant/moderate Desmoplasia Minimal/absent Abundant/moderate Depth of wall penetration T1 (mucosa 1 submucosa) T2 (muscular 1 subserosa) $T3 (serosa 1 adjacent organs) Venous invasion Absent Present Lymphatic invasion Absent Present Nodal metastases Absent Present pTNM stage IA IB II IIIA IIIB IV Survival a (mo) ,36 $36 Total
Cases (n) (% )
No expression (n) (% )
#25% (n) (% )
9 (9.3) 49 (50.5) 39 (40.2)
5 (55.6) 26 (53.1) 21 (53.8)
3 (33.3) 13 (26.5) 11 (28.2)
1 (11.1) 10 (20.4) 7 (17.9)
0.97
62 (63.9) 35 (36.1)
32 (51.6) 20 (57.1)
19 (30.6) 8 (22.9)
11 (17.7) 7 (20.0)
0.71
53 (54.6) 23 (23.7) 1 (1.0) 17 (17.5) 3 (3.1)
32 (60.4) 11 (47.8) 0 (0) 7 (41.2) 2 (66.7)
13 (24.5) 8 (34.8) 0 (0) 5 (29.4) 1 (33.3)
8 (15.1) 4 (17.4) 1 (100.0) 5 (29.4) 0 (0)
0.41
41 (42.3) 51 (52.6) 5 (5.2)
21 (51.2) 31 (60.8) 0 (0)
13 (31.7) 12 (23.5) 2 (40.0)
7 (17.1) 8 (15.7) 3 (60.0)
0.06
51 (52.6) 29 (29.9) 17 (17.5)
25 (49.0) 14 (48.3) 13 (76.5)
15 (29.4) 10 (34.5) 2 (11.8)
11 (21.6) 5 (17.2) 2 (11.8)
0.31
9 (9.3) 44 (45.4) 9 (9.3) 35 (36.1)
6 (66.7) 27 (61.4) 7 (77.8) 12 (34.3)
2 (22.2) 13 (29.5) 1 (11.1) 11 (31.4)
1 (11.1) 4 (9.1) 1 (11.1) 12 (34.3)
0.047
42 (43.3) 55 (56.7)
25 (59.5) 27 (49.1)
13 (31.0) 14 (25.5)
4 (9.5) 14 (25.5)
0.14
61 (62.9) 36 (37.1)
29 (47.5) 23 (63.9)
20 (32.8) 7 (19.4)
12 (19.7) 6 (16.7)
0.26
67 (69.1) 30 (30.9)
35 (52.2) 17 (56.7)
18 (26.9) 9 (30.0)
14 (20.9) 4 (13.3)
0.67
10 (10.3) 41 (42.3) 46 (47.4)
7 (70.0) 24 (58.5) 21 (45.7)
3 (30.0) 12 (29.3) 12 (26.1)
0 (0) 5 (12.2) 13 (28.3)
0.17
40 (41.2) 57 (58.8)
27 (67.5) 25 (43.9)
12 (30.0) 15 (26.3)
1 (2.5) 17 (29.8)
0.0025
9 (9.3) 88 (90.7)
6 (66.7) 46 (52.3)
2 (22.2) 25 (28.4)
1 (11.1) 17 (19.3)
0.70
38 (39.2) 59 (60.8)
20 (52.6) 32 (54.2)
12 (31.6) 15 (25.4)
6 (15.8) 12 (20.3)
0.75
8 (8.2) 23 (23.7) 18 (18.6) 26 (26.8) 19 (19.6) 3 (3.1)
5 (62.5) 12 (52.2) 14 (77.8) 14 (53.8) 6 (31.6) 1 (33.3)
3 (37.5) 8 (34.8) 1 (5.6) 8 (30.8) 6 (31.6) 1 (33.3)
0 (0) 3 (13.0) 3 (16.7) 4 (15.4) 7 (36.8) 1 (33.3)
0.17
56 (61.5) 35 (38.5) 97 (100)
26 (46.4) 23 (65.7) 52 (53.6)
14 (25.0) 11 (31.4) 27 (27.8)
16 (28.6) 1 (2.9) 18 (18.6)
Fungat., fungating; ulcerofungat., ulerofungating; ulceroinfilt., ulceroinfiltrative. six postoperative deaths (n 5 91).
aExcluding
.25% (n) (% )
P value
0.009
466
AMADO ET AL.
GASTROENTEROLOGY Vol. 114, No. 3
Table 2. Distribution of 91 Cases of Gastric Carcinoma According to the Different Clinicopathologic Parameters and Survival Rates (3 and 5 Years) in Univariate Analysis
Category
Figure 3. High expression of dimeric S-Lex (A ) in neoplastic cells within a blood vessel and (B ) in a metastatic focus in the liver (original magnification: A, 623; and B, 1553).
rate of patients with carcinomas showing high expression of dimeric S-Lex was 0%, whereas that of patients with carcinomas with low or no expression was 34%. In univariate analysis, the other clinicopathologic features significantly associated with survival of the patients were as follows: depth of wall penetration (P 5 0.003), venous invasion (P 5 0.004), and pTNM stage (P 5 0.02). The prognostic value of venous invasion was highlighted by the univariate analysis of the survival rate of patients in each pTNM stage. The relative importance of venous invasion was particularly noticeable in the lower pTNM stages: the 5-year survival rate of patients with tumors in stages IA and IB with venous invasion (30%) was significantly worse (P 5 0.04) than that of patients with tumors in the same stages and without venous invasion (68%). The multivariate analysis of the whole series showed that pTNM stage was the strongest independent prognos-
Figure 4. Survival curves of 91 patients with gastric carcinoma (excluding postoperative deaths) according to S-Lex expression. g, Negative; d, low expression; j, high expression.
Total Age (yr ) ,40 40–65 .65 Sex Men Women Site Antrum Body Fundus Gastric stump Macroscopy Fungat./ulcerofungat. Ulcerating/ulceroinfilt. Infiltrative Laure´n’s classification Intestinal Diffuse Unclassified Carneiro et al. classification Isolated cell carcinoma Glandular carcinoma Solid carcinoma Mixed carcinoma Ming’s classification Expanding Infiltrative Lymphoid infiltration Minimal/absent Abundant/moderate Desmoplasia Minimal/absent Abundant/moderate Depth of wall penetration T1 (mucosa 1 submucosa) T2 (muscular 1 subserosa) $T3 (serosa 1 adjacent organs) Venous invasion Absent Present Lymphatic invasion Absent Present pTNM stage IA IB II IIIA IIIB IV S-Lex No expression Low expression High expression
Significance 3-Year 5-Year by generalized survival survival Savage rate (% ) rate (% ) (Mantel–Cox) test 54.6
39.6
57.1 53.7 56.9
57.1 40.3 32.2
0.35
50.7 61.3
39.8 37.4
0.53
60.4 40.0 56.0 33.3
43.7 30.0 37.9 33.3
0.52
49.3 56.6 71.4
33.0 45.7 0
0.37
49.4 57.7 71.4
34.2 50.9 42.9
0.71
72.9 65.1 66.7 29.4
56.7 40.9 50.0 29.4
0.31
59.2 50.7
38.1 41.0
0.82
48.7 63.3
39.6 40.7
0.29
54.0 55.8
38.0 43.7
0.47
80.0 70.7
60.0 34.6
0.003
32.1
27.5
74.4 40.1
53.3 29.6
0.004
62.6 49.0
51.4 28.6
0.09
75.0 70.4 64.7 45.5 16.1 50.0
75.0 46.3 41.4 32.5 16.1 0
0.02
67.0 56.1 11.0
46.4 43.8 0
0.03
NOTE. Postoperative deaths were excluded from this analysis (n 5 6).
DIMERIC S–LEx IN GASTRIC CARCINOMA
March 1998
467
Table 4. Summary of the Results Obtained With the Stepwise Multivariate Regression of Prognostic Factors With the Maximum Partial Likelihood Ratio Method in 91 Cases of the Whole Series Step factors
Log likelihood
P value
0 1 Stage (pTNM) 2 Venous invasion
2169.532 2163.868 2162.000
0.001 0.053
NOTE. Postoperative deaths were excluded from this analysis (n 5 6).
