1836

Predictive Value of Preoperative Serum Sialyl Tn Antigen Levels in Prognosis of Patients with Gastric Cancer

lkuo Takahashi, M.D.,* Yoshihiko Maehara, M.D.,t Tefsuya Kusumoto, M.D.,* Motofumi Yoshida, M.D.," Yoshihiro Kakeji, M.D.,$. Hiroki Kusumoto, M.D.,* Motonosuke Furusawa, M.D.,$. and Keizo Sugimachi, M.D., F.A.C.S.*,t

Background. Expression of sialyl Tn antigen (STN) or serum STN levels were reported to be the independent prognostic factors of colon and ovarian cancers, respec- tively. The clinical significance of serum STN was evalu- ated as a tumor marker in gastric cancer.

Methods. Preoperative STN levels were examined in 350 patients with gastric cancer, surgically treated in the Department of Surgery 11, Kyushu University Hospital, and in the National Kyushu Cancer Center, between April, 1981 and April, 1986. The patients were divided into two groups: a high STN group (n = 304), with over 45 U/ml of STN, and a low STN group (n = 46), with less than or equal to 45 U/ml.

Results. In the high STN group, the patients were older and the tumors were larger, compared to the low STN group. Histologically, tumors in the high STN group were deeply penetrating and the rates of lymphatic in- volvement, vascular involvement, and lymph node and hepatic metastases were higher. Infiltrative growth pat- terns dominated. Thus, the tumors were in an advanced stage. The 5-year survival rate for patients in the high STN group was significantly less than that of patients in the low STN group (44.8% ? 7.9% versus 75.1% ? 2.6%, P < 0.05).

Conclusions. Because a high STN level correlates with an advanced tumor stage and a poorer prognosis, close follow-up and aggressive therapy are recom- mended. Cancer 1993: 72:1836-40.

From the *Cancer Center of Kyushu University Hospital, the ?Department of Surgery 11, Faculty of Medicine, Kyushu University, and the $Clinical Research Division, National Kyushu Cancer Center, Fukuoka, Japan.

Supported in part by Grants-in-Aid from the Uehara Memorial Foundation.

Address for reprints: Yoshihiko Maehara, M.D., Department of Surgery 11, Kyushu University, 3-1 - I Maidashi, Higashi-Ku, Fukuoka 81 2, Japan.

Accepted for publication April 22, 1993.

Key words: tumor marker, serum sialyl Tn antigen, gas- tric cancer, prognosis.

An ideal tumor marker should signal the presence of tumor and also define the site of the lesion and its mor- phologic type. In gastrointestinal tract malignancies, carcinoembryonic antigen (CEA) and carbohydrate an- tigen 19-9 (CA 19-9) are used commonly for postopera- tive follow-up or detection of re~urrence.'-~ Shimizu et al.' reported that in patients with gastric cancer, the preoperative and periodic postoperative assay of CEA levels has predictive value in determining cancer stage and prognosis. Our group' reported that CEA is useful for monitoring the postoperative course of well differ- entiated gastric cancer, but is not useful in cases of poorly differentiated gastric cancer. Maeta et aL5 exam- ined CA 19-9 in tissue and sera from gastric cancer pa- tients and evaluated the clinical significance of this marker as a detector, monitor, or tumor-associated anti- gen of gastric cancer. Markers currently available, how- ever, do not have sufficient sensitivity or specificity. The positive rates of elevated preoperative CEA in pa- tients with gastric cancer were reported to be 18.7%' and 2 5 O / 0 . ~ The rates of elevated CA 19-9 in gastric cancer were reported to be 27.5%,' 38%,6 or 52y0,~ but CA 19-9 is inferior in sensitivity to CEA.6 Thus, to ob- tain more information leading to a diagnosis, the combi- nation of these markers or the evaluation of other markers is r e q ~ i r e d . ~ - ~

Incomplete synthesis of glycolipids or glycopro- teins in cell membranes often occurs in cancer cell^.'^^^' Sialyl Tn antigen (STN) is one of these abnormal 0- linked glycoproteins. The Tn antigen is a precursor of the Thomsen-Fridenreich antigen, the T antigen, and STN is a sialic acid-bound Tn antigen."-" Immunohis- tologic expression of sialyl Tn or elevated serum STN