Figure 5. Survival curves of 60 patients with gastric carcinomas in stages II and III (excluding postoperative deaths) according to S-Lex expression. g, Negative; d, low expression, j, high expression.
tic factor followed by venous invasion (Tables 3 and 4). The multivariate analysis of the 60 cases in pTNM stages II and III showed that high dimeric S-Lex expression was the only independent prognostic factor influencing survival rate (Tables 5 and 6). When dimeric S-Lex expression was removed from the stepwise multivariate analysis, pTNM stages II and III stood up as the strongest independent prognostic factor followed by venous invasion (data not shown).
Discussion In the present study, we analyzed the immunohistochemical expression of dimeric S-Lex in gastric carcinomas and its relationship with the clinicopathologic features of the tumors and the survival rate of the patients. In nonneoplastic mucosa, the expression of dimeric S-Lex, when present, was only observed in the neck cells. This finding is in agreement with the previously reported Table 3. Summary of the Results Obtained With the Multivariate Analysis of Prognostic Factors With Cox’s Model in 91 Cases of the Whole Series Factors
P value
Age Sex Site Macroscopy Laure´n’s classification Carneiro et al. classification Ming’s classification Lymphoid infiltration Desmoplasia Venous invasion Stage (pTNM) S-Lex (no, low, and high expression)
0.07 0.53 0.60 0.20 0.83 0.08 0.83 0.30 0.47 0.004 0.0008 0.04
NOTE. Postoperative deaths were excluded from this analysis (n 5 6).
restricted immunoreactivity of FH6 antibody in normal adult tissues.25 In gastric carcinomas, we detected dimeric S-Lex expression in 46.4% of the cases. This rate of expression falls within the range of those previously reported in gastric carcinoma, from 25.5% to 74.1%.18,26–29 The varying expression level of dimeric S-Lex in gastric carcinoma found by different groups may reflect the heterogeneity of the series regarding the histological types of tumors and the different specificities of the antibodies. It has been reported previously that differentiated carcinomas (roughly corresponding to glandular [intestinal] carcinomas) show a higher expression of dimeric S-Lex than undifferentiated carcinomas (the latter group encompassing isolated–cell type [diffuse] carcinomas and solid carcinomas).26,28,29 Because of the use of different histological classifications, it is, however, difficult to compare the results of the different series. In our study, we found no significant relationship between dimeric S-Lex expression and the histological type of the tumors classified according to Laure´n, which was at variance with the results obtained when using the classification proposed by Carneiro et al.20: dimeric S-Lex expression was significantly greater in mixed carcinomas (encompassing Table 5. Summary of the Results Obtained With the Multivariate Analysis of Prognostic Factors With Cox’s Model in 60 Patients With Tumors in pTNM Stages II and III Factors
P value
Age Sex Site Macroscopy Laure´n’s classification Carneiro et al. classification Ming’s classification Lymphoid infiltration Desmoplasia Venous invasion Stage (pTNM) S-Lex (no, low, and high expression)
0.66 0.71 0.82 0.93 0.59 0.30 0.15 0.36 0.89 0.15 0.07 0.03
NOTE. Cases in pTNM stages I and IV (n 5 34) and postoperative deaths of patients with tumors in stages II and III (n 5 3) were excluded from this analysis.
468
AMADO ET AL.
GASTROENTEROLOGY Vol. 114, No. 3
Table 6. Summary of the Results Obtained With the Stepwise Multivariate Regression of Prognostic Factors With the Maximum Partial Likelihood Ratio Method in 60 Patients With Tumors in pTNM Stages II and III Step factors
Log likelihood
P value
0 1 S-Lex (no, low, and high expression)
2104.083 2101.779
0.03
NOTE. Cases in pTNM stages I and IV (n 5 34) and postoperative deaths of patients with tumors in stages II and III (n 5 3) were excluded from this analysis.
more than one morphological type) than in tumors with a ‘‘pure’’ histological type (glandular, isolated cell type, or solid carcinomas). We have also observed, within the group of mixed carcinomas of our series, that the prevalence of cases strongly expressing dimeric S-Lex was greater, although not significantly, in cases predominantly composed of glandular (intestinal) areas than in cases in which isolated cells were the predominant component. Interestingly, the prevalence of a high expression of dimeric S-Lex in the latter subgroup was greater than that observed in carcinomas of pure glandular (intestinal) phenotype. It remains to be seen if the high expression of dimeric S-Lex in mixed carcinomas may partly account for the reported worse prognosis of this tumor type in comparison with those of carcinomas of ‘‘pure’’ histological type, as shown previously in a larger series of gastric carcinomas.20 Interestingly, a significant correlation was found between dimeric S-Lex expression and venous invasion. To the best of our knowledge, this is the first study to show such a relationship in gastric carcinoma, and we admit that the systematic search for venous invasion that we performed in the present study, using orcein-stained sections, may have contributed to increase the accuracy of the detection of venous invasion and to highlight its relationship with dimeric S-Lex expression. However, apart from the putative effect on angiogenesis, the mechanisms through which the neoplastic cells expressing dimeric S-Lex acquire vascular invasive properties remain to be explored. Further, the relationship between dimeric S-Lex expression and vascular invasion may also reflect the ability to detect thrombi of neoplastic cells within the vessels. Dimeric S-Lex being a ligand for P-selectin, which is also expressed in platelets,30 may thus contribute to the formation of neoplastic thrombi and eventually facilitate the visualization of neoplastic cells in the vascular lumina. In keeping with the findings of Ikeda et al.29 and Nakamori et al.,18 and at variance with those obtained by Dohi et al.26 who observed a correlation between expres-
sion of dimeric S-Lex (recognized in all these studies by FH6) and depth of penetration of gastric wall, we did not observe any significant relationship between expression of dimeric S-Lex and other features of tumor aggressiveness, apart from the aforementioned vascular invasion. In the present study, high expression of dimeric S-Lex was found to be a prognostic marker. In univariate analysis, dimeric S-Lex expression was found to be significantly related to survival rate together with depth of wall penetration, venous invasion, and pTNM staging. High expression of dimeric S-Lex kept its prognostic meaning when the univariate analysis was restricted to patients with carcinomas in intermediate pTNM stages and to patients with carcinomas showing venous invasion, thus supporting its identification as a prognostic marker in gastric carcinoma. Despite the aforementioned prognostic value at univariate analysis, dimeric S-Lex expression did not keep an independent prognostic significance in the multivariate analysis of the whole series. In such analysis, only pTNM staging and venous invasion kept prognostic significance, which was in agreement with data on record.20 When the influence of pTNM staging was reduced markedly by restricting the multivariate analysis to the 60 cases with tumors in intermediate stages (pTNM stages II and III) dimeric S-Lex expression emerged as the single significant prognostic factor. The appearance of either venous invasion or dimeric S-Lex expression as independent prognostic factors in two different settings of stepwise multivariate analysis fits with the strong correlation between them. The reason for the prominence of venous invasion in the multivariate analysis of the whole series stems from the extreme importance from a prognostic standpoint of the presence of venous invasion in the lower pTNM stages. The same does not hold true when the multivariate analysis is restricted to pTNM intermediate stages. In this setting, dimeric S-Lex expression shows an advantage over venous invasion as an independent prognostic factor partly because the influence of venous invasion is substantially reduced within the group of carcinomas in intermediate stages and partly because dimeric S-Lex plays a role in prognosis per se apart from that linked to venous invasion. In fact, dimeric S-Lex antigen, as a high-affinity ligand to E-selectin,5 is thought to facilitate the extravasation from vessels of neoplastic cells expressing this antigen.4 It is tempting to suggest that this functional role of dimeric S-Lex may explain, in the intermediate pTNM stages, the significantly worse prognosis of patients harboring tumors with high expression of dimeric S-Lex through the ability to give rise to distant metastases that may have not been detected at the time of initial diagnosis. This added
March 1998
value of dimeric S-Lex is supported by the trend towards the worse prognosis of patients with carcinomas showing high dimeric S-Lex expression when the comparison is restricted to cases with venous invasion. Our findings on the relationship between the expression of dimeric S-Lex and the survival rate of patients do not fit entirely with those previously reported by Nakamori et al.18 These investigators coanalyzed the putative role of dimeric S-Lex and S-Lea as prognostic factors in a series of 137 gastric carcinomas, and in their hands, S-Lea but not dimeric S-Lex was found to be independently related to the survival rate of the patients. The discrepancy between our findings and those of Nakamori et al. may partly depend on the different scoring systems of immunoreactivity used in the two series. While Nakamori et al.18 just scored cases as negative or positive, we scored positive cases in two levels of immunoreactivity, i.e., low and high rates of expression, and we have shown that cases with low expression of dimeric S-Lex are very similar from the prognostic standpoint to cases without any expression at all and distinct from cases with high expression of dimeric S-Lex. It remains to be shown whether or not the increased biological and clinical aggressiveness of the tumors of the latter group may be ascribed to the increased availability of the dimeric S-Lex antigen at the cell surface. Altogether, the findings of the present study lead us to conclude that the clinical significance of dimeric S-Lex expression is twofold: relationship with venous invasion and relevance as prognostic factor in cases with intermediate pTNM stages. Based on this clinical significance, we think that it is important to perform the immunohistochemical detection of dimeric S-Lex in every gastric carcinoma. Whenever present, it indicates the need to search for gross vascular invasion; if there would be different therapy protocols for cases with vascular invasion and for cases without vascular invasion, cases with high dimeric S-Lex expression should be included in the former. One should also consider the possibility of including such cases within a subgroup of patients benefiting from adjuvant therapy targeted for occult distant metastases.