Predictive Value of STN in Gastric Cancer/Takahashi et al. 1837

levels have been reported to be prognostic factors in

We report here the predictive value of preoperative STN levels, and the relationship between the clinico- pathologic features and preoperative STN level in pa- tients with gastric cancer.

and ovarian cancers.I4

Patients and Methods

From April, 1981 to April, 1986, 350 Japanese patients with primary gastric cancer were surgically treated in the Department of Surgery 11, Faculty of Medicine, Kyushu University and at the National Kyushu Cancer Center, Fukuoka, Japan, and the serum STN levels in these patients were determined. In these patients, there was no evidence of another malignancy, and no patient had been given preoperative treatment with anticancer drugs. The pathologic diagnoses and classifications for each patient were carried out according to the General Rules for the Gastric Cancer Study in Surgery and Pa- thology in

The serum STN levels were measured using a one- step radioimmunoassay kit (S-Tn Otsuka; Otsuka As- say Laboratories, Tokushima, Japan).17 Competitive binding to the radiolabeled monoclonal antibody TKH- 2 between serum STN and STN-coated beads was used (an immunoradiometric competitive inhibition assay)." Serum was obtained from venous blood after an over- night fast. The blood samples were separated immedi- ately by centrifugation and placed in liquid nitrogen until this assay (October, 1991). The cutoff line between normal and elevated STN titers was set at 45 U/ml. This cutoff line, 45 U/ml, is the mean f 2 standard deviations (SD) of the STN value in the normal volun- teer, as reported by Imura et al.I7 We classified the pa- tients into two groups: a high STN group, with serum STN over 45 U/ml, and a low STN group, with 5.0 to 45.0 U/ml. This assay had a lower limit of detection of 5.0 U/ml serum STN.

Clinicopathologic data were stored in an IBM (Ar- monk, NY) 4381 mainframe computer. The Biomedical Computer Program (BMDP Statistical Package Pro- gram, Los Angeles, CA) was used for all statistical analy- ses.18 Data were analyzed using the chi-square test and the Mann-Whitney test to compare data on patients in the high group with those in the low group. For these analyses, the BMDP P4F and P3F programs were used. Survival curves were calculated by the Kaplan-Meier method, using the BMDP P1L program. Comparisons between the two groups were performed using the gen- eralized Wilcoxon test to analyze equality of the sur- vival curves. A P value of less than 0.05 was considered to be statistically significant. In the statistical analysis, deaths due to causes other than gastric carcinoma were

considered censored cases. Unknown data also were excluded from the statistical analyses.

Results

Clinicopathologic Factors

Clinicopathologic data on the 350 patients are listed in Table 1. Of these patients, 46 (13%) had a high STN level. Significant differences between the two groups existed with respect to the following factors: mean age (P < 0.01), maximum tumor diameter (P < 0.01), stage (P < 0.01), serosal invasion (P < 0.01), lymphatic in- volvement (P < 0.01), vascular involvement (P < 0.01), histologic growth pattern, lymph node metastasis (P < 0.01), and frequency of liver metastasis (P < 0.01). In the high STN group, patients were significantly older and the tumors were larger. Histologically, tumors were more likely to have penetrated deeply, and positive rates of lymphatic involvement, vascular involvement, and lymph node metastasis were higher. Tumors tended to show an infiltrative growth pattern, and liver metastasis was more frequent. Tumors in the high STN group were in a more advanced stage.

STN levels were compared between high and low STN groups, according to cancer stage (Table 2). With stage progression, the STN level in the high STN group increased (Fig. 1).

Survival Rates

No patient was lost to follow-up. The mean follow-up duration (+ SD) at the time of analysis (November, 1991) was 6.07 * 0.92 years. Two hundred eight of the 350 patients were living.

The postoperative survival curves between the high STN group and the low STN group were com- pared (Fig. 2). The 5-year survival rate was 75.1% f 2.6% in the low STN group, and 44.8% k 7.9% in the high STN group. Survival rates between the two groups significantly differed (P < 0.05).

Discussion

0-linked oligosaccharides were found to be autocarcin- ogens." Mucins, the major component of mucus, are high-molecular-weight glycoproteins containing oligo- saccharide side chains connected by 0-glycosidic link- age to serine or threonine residues in the apomucin backbone. The addition of N-acethylgalactosamine to apomucin results in the formation of Tn antigen, and the addition of sialic acid forms STN." In cancer cells, glycosylation often is incomplete, resulting in a storage

1838 CANCER September 25, 2993, Volume 72, No. 6

Table 1. Clinicopathologic Features of Tumors in Relationship to Serum Sialyl-Tn Level Serum STN level (U/ml) Serum STN level (U/ml)

4 45 (Yo) > 45 (Yo) 4 45 (Yo) > 45 (Yo) Variable (n = 304) (n = 46) P values Variable (n = 304) (n = 46) P value*

Age (yr) Sex

Male Female

Maximum tumor diameter

Tumor location (cm)

Upper Middle Lower

Stage I I1 I11 IV

Negative Positive

Histologic type Differentiated Undifferentiated Other

Negative Positive Unknown 11

Negative Positive Unknown11

Serosal invasion

Lymphatic involvement

Vascular involvement

60.1 f 11.95

205 (67.4) 99 (32.6)

5.4 * 3.75

52 (17.1) 102 (33.6) 150 (49.3)

153 (50.3) 32 (10.5) 61 (20.1) 58 (19.1)

201 (66.1) 103 (33.9)

193 (63.5) 105 (34.5)

6 (2.0)

107 (35.2) 196 (64.5)

l (0 .3)

190 (62.5) 111 (36.5)

3 (0.7)

65.8 ? 9.35 < 0.01 NS

34 (73.9) 12 (26.1)

8.0 f 3.95 < 0.01 NS

4 (8.7) 18 (39.1) 24 (52.2)

9 (19.6)

12 (26.1) 22 (47.8)

17 (37.0) 29 (63.0)

28 (60.9) 18 (39.1)

< 0.01

3 (9.5)

< 0.01

NS

0 (0.0) < 0.01

7 (15.2) 39 (84.8)

0 (0.0) < 0.01

13 (28.3) 33 (71.7)

0 (0.0)

Histologic growth pattern Expansive Intermediate Infiltrative Unknown11

Histologic lymph node metastasis

Negative Positive

Negative Positive

Negative Positive

Partial Total

R1 R2 and R 3

Curative Noncurative Unknown11

Peritoneal dissemination

Hepatic metastasis

Gastric resection

Lymph node dissection7

Curability

131 (43.1) 117 (38.5) 53 (17.4)

3 (1.0)

169 (55.6) 135 (44.4)

279 (91.8) 25 (8.2)

291 (95.7) 13 (4.3)

227 (74.7) 77 (25.3)

49 (16.1) 255 (83.9)

249 (81.9) 54 (17.8)

l(0.3)

11 (23.9) 17 (37.0) 18 (39.1) 0 (0.0)

9 (19.6) 37 (80.4)

41 (89.1) 5 (10.9)

39 (84.8) 7 (15.2)

30 (65.2) 16 (34.8)

14 (30.4) 32 (69.6)

29 (63.0) 17 (37.0) 0 (0.0)

< 0.01

< 0.01

NS

< 0.01

NS

NS

NS

STN: sialyl-Tn; NS: not significant; R1: gastric resection including the complete removal of group 1 lymph nodes alone; R2: gastric resection including the complete removal of group 1 and 2 lymph nodes; R3: gastric resection including the complete removal of group 1, 2, and 3 lymph nodes. $ Based on Mann-Whitney U test or chi-square test. 5 Mean rL- standard deviation. 11 Unknown data were excluded in the comparative analysis and survival analysis. B According to the General Rules for the Gastric Cancer Study in Surgery and Pathology in ]apan.l5

of precursor structures that, in normal cells, would re- main cryptic because of further elongation.

In patients with colon or ovarian cancer, serum STN levels or the immunohistologic expression of STN were found to be related to clinicopathologic features and to the progn~sis. '~, '~, '~ Itzkowitz et al.13 found in immunohistologic studies that normal colonic mucosa did not express Tn, sialyl-Tn, or T antigens, whereas colon cancer cells highly expressed these antigens. They analyzed data on 137 patients with colon cancer who underwent surgical resection, and reported that the 5- year survival rate for patients with STN-negative ver- sus STN-positive tumors was 100% versus 73% ( P < 0.05, Mantel-Cox analysis). Multivariate regression

analysis revealed that STN expression ( P < 0.05) and tumor ploidy (P < 0.001) were the two independent prognostic factors for patients with colon cancer. Inoue et al.19 reported that serum STN was significantly ele- vated in patients with ovarian cancer. This antigen level correlated well with response to treatment, and they concluded that it is useful for monitoring patients and for evaluating the effect of therapy. Kobayashi et al.I4 reported that the 5-year survival of patients with ovar- ian cancer with low serum STN (STN < 50) versus those with high STN (STN > 50) was 76.9% and 10.8%, respectively, and a multivariate analysis showed that serum STN is an independent prognostic factor.