References 1. Hakomori S. Aberrant glycosylation in tumors and tumor-associated carbohydrate antigens. Adv Cancer Res 1989;52:257–331. 2. Dabelsteen E. Cell surface carbohydrates as prognostic markers in human carcinomas. J Pathol 1996;179:358–369. 3. Carneiro F, Amado M, David L, Clausen H, Sobrinho-Simo ˜es M. Glycosylation features of gastric carcinoma initiation and progression. A review with emphasis on simple mucin-type carbohydrates and histo-blood group antigens of the Lewis system. Eur J Cancer Prev 1994;3(Suppl 2):39–46.
DIMERIC S–LEx IN GASTRIC CARCINOMA
469
4. Varki A. Selectin ligands. Proc Natl Acad Sci USA 1994;91:7390– 7397. 5. Patel T, Goelz S, Lobb R, Parekh R. Isolation and characterization of natural protein-associated carbohydrate ligands for E-selectin. Biochemistry 1994;33:14815–14824. 6. Irimura T. Cancer metastasis determined by carbohydratemediated cell adhesion. Adv Exp Med Biol 1994;353:27–34. 7. Springer TA. Traffic signals for lymphocyte recirculation and leukocyte emigration: the multistep paradigm. Cell 1994;76:301– 314. 8. Nguyen M, Eilber F, Defrees S. Novel synthetic analogs of sialyl Lewis X can inhibit angiogenesis in vitro and in vivo. Biochem Biophys Res Commun 1996;228:716–723. 9. Tanigawa N, Amaya H, Matsumura M, Shimomatsuya T, Horiuchi T, Muraoka R, Iki M. Extent of tumor vascularization correlates with prognosis and hematogenous metastasis in gastric carcinomas. Cancer Res 1996;56:2671–2676. 10. Hoff S, Irimura T, Matsushita Y, Ota D, Cleary K, Hakomori S. Metastatic potential of colon carcinoma. Expression of ABO/Lewisrelated antigens. Arch Surg 1990;125:206–209. 11. Irimura T, Nakamori S, Matsushita Y, Taniuchi Y, Todoroki N, Tsuji T, Izumi Y, Kawamura Y, Hoff S, Cleary K, Ota D. Colorectal cancer metastasis determined by carbohydrate-mediated cell adhesion: role of sialyl-Lex antigens. Semin Cancer Biol 1993;4:319–324. 12. Ogawa J, Sano A, Koide S, Shohtsu A. Relation between recurrence and expression of proliferating cell nuclear antigen, sialyl Lewisx and sialyl Lewisa in lung cancer. J Thorac Cardiovasc Surg 1994;108:329–336. 13. Inufusa H, Kojima N, Yasutomi M, Hakomori S. Human lung adenocarcinoma cell lines with different lung colonization potential (LCP), and a correlation between expression of sialosyl dimeric Le (x) (defined by Mab FH6) and LCP. Clin Exp Metastasis 1991;9:245–257. 14. Nakamori S, Kameyama M, Imaoka S, Furukawa H, Ishikawa O, Sasaki Y, Kabuto T, Iwanaga T, Matsushita Y, Irimura T. Increased expression of sialyl Lewisx antigen correlates with poor survival in patients with colorectal carcinoma: clinicopathological and immunohistochemical study. Cancer Res 1993;53:3632–3637. 15. Narita T, Funahashi H, Satoh Y, Watanabe T, Sakamoto J, Takagi H. Association of expression of blood group-related carbohydrate antigens with prognosis in breast cancer. Cancer 1993;71:3044– 3053. 16. Jørgensen T, Berner A, Kaalhus O, Tveter K, Danielsen H, Bryne M. Up-regulation of the oligosaccharide Sialyl Lewisx: a new prognostic parameter in metastatic prostate cancer. Cancer Res 1995;55:1817–1819. 17. Nakayama T, Watanabe M, Katsumata T, Teramoto T, Kitajima M. Expression of sialyl Lewisa as a new prognostic factor for patients with advanced colorectal carcinoma. Cancer 1995;75:2051– 2056. 18. Nakamori S, Furukawa H, Hiratsuka M, Iwanaga T, Imaoka S, Ishikawa O, Kabuto T, Sasaki Y, Kameyama M, Ishiguro S, Irimura T. Expression of carbohydrate antigen sialyl Lea: a new functional prognostic factor in gastric cancer. J Clin Oncol 1997;15:816– 825. 19. Laure´n P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histoclinical classification. Acta Pathol Microbiol Scand 1965;64: 31–49. 20. Carneiro F, Seixas M, Sobrinho-Simo ˜es M. New elements for an updated classification of the carcinomas of the stomach. Pathol Res Pract 1995;191:571–584. 21. Ming S-C. Gastric carcinoma: a pathobiological classification. Cancer 1977;39:2475–2485. 22. Fukushi Y, Nudelman E, Levery SB, Hakomori S, Rauvala H. Novel fucolipids accumulating in human adenocarcinoma: III a hybridoma antibody (FH6) defining a human cancer-associated difuco-
470
23.
24. 25.
26.
27.
28.
AMADO ET AL.
ganglioside (VI3NeuAcV3III3Fuc2nLc 6). J Biol Chem 1984;259: 10511–10517. Hsu SM, Faine L, Fanger H. A comparative study of the peroxidaseantiperoxidase method and the avidin-biotin complex method for studying polypeptide hormones with radioimmunoassay antibodies. Am J Clin Pathol 1981;75:734–738. Cox DR. Regression models and life tables. J R Statist Soc 1972;34:187–202. Fukushi Y, Kannagi R, Hakomori S, Shepard T, Kulander B, Singer J. Location and distribution of difucoganglioside (VI3NeuAcV3III3Fuc2nLc6) in normal and tumor tissues defined by its monoclonal antibody FH6. Cancer Res 1985;45:3711–3717. Dohi T, Abe K, Ohshiba S. Immunohistochemical study of carbohydrate antigen expression in gastric carcinoma. Gastroenterol Jpn 1989;24:239–245. Sakamoto J, Watanabe T, Tokumaru T, Takagi H, Nakazato H, Lloyd KO. Expression of Lewisa, Lewisb, Lewisx, Lewisy, sialyl-Lewisa and sialyl-Lewisx blood group antigens in human gastric carcinoma and in normal gastric tissue. Cancer Res 1989;49:745–752. Murata K, Egami H, Shibata Y, Sakamoto K, Misumi A, Ogawa M.
GASTROENTEROLOGY Vol. 114, No. 3
Expression of blood group-related antigens, ABH, Lewisa, Lewisb, Lewisx, Lewisy, CA 19-9 and CSLEX1 in early cancer, intestinal metaplasia, and uninvolved mucosa of the stomach. Am J Clin Pathol 1992;98:67–75. 29. Ikeda Y, Mori M, Kajiyama K, Haraguchi Y, Sasaki O, Sugimachi K. Immunohistochemical expression of sialyl Tn, sialyl Lewisa, sialyl Lewisa-b-, and sialyl Lewisx in primary tumor and metastatic lymph nodes in human gastric cancer. J Surg Oncol 1996;62:171–176. 30. Palabrica T, Lobb R, Furie B, Aronovitz M, Benjamin C, Hsu Y-M, Sajer S, Furie B. Leukocyte accumulation promoting fibrin deposition is mediated in vivo by P-selectin on adherent platelets. Nature 1992;359:848–851.
Received August 13, 1997. Accepted November 18, 1997. Address requests for reprints to: Margarida Amado, Ph.D., Institute of Molecular Pathology and Immunology of University of Porto, IPATIMUP, Rua Dr. Roberto Frias s/n, 4200 Porto, Portugal. Fax: (351) 2-557-0799. Supported by project 2/2.1/SAU/1356/95 from PRAXIS XXI.
Dimeric Sialyl-Lex Expression in Gastric Carcinoma Correlates With Venous Invasion and Poor Outcome MARGARIDA AMADO,* FA´TIMA CARNEIRO,* MA´RIO SEIXAS,* HENRIK CLAUSEN,‡ ˜ ES* and MANUEL SOBRINHO–SIMO *Institute of Molecular Pathology and Immunology of University of Porto, IPATIMUP, Medical Faculty of Porto, Hospital S. Joa ˜o, Porto, Portugal; and ‡Department of Oral Diagnostics, School of Dentistry, Copenhagen, Denmark
Background & Aims: High expression of sialyl-Lex in tumors of different organs correlates with hematogenous metastasis and adverse outcome. Dimeric sialylLex expression in gastric carcinoma was evaluated, and its prognostic significance within this setting was determined. Methods: Dimeric sialyl-Lex immunohistochemical expression in 97 gastric carcinomas was analyzed using the FH6 monoclonal antibody. Scoring was based on the percentage of immunoreactive cells: negative, low expression (I25%), and high expression (G25%). Results: Immunoreactivity was observed in 45 cases (46.4%), encompassing 27 and 18 cases with low and high expression, respectively. Significant relationships were found between dimeric sialyl-Lex expression and venous invasion (P 5 0.0025) and histological classification (P 5 0.05). No correlation was observed with other clinicopathologic features. Patients with tumors showing high expression of dimeric sialyl-Lex had a significantly shorter survival time than those with low or no expression (P 5 0.03). By multivariate analysis, pathological TNM (pTNM) staging and venous invasion emerged as independent prognostic factors in the whole series. Within the group of patients with tumors in pTNM stages II and III, dimeric sialyl-Lex was the only independent prognostic factor. Conclusions: High expression of dimeric sialyl-Lex correlates with venous invasion and poor outcome in gastric carcinoma.
he development and progression of carcinomas is often associated with dramatic alterations in cell surface carbohydrates.1,2 One of the most frequent glycosylation changes in gastric carcinoma cells is the increased expression of sialylated antigens (sialyl-Tn, sialyl-T, sialyl-Lex [S-Lex], sialyl-Lea [S-Lea], and CDw75), which has been shown to correlate with features of biological aggressiveness, namely, invasiveness and/or metastatic potential (for review, see Carneiro et al.3). S-Lex in both monomeric and dimeric forms as well as S-Lea are, among sialylated carbohydrates, particularly relevant from a biological standpoint. These antigens are ligands of E- and P-selectins in the vascular endothe-
T
lium,4 and dimeric S-Lex was shown to be a high-affinity ligand for E-selectin.5 It has been hypothesized that the selectin/S-Lex interaction mediates the sequence of adhesion of tumor cells to endothelia and their subsequent extravasation in a process that mimics the interaction between leukocytes and endothelial cells, a key step preceding diapedesis through vascular wall.6,7 Recently, it was shown that S-Lex analogues as well as antibodies anti–S-Lex inhibit the formation of new vessels both in in vitro and in vivo assays.8 The involvement of S-Lex in angiogenesis supports the notion that it may be related to hematogenous metastasis.9 Studies showing an association between the increased expression of S-Lex and/or S-Lea in tumor cells and blood-borne metastases10–12 give further support for the putative role of these antigens in the processes of angiogenesis and hematogenous metastasis. Moreover, evidence collected in several clinicopathologic studies showed that S-Lex and S-Lea antigens are potential prognostic markers in patients with tumors of different organs.13–18 In the present study, we analyzed the expression of dimeric S-Lex in a series of 97 gastric carcinomas. The relationship between the expression level of dimeric S-Lex and the clinicopathologic features of the cases, as well as the significance of dimeric S-Lex as a prognostic factor, were evaluated. The results indicate that high expression of dimeric S-Lex in gastric carcinomas is associated with venous invasion and may be useful as a potential prognostic factor in patients with intermediate pathological (pTNM) stages of gastric carcinoma.
Materials and Methods Surgical specimens from 97 gastric carcinomas resected consecutively at Hospital S. Joa˜o-Medical Faculty of Porto were included in the study. All tissue samples were collected after informed consent. Tissue fragments were fixed in 10% formalAbbreviations used in this paper: S-Lea, sialyl-Lea; S-Lex, sialyl-Lex; TBS, Tris-buffered saline. r 1998 by the American Gastroenterological Association 0016-5085/98/$3.00
March 1998
dehyde and embedded in paraffin. Sections of 4 µm were cut from each representative paraffin block and stained with H&E to evaluate and classify the lesions. Serial sections of each case were used for orcein staining (detection of vascular invasion) and immunohistochemistry.
Histological Classification Carcinomas were classified according to Laure´n19 as intestinal (n 5 51), diffuse (n 5 29), and unclassified carcinomas (n 5 17) and according to the classification proposed by Carneiro et al.20 as glandular (n 5 44), isolated cell (n 5 9), solid (n 5 9), and mixed carcinomas (n 5 35). The growth pattern was classified according to the criteria of Ming.21 The pathological staging was achieved using the unified 1987 TNM system for gastric carcinoma. Venous invasion was detected in 57 of the 97 cases. Penetration of the gastric wall and lymph node metastases were recorded in every case.
Immunohistochemistry The monoclonal antibody FH6 generated against dimeric S-Lex epitope22 was used to study the 97 cases and the normal-appearing gastric mucosa, at distance from the tumors, in 30 of these cases. The series of normal-appearing mucosa was not expanded further because the results were consistently similar in the 30 cases. A modification of the avidin-biotinperoxidase complex method23 was applied. The paraffin sections (4 µm thick) were dewaxed and treated with 0.3% hydrogen peroxide in methanol for 30 minutes to block the endogenous peroxidase, washed in Tris-buffered saline (TBS), and then incubated for 20 minutes with normal rabbit serum at a dilution of 1:5 in TBS containing 25% bovine serum albumin. Excess normal serum was removed and replaced by the monoclonal antibody FH6 diluted at 1:5. After overnight incubation (<18 hours) at 4°C, slides were washed with TBS, and the sections were incubated with a 1:200 dilution of biotin-labeled secondary antibody for 30 minutes. After washing with TBS, sections were incubated with avidin-biotinperoxidase complex (10 mg/mL of biotin-labeled peroxidase) for 60 minutes. This was followed by staining the sections for 5 minutes with 0.05% 3,3-diaminobenzidine tetrahydrochloride, freshly prepared in 0.05 mol/L Tris buffer, pH 7.6, containing 0.01% hydrogen peroxide. Finally, sections were counterstained with hematoxylin, dehydrated, and mounted. Dilution of primary antibodies, biotin-labeled secondary antibodies, and avidin-biotin-peroxidase complex were made with TBS containing 12.5% bovine serum albumin. All series included as a positive control a colon adenocarcinoma previously shown to display prominent expression of dimeric S-Lex. Negative controls were performed by substitution of the primary antibody with immunoglobulins of the same subclass and concentration as the monoclonal antibody.