Mofoo et a1." reported in their analysis of 257 pa-

Predictive Value of STN in Gastric CancerlTakahashi et al . 1839

Table 2. Comparison of Sialyl-Tn (STN) Levels Between High STN Group and Low STN Group According to Tumor Stage

STN levels (no. of group)

Low STN group High STN group Variable (STN 5 45 U / m l ) (STN > 45 U/ml i

Total Stage

I 11 I11 IV

24.2 f 7.8 (n = 304)

24.4 k 8.1 (n = 153) 24.1 1 7.1 (n = 32) 23.7 f 7.9 (n = 61) 25.6 f 8.1 (n = 58)

321.1 f 606.5 (n = 46)*

71.7 f 52.8 (n = 9)* 165.4 f 163.7 (n = 3) 226.1 k 291.5 (n = 12)* 496.2 f 820.1 (n = 22)*-t

* P < 0.05 compared with low STN group. t P < 0.05 comuared with Stage 1 high STN group.

tients and 121 normal control subjects that the positive rate (STN > 45) in patients with gastric cancer (n = 45) was 28.1%, and that the positive rate of serum STN increases as the clinical stage progresses, but the clini- copathologic features of these patients were not given. We have no documentation of the relationship between clinicopathologic features of patients with gastric cancer and STN levels.

As shown in Table 1, the high STN group had tu- mors of more advanced stage, and the prognosis of this group was poorer than that of the low STN group. Be- cause this study was restricted to patients who under- went surgical resection, the positive rate of 13% in this study is lower than that reported by Motoo et a1.” If unresectable cases, which usually are in an advanced stage, had been included, the positive rate would be higher.

I 100,r

L

n - E 3

f 500- v)

v

0 3 6 1 7 0 1405 N N

t 0

0

0

Iv 0 stage I II III

Figure 1. STN levels in each of stage for the high STN group. Data on STN levels in Stages I-IV are included.

‘ O 0 1

i

I I 0 1 2 3 4 5

Time after operation (years) Figure 2. Survival curve for patients with gastric cancer with high or low STN levels. There were 46 patients in the high STN group (light line), and 304 patients in the low STN group (dark line). Survival time between the two groups was significantly different (P < 0.05).

Depth of penetration, invasion of the duodenum, macroscopic appearance, intramural and extramural growth, pattern of the lesion, lymphatic and vascular invasion, stage, and lymph node metastasis are well known independent prognostic factors in patients with gastric cancer.21,22 Because significant differences in the clinicopathologic features between two groups existed with respect to mean age, maximum tumor diameter, stage, serosal invasion, lymphatic involvement, vascu- lar involvement, histologic growth pattern, histologic lymph node metastasis, lymph node metastasis, and he- patic metastasis, as shown in Table 1, patients with a high STN will have a poor prognosis.

In conclusion, the preoperative STN level is valu- able for predicting the prognosis of patients with gastric cancers. Aggressive therapy and close postoperative follow-up are mandatory for patients with high STN levels.

References

Shimizu N, Wakatsuki T, Murakami A, Yoshioka H, Hamazoe R, Kanayama H, et al. Carcinoembryonic antigen in gastric cancer patients. Oncology 1987; 44:240-4. Maehara Y, Sugimachi K, Akagi M, Kakegawa T, Shimazu H, Tomita M. Serum carcinoembryonic antigen level increases cor- relate with tumor progression in patients with differentiated gastric carcinoma following noncurative resection. Cancer Res

Koga T, Kano T, Souda K, Oka N, Inokuchi K. The clinical useful- ness of preoperative CEA determination in gastric cancer. ]pn ! Surg 1987; 17:342-7. Tamada R, Hiramoto Y, Abe Y, Nouzuka T, Okamura T, Ma- suda H, et al. Serial determinations of carcinoembryonic antigen

1990; 50~3952-5.

1840 CANCER September 25, 2993, Volume 72, No. 6

5.

6.

7.