Scoring A semiquantitative approach was used to score the staining reactions. Expression of dimeric S-Lex in carcinomas was classified in the following manner: negative, no immunore-
DIMERIC S–LEx IN GASTRIC CARCINOMA
463
activity or immunostaining in very rare cells; low expression, #25% of immunoreactive cells; and high expression, .25% immunoreactive cells. The cutoff value of 25% positive cells to distinguish between low and high expression was chosen taking into account the tendency for a bimodal distribution of the immunoreactivity scores in the whole series, evaluated in 5% intervals, showing an intermediate lowest level at 25%. Scoring of immunoreactivity was evaluated irrespectively of localization of positive cells and intensity of the staining. Evaluations were performed independently by two observers (M.A. and F.C.) blinded to the clinicopathologic features and follow-up data. In the event of disagreement, slides were reviewed by the observers, and a consensus was obtained.
Statistical Analysis Statistical analysis was performed using a x2 test employing Statview 4.01 software (Abacus Concepts, Inc., Berkeley, CA). Differences were considered statistically significant at P values of ,0.05. The relationship between the expression of dimeric S-Lex and survival rate of patients was assessed using both univariate and multivariate analysis (BMDP Statistical Software, Cork, Ireland). The following parameters were taken into consideration in these analyses: age and sex of the patients, tumor site, gross appearance of the tumors, histological classifications according to Laure´n19 and Carneiro et al.,20 the classification of Ming,21 amount of lymphoid infiltration, degree of desmoplastic response, depth of wall penetration, venous and lymphatic invasion, pathological staging (pTNM), and expression of dimeric S-Lex. Any death occurring within the first month of surgery was considered a postoperative death. Survival curves were calculated using the Berkson’s actuarial method in a series of patients from which the six postoperative deaths were excluded (n 5 91) and in the series of patients with tumors in pTNM stages II and III (n 5 60). Survival curves were compared statistically using the generalized Savage (Mantel– Cox) test. The evaluation of the prognostic significance of several factors was performed by multivariate regression techniques. The Cox’s proportional hazards model24 was used to evaluate the influence on survival of the aforementioned parameters. The maximum partial likelihood ratio method was used to perform a linear stepwise multivariate regression to find out the factors that are significantly or suggestively related to the survival rate.
Results Relationship Between Dimeric S-Lex Expression and Clinicopathologic Features Immunoreactivity was either absent or restricted to mucous neck cells in normal-appearing gastric mucosa distant from tumors. In the gastric mucosa at the immediate periphery of tumors, weak expression of dimeric S-Lex antigen was observed in 78% of the cases, usually in the cytoplasm of neck and foveolar/superficial cells. Dimeric S-Lex was expressed in 45 of 97 gastric
464
AMADO ET AL.
carcinoma cases (46.4%). Immunoreactivity was observed at the cell membrane (Figure 1) or in the cytoplasm (Figure 2) and varied greatly from case to case with either diffuse or focal distribution (Figure 2). No apparent regional differences were found in the expression of dimeric S-Lex between the superficial and invasive edge of the tumors. Table 1 summarizes the clinicopathologic features of the 97 cases and the corresponding dimeric S-Lex expression. Eighteen of 97 tumors (18.6%) showed high dimeric S-Lex expression (.25% immunoreactive cells). These cases encompassed, according to the classification proposed by Carneiro et al.,20 1 of 9 isolated cell carcinomas (11.1%), 4 of 44 glandular carcinomas (9.1%), 1 of 9 solid carcinomas (11.1%), and 12 of 35 mixed carcinomas (34.3%). This difference was statistically significant (P 5 0.047). Within the group of mixed carcinomas, no significant difference was found between the expression of S-Lex observed in cases predominantly composed of isolated cells (high expression in 20.0% of the cases) and those predominantly composed of glandular structures (high expression in 46.7% of the cases) (P 5 0.21). No significant relationship was found between dimeric S-Lex expression and the classification of Laure´n (Table 1). In the whole series, we found a significant relationship between high expression of dimeric S-Lex and the presence of venous invasion (P 5 0.0025) (Table 1). Neoplastic cells within blood vessels showed a pattern of dimeric S-Lex expression similar to that observed in the primary tumors (Figure 3A). The same holds true regarding the comparison between the three cases of liver metastases with the respective primaries (Figure 3B). A suggestive association was found between high
Figure 1. A case of glandular (intestinal) carcinoma showing S-Lex expression at the cell membrane (arrow ) and, to a lesser degree, in the cytoplasm of neoplastic cells (arrowhead ) (original magnification 6203).
GASTROENTEROLOGY Vol. 114, No. 3
Figure 2. In this case of isolated cell (diffuse) carcinoma, one can see a focus of neoplastic cells expressing S-Lex in the cytoplasm, adjacent to a focus of nonimmunoreactive cells (original magnification 3103).
expression of dimeric S-Lex and infiltrative macroscopic appearance (P 5 0.06). No significant association was found between dimeric S-Lex expression and the other clinicopathologic features (Table 1). Relationship Between Dimeric S-Lex Expression and Survival (Excluding Six Postoperative Deaths) Regarding the survival time, the patients were divided into two groups: short-term survivors (surviving time of ,3 years) and long-term survivors (surviving time of $3 years). The percentage of cases with high dimeric S-Lex expression was significantly greater in short-term survivors (28.6%) than in long-term survivors (2.9%) (P 5 0.009). In univariate analysis, patients with tumors showing high dimeric S-Lex expression had a significantly shorter survival time than those with low or no expression (P 5 0.03) (Figure 4). Survival rates were evaluated both at 3 and 5 years. The 3-year survival rate of patients with carcinomas showing high dimeric S-Lex expression (11.0%) was much lower than those of patients with carcinomas showing low expression (56.1%) and no expression (67.0%) (Table 2). The same trend was observed in the 5-year survival rates: 0%, 43.8%, and 46.4% for patients with carcinomas with high, low, and no expression, respectively. When only the patients with intermediate pTNM stages II and III were studied, those with carcinomas showing high dimeric S-Lex expression had a significantly shorter survival time than those with carcinomas showing low or no expression (P 5 0.02) (Figure 5). Similar results, although not reaching the threshold of statistical significance (P 5 0.2), were obtained when the study was restricted to patients with carcinomas showing venous invasion: the 5-year survival
DIMERIC S–LEx IN GASTRIC CARCINOMA
March 1998
465
Table 1. Summary of the Data on the Expression of S-Lex and the Clinicopathologic Features of the Cases Category Age (yr ) ,40 40–65 .65 Sex Men Women Site Antrum Body Fundus Cardia Gastric stump Macroscopy Fungat./ulcerofungat. Ulcerating/ulceroinfilt. Infiltrative Laure´n’s classification Intestinal Diffuse Unclassified Carneiro et al. classification Isolated cell carcinoma Glandular carcinoma Solid carcinoma Mixed carcinoma Ming’s classification Expanding Infiltrative Lymphoid infiltration Minimal/absent Abundant/moderate Desmoplasia Minimal/absent Abundant/moderate Depth of wall penetration T1 (mucosa 1 submucosa) T2 (muscular 1 subserosa) $T3 (serosa 1 adjacent organs) Venous invasion Absent Present Lymphatic invasion Absent Present Nodal metastases Absent Present pTNM stage IA IB II IIIA IIIB IV Survival a (mo) ,36 $36 Total
Cases (n) (% )
No expression (n) (% )
#25% (n) (% )
9 (9.3) 49 (50.5) 39 (40.2)
5 (55.6) 26 (53.1) 21 (53.8)
3 (33.3) 13 (26.5) 11 (28.2)
1 (11.1) 10 (20.4) 7 (17.9)
0.97
62 (63.9) 35 (36.1)
32 (51.6) 20 (57.1)
19 (30.6) 8 (22.9)
11 (17.7) 7 (20.0)
0.71
53 (54.6) 23 (23.7) 1 (1.0) 17 (17.5) 3 (3.1)
32 (60.4) 11 (47.8) 0 (0) 7 (41.2) 2 (66.7)
13 (24.5) 8 (34.8) 0 (0) 5 (29.4) 1 (33.3)
8 (15.1) 4 (17.4) 1 (100.0) 5 (29.4) 0 (0)
0.41
41 (42.3) 51 (52.6) 5 (5.2)
21 (51.2) 31 (60.8) 0 (0)
13 (31.7) 12 (23.5) 2 (40.0)
7 (17.1) 8 (15.7) 3 (60.0)
0.06
51 (52.6) 29 (29.9) 17 (17.5)
25 (49.0) 14 (48.3) 13 (76.5)
15 (29.4) 10 (34.5) 2 (11.8)
11 (21.6) 5 (17.2) 2 (11.8)
0.31
9 (9.3) 44 (45.4) 9 (9.3) 35 (36.1)
6 (66.7) 27 (61.4) 7 (77.8) 12 (34.3)
2 (22.2) 13 (29.5) 1 (11.1) 11 (31.4)
1 (11.1) 4 (9.1) 1 (11.1) 12 (34.3)
0.047
42 (43.3) 55 (56.7)
25 (59.5) 27 (49.1)
13 (31.0) 14 (25.5)
4 (9.5) 14 (25.5)
0.14
61 (62.9) 36 (37.1)
29 (47.5) 23 (63.9)
20 (32.8) 7 (19.4)
12 (19.7) 6 (16.7)
0.26
67 (69.1) 30 (30.9)
35 (52.2) 17 (56.7)
18 (26.9) 9 (30.0)
14 (20.9) 4 (13.3)
0.67
10 (10.3) 41 (42.3) 46 (47.4)
7 (70.0) 24 (58.5) 21 (45.7)
3 (30.0) 12 (29.3) 12 (26.1)
0 (0) 5 (12.2) 13 (28.3)
0.17
40 (41.2) 57 (58.8)
27 (67.5) 25 (43.9)
12 (30.0) 15 (26.3)
1 (2.5) 17 (29.8)
0.0025
9 (9.3) 88 (90.7)
6 (66.7) 46 (52.3)
2 (22.2) 25 (28.4)
1 (11.1) 17 (19.3)
0.70
38 (39.2) 59 (60.8)
20 (52.6) 32 (54.2)
12 (31.6) 15 (25.4)
6 (15.8) 12 (20.3)
0.75
8 (8.2) 23 (23.7) 18 (18.6) 26 (26.8) 19 (19.6) 3 (3.1)
5 (62.5) 12 (52.2) 14 (77.8) 14 (53.8) 6 (31.6) 1 (33.3)
3 (37.5) 8 (34.8) 1 (5.6) 8 (30.8) 6 (31.6) 1 (33.3)
0 (0) 3 (13.0) 3 (16.7) 4 (15.4) 7 (36.8) 1 (33.3)
0.17
56 (61.5) 35 (38.5) 97 (100)
26 (46.4) 23 (65.7) 52 (53.6)
14 (25.0) 11 (31.4) 27 (27.8)
16 (28.6) 1 (2.9) 18 (18.6)
Fungat., fungating; ulcerofungat., ulerofungating; ulceroinfilt., ulceroinfiltrative. six postoperative deaths (n 5 91).
aExcluding
.25% (n) (% )
P value
0.009
466
AMADO ET AL.
GASTROENTEROLOGY Vol. 114, No. 3
Table 2. Distribution of 91 Cases of Gastric Carcinoma According to the Different Clinicopathologic Parameters and Survival Rates (3 and 5 Years) in Univariate Analysis
Category
Figure 3. High expression of dimeric S-Lex (A ) in neoplastic cells within a blood vessel and (B ) in a metastatic focus in the liver (original magnification: A, 623; and B, 1553).
rate of patients with carcinomas showing high expression of dimeric S-Lex was 0%, whereas that of patients with carcinomas with low or no expression was 34%. In univariate analysis, the other clinicopathologic features significantly associated with survival of the patients were as follows: depth of wall penetration (P 5 0.003), venous invasion (P 5 0.004), and pTNM stage (P 5 0.02). The prognostic value of venous invasion was highlighted by the univariate analysis of the survival rate of patients in each pTNM stage. The relative importance of venous invasion was particularly noticeable in the lower pTNM stages: the 5-year survival rate of patients with tumors in stages IA and IB with venous invasion (30%) was significantly worse (P 5 0.04) than that of patients with tumors in the same stages and without venous invasion (68%). The multivariate analysis of the whole series showed that pTNM stage was the strongest independent prognos-
Figure 4. Survival curves of 91 patients with gastric carcinoma (excluding postoperative deaths) according to S-Lex expression. g, Negative; d, low expression; j, high expression.
Total Age (yr ) ,40 40–65 .65 Sex Men Women Site Antrum Body Fundus Gastric stump Macroscopy Fungat./ulcerofungat. Ulcerating/ulceroinfilt. Infiltrative Laure´n’s classification Intestinal Diffuse Unclassified Carneiro et al. classification Isolated cell carcinoma Glandular carcinoma Solid carcinoma Mixed carcinoma Ming’s classification Expanding Infiltrative Lymphoid infiltration Minimal/absent Abundant/moderate Desmoplasia Minimal/absent Abundant/moderate Depth of wall penetration T1 (mucosa 1 submucosa) T2 (muscular 1 subserosa) $T3 (serosa 1 adjacent organs) Venous invasion Absent Present Lymphatic invasion Absent Present pTNM stage IA IB II IIIA IIIB IV S-Lex No expression Low expression High expression
Significance 3-Year 5-Year by generalized survival survival Savage rate (% ) rate (% ) (Mantel–Cox) test 54.6
39.6
57.1 53.7 56.9
57.1 40.3 32.2
0.35
50.7 61.3
39.8 37.4
0.53
60.4 40.0 56.0 33.3
43.7 30.0 37.9 33.3
0.52
49.3 56.6 71.4
33.0 45.7 0
0.37
49.4 57.7 71.4
34.2 50.9 42.9
0.71
72.9 65.1 66.7 29.4
56.7 40.9 50.0 29.4
0.31
59.2 50.7
38.1 41.0
0.82
48.7 63.3
39.6 40.7
0.29
54.0 55.8
38.0 43.7
0.47
80.0 70.7
60.0 34.6
0.003
32.1
27.5
74.4 40.1
53.3 29.6
0.004
62.6 49.0
51.4 28.6
0.09
75.0 70.4 64.7 45.5 16.1 50.0
75.0 46.3 41.4 32.5 16.1 0
0.02
67.0 56.1 11.0
46.4 43.8 0
0.03
NOTE. Postoperative deaths were excluded from this analysis (n 5 6).
DIMERIC S–LEx IN GASTRIC CARCINOMA
March 1998
467
Table 4. Summary of the Results Obtained With the Stepwise Multivariate Regression of Prognostic Factors With the Maximum Partial Likelihood Ratio Method in 91 Cases of the Whole Series Step factors
Log likelihood
P value
0 1 Stage (pTNM) 2 Venous invasion
2169.532 2163.868 2162.000
0.001 0.053
NOTE. Postoperative deaths were excluded from this analysis (n 5 6).
Figure 5. Survival curves of 60 patients with gastric carcinomas in stages II and III (excluding postoperative deaths) according to S-Lex expression. g, Negative; d, low expression, j, high expression.
tic factor followed by venous invasion (Tables 3 and 4). The multivariate analysis of the 60 cases in pTNM stages II and III showed that high dimeric S-Lex expression was the only independent prognostic factor influencing survival rate (Tables 5 and 6). When dimeric S-Lex expression was removed from the stepwise multivariate analysis, pTNM stages II and III stood up as the strongest independent prognostic factor followed by venous invasion (data not shown).