8.

9.

10.

11.

12.

for early detection of recurrent gastric cancer. ] p n ] Surg 1982;

Maeta M, Yoshioka H, Shimizu T, Murakami A, Hamazoe R, Koga S. Carbohydrate antigen 19-9 in tissues and sera from patients with gastric cancer. Oncology 1990; 47:229-33. Nishida K, Yamamoto T, Yoshioka K, Iida M, Miyagawa H, Yo- shikawa T, et al. Assay of serum carbohydrate antigen (CA) 19-9 in the diagnosis of gastric cancer [in Japanese with English ab- stract]. ]pn ]Cancer Chemother 1986; 13:lOl-8. Heptner G, Domschke S, Domschke W. Comparison of CA 72-4 with CA 19-9 and carcinoembryonic antigen in the serodiagnos- tics of gastrointestinal malignancies. Scand ] Gastroenterol 1989; 24:745-50. Quentmeier A, Schlag P, Geisen HP, Schmidt-Gayk H. Evalua- tion of Ca 12-5 as a tumor marker for gastric and colorectal cancer in comparison to CEA and Ca 19-9. Eur I Surg Oncol

Kornek G, Depisch D, Temsch EM, Scheithauer W. Compara- tive analysis of cancer-associated antigen CA-195, CA 19-9 and carcinoembryonic antigen in diagnosis, follow-up and monitor- ing of response to chemotherapy in patients with gastrointesti- nal cancer. ] Cancer Res Clin Oncol 1991; 117:493-6. Springer GF. T and Tn, general carcinoma autoantigens. Science 1984; 224:1198-1206. Hakomori S, Kannagi R. Glycosphingolipids as tumor-asso- ciated and differentiation markers. I N a f l Cancer Znst 1983; 71931-51. Kjeldsen T, Clausen H, Hirohashi S, Ogawa T, Iijima H, Hako- mori S. Preparation and characterization of monoclonal antibod- ies directed to the tumor-associated 0-linked sialosyl-2-6-N- acetylgalactosaminyl (sialosyl-Tn) epitope. Cancer Res 1988;

121429-33.

1987; 13:197-201.

48:2214-20.

13

14

15

16

17

18

19

20

21

22

Itzkowitz SH, Bloom EJ, Kokal WA, Modin G, Hakomori S, Kim YS. A novel mucin antigen associated with prognosis in colorec- tal cancer patients. Cancer 1990; 66:1960-6. Kobayashi H, Terao T, Kawashima Y. Circulating serum STN antigen as a prognostic marker in epithelial ovarian cancer [in Japanese with English abstract]. ] p n ] Cancer Chemother 1991;

Japanese Research Society for Gastric Cancer. The general rules for the gastric cancer study in surgery and pathology: Part I. clinical classification. ] p n ] Surg 1981; 11:127-39. Japanese Research Society for Gastric Cancer. The general rules for the gastric cancer study in surgery and pathology: Part 11. histological classification of gastric cancer. ]pn ] Surg 1981;

Imura H, Mori T, Ohkura H, Ishii M, Ariyoshi H, Endo J, et al. Basic and clinical evaluation of an immunoradiometric competi- tive inhibition assay for sialyl Tn antigen [in Japanese with En- glish abstract]. ]pn ] Cancer Chemother 1989; 16: 3213-9. Dixon WJ. BMDP statistical software. Berkeley, California: Uni- versity of California Press, 1988229-744. Inoue M, Ogawa H, Nakanishi K, Tanizawa 0, Karino K, Endo J. Clinical value of sialyl Tn antigen in patients with gynecologic tumor. Obstet Gynecol 1990; 75:1032-6. Motoo Y, Kawakami H, Watanabe H, Satomura Y, Ohta H, Okai T, et al. Serum sialyl-Tn antigen level in patients with digestive cancers. Oncology 1991; 48:321-6. Okada M, Kojima S, Murakami M, Fuchigami T, Yao T, Omae T, et al. Human gastric carcinoma: prognosis in relation to macro- scopic and microscopic features of the primary tumor. Natl Cancer [nst 1983; 71:257-9. Baba H, Korenaga D, Okamura T, Sugimachi K. Prognostic fac- tors in gastric cancer with serosal invasion: univariate and multi- variate analyses. Arch Surg 1989; 124:1061-4.

18: 1651-5.

11: 140-5.