Discussion In the present study, we analyzed the immunohistochemical expression of dimeric S-Lex in gastric carcinomas and its relationship with the clinicopathologic features of the tumors and the survival rate of the patients. In nonneoplastic mucosa, the expression of dimeric S-Lex, when present, was only observed in the neck cells. This finding is in agreement with the previously reported Table 3. Summary of the Results Obtained With the Multivariate Analysis of Prognostic Factors With Cox’s Model in 91 Cases of the Whole Series Factors
P value
Age Sex Site Macroscopy Laure´n’s classification Carneiro et al. classification Ming’s classification Lymphoid infiltration Desmoplasia Venous invasion Stage (pTNM) S-Lex (no, low, and high expression)
0.07 0.53 0.60 0.20 0.83 0.08 0.83 0.30 0.47 0.004 0.0008 0.04
NOTE. Postoperative deaths were excluded from this analysis (n 5 6).
restricted immunoreactivity of FH6 antibody in normal adult tissues.25 In gastric carcinomas, we detected dimeric S-Lex expression in 46.4% of the cases. This rate of expression falls within the range of those previously reported in gastric carcinoma, from 25.5% to 74.1%.18,26–29 The varying expression level of dimeric S-Lex in gastric carcinoma found by different groups may reflect the heterogeneity of the series regarding the histological types of tumors and the different specificities of the antibodies. It has been reported previously that differentiated carcinomas (roughly corresponding to glandular [intestinal] carcinomas) show a higher expression of dimeric S-Lex than undifferentiated carcinomas (the latter group encompassing isolated–cell type [diffuse] carcinomas and solid carcinomas).26,28,29 Because of the use of different histological classifications, it is, however, difficult to compare the results of the different series. In our study, we found no significant relationship between dimeric S-Lex expression and the histological type of the tumors classified according to Laure´n, which was at variance with the results obtained when using the classification proposed by Carneiro et al.20: dimeric S-Lex expression was significantly greater in mixed carcinomas (encompassing Table 5. Summary of the Results Obtained With the Multivariate Analysis of Prognostic Factors With Cox’s Model in 60 Patients With Tumors in pTNM Stages II and III Factors
P value
Age Sex Site Macroscopy Laure´n’s classification Carneiro et al. classification Ming’s classification Lymphoid infiltration Desmoplasia Venous invasion Stage (pTNM) S-Lex (no, low, and high expression)
0.66 0.71 0.82 0.93 0.59 0.30 0.15 0.36 0.89 0.15 0.07 0.03
NOTE. Cases in pTNM stages I and IV (n 5 34) and postoperative deaths of patients with tumors in stages II and III (n 5 3) were excluded from this analysis.
468
AMADO ET AL.
GASTROENTEROLOGY Vol. 114, No. 3
Table 6. Summary of the Results Obtained With the Stepwise Multivariate Regression of Prognostic Factors With the Maximum Partial Likelihood Ratio Method in 60 Patients With Tumors in pTNM Stages II and III Step factors
Log likelihood
P value
0 1 S-Lex (no, low, and high expression)
2104.083 2101.779
0.03
NOTE. Cases in pTNM stages I and IV (n 5 34) and postoperative deaths of patients with tumors in stages II and III (n 5 3) were excluded from this analysis.
more than one morphological type) than in tumors with a ‘‘pure’’ histological type (glandular, isolated cell type, or solid carcinomas). We have also observed, within the group of mixed carcinomas of our series, that the prevalence of cases strongly expressing dimeric S-Lex was greater, although not significantly, in cases predominantly composed of glandular (intestinal) areas than in cases in which isolated cells were the predominant component. Interestingly, the prevalence of a high expression of dimeric S-Lex in the latter subgroup was greater than that observed in carcinomas of pure glandular (intestinal) phenotype. It remains to be seen if the high expression of dimeric S-Lex in mixed carcinomas may partly account for the reported worse prognosis of this tumor type in comparison with those of carcinomas of ‘‘pure’’ histological type, as shown previously in a larger series of gastric carcinomas.20 Interestingly, a significant correlation was found between dimeric S-Lex expression and venous invasion. To the best of our knowledge, this is the first study to show such a relationship in gastric carcinoma, and we admit that the systematic search for venous invasion that we performed in the present study, using orcein-stained sections, may have contributed to increase the accuracy of the detection of venous invasion and to highlight its relationship with dimeric S-Lex expression. However, apart from the putative effect on angiogenesis, the mechanisms through which the neoplastic cells expressing dimeric S-Lex acquire vascular invasive properties remain to be explored. Further, the relationship between dimeric S-Lex expression and vascular invasion may also reflect the ability to detect thrombi of neoplastic cells within the vessels. Dimeric S-Lex being a ligand for P-selectin, which is also expressed in platelets,30 may thus contribute to the formation of neoplastic thrombi and eventually facilitate the visualization of neoplastic cells in the vascular lumina. In keeping with the findings of Ikeda et al.29 and Nakamori et al.,18 and at variance with those obtained by Dohi et al.26 who observed a correlation between expres-
sion of dimeric S-Lex (recognized in all these studies by FH6) and depth of penetration of gastric wall, we did not observe any significant relationship between expression of dimeric S-Lex and other features of tumor aggressiveness, apart from the aforementioned vascular invasion. In the present study, high expression of dimeric S-Lex was found to be a prognostic marker. In univariate analysis, dimeric S-Lex expression was found to be significantly related to survival rate together with depth of wall penetration, venous invasion, and pTNM staging. High expression of dimeric S-Lex kept its prognostic meaning when the univariate analysis was restricted to patients with carcinomas in intermediate pTNM stages and to patients with carcinomas showing venous invasion, thus supporting its identification as a prognostic marker in gastric carcinoma. Despite the aforementioned prognostic value at univariate analysis, dimeric S-Lex expression did not keep an independent prognostic significance in the multivariate analysis of the whole series. In such analysis, only pTNM staging and venous invasion kept prognostic significance, which was in agreement with data on record.20 When the influence of pTNM staging was reduced markedly by restricting the multivariate analysis to the 60 cases with tumors in intermediate stages (pTNM stages II and III) dimeric S-Lex expression emerged as the single significant prognostic factor. The appearance of either venous invasion or dimeric S-Lex expression as independent prognostic factors in two different settings of stepwise multivariate analysis fits with the strong correlation between them. The reason for the prominence of venous invasion in the multivariate analysis of the whole series stems from the extreme importance from a prognostic standpoint of the presence of venous invasion in the lower pTNM stages. The same does not hold true when the multivariate analysis is restricted to pTNM intermediate stages. In this setting, dimeric S-Lex expression shows an advantage over venous invasion as an independent prognostic factor partly because the influence of venous invasion is substantially reduced within the group of carcinomas in intermediate stages and partly because dimeric S-Lex plays a role in prognosis per se apart from that linked to venous invasion. In fact, dimeric S-Lex antigen, as a high-affinity ligand to E-selectin,5 is thought to facilitate the extravasation from vessels of neoplastic cells expressing this antigen.4 It is tempting to suggest that this functional role of dimeric S-Lex may explain, in the intermediate pTNM stages, the significantly worse prognosis of patients harboring tumors with high expression of dimeric S-Lex through the ability to give rise to distant metastases that may have not been detected at the time of initial diagnosis. This added
March 1998
value of dimeric S-Lex is supported by the trend towards the worse prognosis of patients with carcinomas showing high dimeric S-Lex expression when the comparison is restricted to cases with venous invasion. Our findings on the relationship between the expression of dimeric S-Lex and the survival rate of patients do not fit entirely with those previously reported by Nakamori et al.18 These investigators coanalyzed the putative role of dimeric S-Lex and S-Lea as prognostic factors in a series of 137 gastric carcinomas, and in their hands, S-Lea but not dimeric S-Lex was found to be independently related to the survival rate of the patients. The discrepancy between our findings and those of Nakamori et al. may partly depend on the different scoring systems of immunoreactivity used in the two series. While Nakamori et al.18 just scored cases as negative or positive, we scored positive cases in two levels of immunoreactivity, i.e., low and high rates of expression, and we have shown that cases with low expression of dimeric S-Lex are very similar from the prognostic standpoint to cases without any expression at all and distinct from cases with high expression of dimeric S-Lex. It remains to be shown whether or not the increased biological and clinical aggressiveness of the tumors of the latter group may be ascribed to the increased availability of the dimeric S-Lex antigen at the cell surface. Altogether, the findings of the present study lead us to conclude that the clinical significance of dimeric S-Lex expression is twofold: relationship with venous invasion and relevance as prognostic factor in cases with intermediate pTNM stages. Based on this clinical significance, we think that it is important to perform the immunohistochemical detection of dimeric S-Lex in every gastric carcinoma. Whenever present, it indicates the need to search for gross vascular invasion; if there would be different therapy protocols for cases with vascular invasion and for cases without vascular invasion, cases with high dimeric S-Lex expression should be included in the former. One should also consider the possibility of including such cases within a subgroup of patients benefiting from adjuvant therapy targeted for occult distant metastases.
References 1. Hakomori S. Aberrant glycosylation in tumors and tumor-associated carbohydrate antigens. Adv Cancer Res 1989;52:257–331. 2. Dabelsteen E. Cell surface carbohydrates as prognostic markers in human carcinomas. J Pathol 1996;179:358–369. 3. Carneiro F, Amado M, David L, Clausen H, Sobrinho-Simo ˜es M. Glycosylation features of gastric carcinoma initiation and progression. A review with emphasis on simple mucin-type carbohydrates and histo-blood group antigens of the Lewis system. Eur J Cancer Prev 1994;3(Suppl 2):39–46.
DIMERIC S–LEx IN GASTRIC CARCINOMA
469
4. Varki A. Selectin ligands. Proc Natl Acad Sci USA 1994;91:7390– 7397. 5. Patel T, Goelz S, Lobb R, Parekh R. Isolation and characterization of natural protein-associated carbohydrate ligands for E-selectin. Biochemistry 1994;33:14815–14824. 6. Irimura T. Cancer metastasis determined by carbohydratemediated cell adhesion. Adv Exp Med Biol 1994;353:27–34. 7. Springer TA. Traffic signals for lymphocyte recirculation and leukocyte emigration: the multistep paradigm. Cell 1994;76:301– 314. 8. Nguyen M, Eilber F, Defrees S. Novel synthetic analogs of sialyl Lewis X can inhibit angiogenesis in vitro and in vivo. Biochem Biophys Res Commun 1996;228:716–723. 9. Tanigawa N, Amaya H, Matsumura M, Shimomatsuya T, Horiuchi T, Muraoka R, Iki M. Extent of tumor vascularization correlates with prognosis and hematogenous metastasis in gastric carcinomas. Cancer Res 1996;56:2671–2676. 10. Hoff S, Irimura T, Matsushita Y, Ota D, Cleary K, Hakomori S. Metastatic potential of colon carcinoma. Expression of ABO/Lewisrelated antigens. Arch Surg 1990;125:206–209. 11. Irimura T, Nakamori S, Matsushita Y, Taniuchi Y, Todoroki N, Tsuji T, Izumi Y, Kawamura Y, Hoff S, Cleary K, Ota D. Colorectal cancer metastasis determined by carbohydrate-mediated cell adhesion: role of sialyl-Lex antigens. Semin Cancer Biol 1993;4:319–324. 12. Ogawa J, Sano A, Koide S, Shohtsu A. Relation between recurrence and expression of proliferating cell nuclear antigen, sialyl Lewisx and sialyl Lewisa in lung cancer. J Thorac Cardiovasc Surg 1994;108:329–336. 13. Inufusa H, Kojima N, Yasutomi M, Hakomori S. Human lung adenocarcinoma cell lines with different lung colonization potential (LCP), and a correlation between expression of sialosyl dimeric Le (x) (defined by Mab FH6) and LCP. Clin Exp Metastasis 1991;9:245–257. 14. Nakamori S, Kameyama M, Imaoka S, Furukawa H, Ishikawa O, Sasaki Y, Kabuto T, Iwanaga T, Matsushita Y, Irimura T. Increased expression of sialyl Lewisx antigen correlates with poor survival in patients with colorectal carcinoma: clinicopathological and immunohistochemical study. Cancer Res 1993;53:3632–3637. 15. Narita T, Funahashi H, Satoh Y, Watanabe T, Sakamoto J, Takagi H. Association of expression of blood group-related carbohydrate antigens with prognosis in breast cancer. Cancer 1993;71:3044– 3053. 16. Jørgensen T, Berner A, Kaalhus O, Tveter K, Danielsen H, Bryne M. Up-regulation of the oligosaccharide Sialyl Lewisx: a new prognostic parameter in metastatic prostate cancer. Cancer Res 1995;55:1817–1819. 17. Nakayama T, Watanabe M, Katsumata T, Teramoto T, Kitajima M. Expression of sialyl Lewisa as a new prognostic factor for patients with advanced colorectal carcinoma. Cancer 1995;75:2051– 2056. 18. Nakamori S, Furukawa H, Hiratsuka M, Iwanaga T, Imaoka S, Ishikawa O, Kabuto T, Sasaki Y, Kameyama M, Ishiguro S, Irimura T. Expression of carbohydrate antigen sialyl Lea: a new functional prognostic factor in gastric cancer. J Clin Oncol 1997;15:816– 825. 19. Laure´n P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histoclinical classification. Acta Pathol Microbiol Scand 1965;64: 31–49. 20. Carneiro F, Seixas M, Sobrinho-Simo ˜es M. New elements for an updated classification of the carcinomas of the stomach. Pathol Res Pract 1995;191:571–584. 21. Ming S-C. Gastric carcinoma: a pathobiological classification. Cancer 1977;39:2475–2485. 22. Fukushi Y, Nudelman E, Levery SB, Hakomori S, Rauvala H. Novel fucolipids accumulating in human adenocarcinoma: III a hybridoma antibody (FH6) defining a human cancer-associated difuco-
470
23.
24. 25.
26.
27.
28.
AMADO ET AL.
ganglioside (VI3NeuAcV3III3Fuc2nLc 6). J Biol Chem 1984;259: 10511–10517. Hsu SM, Faine L, Fanger H. A comparative study of the peroxidaseantiperoxidase method and the avidin-biotin complex method for studying polypeptide hormones with radioimmunoassay antibodies. Am J Clin Pathol 1981;75:734–738. Cox DR. Regression models and life tables. J R Statist Soc 1972;34:187–202. Fukushi Y, Kannagi R, Hakomori S, Shepard T, Kulander B, Singer J. Location and distribution of difucoganglioside (VI3NeuAcV3III3Fuc2nLc6) in normal and tumor tissues defined by its monoclonal antibody FH6. Cancer Res 1985;45:3711–3717. Dohi T, Abe K, Ohshiba S. Immunohistochemical study of carbohydrate antigen expression in gastric carcinoma. Gastroenterol Jpn 1989;24:239–245. Sakamoto J, Watanabe T, Tokumaru T, Takagi H, Nakazato H, Lloyd KO. Expression of Lewisa, Lewisb, Lewisx, Lewisy, sialyl-Lewisa and sialyl-Lewisx blood group antigens in human gastric carcinoma and in normal gastric tissue. Cancer Res 1989;49:745–752. Murata K, Egami H, Shibata Y, Sakamoto K, Misumi A, Ogawa M.
GASTROENTEROLOGY Vol. 114, No. 3
Expression of blood group-related antigens, ABH, Lewisa, Lewisb, Lewisx, Lewisy, CA 19-9 and CSLEX1 in early cancer, intestinal metaplasia, and uninvolved mucosa of the stomach. Am J Clin Pathol 1992;98:67–75. 29. Ikeda Y, Mori M, Kajiyama K, Haraguchi Y, Sasaki O, Sugimachi K. Immunohistochemical expression of sialyl Tn, sialyl Lewisa, sialyl Lewisa-b-, and sialyl Lewisx in primary tumor and metastatic lymph nodes in human gastric cancer. J Surg Oncol 1996;62:171–176. 30. Palabrica T, Lobb R, Furie B, Aronovitz M, Benjamin C, Hsu Y-M, Sajer S, Furie B. Leukocyte accumulation promoting fibrin deposition is mediated in vivo by P-selectin on adherent platelets. Nature 1992;359:848–851.
Received August 13, 1997. Accepted November 18, 1997. Address requests for reprints to: Margarida Amado, Ph.D., Institute of Molecular Pathology and Immunology of University of Porto, IPATIMUP, Rua Dr. Roberto Frias s/n, 4200 Porto, Portugal. Fax: (351) 2-557-0799. Supported by project 2/2.1/SAU/1356/95 from PRAXIS XXI.