ORIGINAL ARTICLE 
Annals of Nuclear Medicine Vol.9, No.1, 7-13, 1995 

Concentration and distribution of tumor associated antigens TAG-72 and CEA in stomach cancer 

June-Key CHUNG,* Myung-Chul LEE,* Hong-Keun CHUNG,** Sang-Moo LIM,*** 
Ja-June JANG**** and Chang-Soon KOH* 

*Department of Internal Medicine, **Department of Biochemistry, ****Department of Pathology, 
Seoul National University, College of Medicine, Seoul Korea 
***Department of Nuclear Medicine, Korea Cancer Center Hospital, Seoul, Korea 

We measured the concentration and distribution of tumor associated antigens, TAG-72 and CEA, in stomach cancer by in vitro quantitative autoradiography (IV-QAR). Frozen sections of 33 specimens were incubated with varying concentrations of 125I-1abeled CEA-79.1 and B72.3 antibodies specific for carcinoembryonic antigen (CEA) and tumor-associated glycoprotein-72 (TAG-72), respectively. Computer analysis of specific antibody binding gave maximal binding 
values which were equal to the concentrations of the antigen or epitope. TAG-72 was detected in 25 specimens, at a concentration ranging from 8.4 to 562.9 pmol/g. CEA was detected in 32 of the 33 specimens and its concentration ranged from 8.8 to 525.3 pmol/g. The distribution of TAG-72 by IV-QAR coincided with that of the tumor cells in 41.4% of the pathologic lesions. The distribution of CEA coincided with the tumor cells in 80.5% of pathologic lesions, nearly twice the TAG-72. The concentration of TAG-72 was significantly higher in mucinous adenocarcinoma and 
mucin containing adenocarcinomas than other types of adenocarcinomas. There was no significant difference in the concentration of CEA among the pathologic types of stomach cancer. In summary, stomach cancer exhibited wide variations in TAG-72 and CEA expression. CEA expression was more frequent and homogeneous than TAG-72. 

Key words: carcinoembryonic antigen, stomach cancer tumor-associated glycoprotein-72, autoradiography, 

INTRODUCTION 
STOMACH CANCER is still the most frequent type of malignant tumor in East Asia. Europe and South America. I The only acceptable management of stomach cancer is early diagnosis and surgery.2,3 Recently, radioimmunoscintigraphy has been used as an investigative tool to detect malignant tumors.4,5 Monoclonal antibodies with specificity for antigens on stomach cancer cells and colon cancer cells are increasingly being tried as specific probes for use in radioimmunoscintigraphy and radioimmunotherapy.6,7 Especially B72.3, which is specific for tumor-associated glycoprotein-72 (TAG-72), and anti-carcino-embryonic antigen (CEA) antibodies have been used widely in the radioimmunoscintigraphy of gastrointestinal carcinomas.7-11 One of the critical factors in successful radioimmunoscintigraphy and immunotherapy is the local concentration and distribution of the target antigen, 12 but a few reports investigated the concentration of these antigens in stomach cancer. 

Immunohistochemical methods have been used to analyze the presence and distribution of various antigens in tumors, but immunohistochemistry alone is limited in its use as an objective, sensitive and quantitative diagnostic tool. We previously reported the use of in vitro quantitative autoradiography for measuring tumor associated antigens in histologic sections.13,14 This study evaluates the concentration and distribution pattern of the tumor associated antigens, TAG-72 and CEA, in human stomach cancer.

MATERIALS AND METHODS 

Cancer tissues 
Thirty-three specimens of stomach cancer were evaluated. All specimens were from cases of advanced stomach cancer and were obtained by surgical resection. 

Monoclonal antibodies 
Two monoclonal antibodies (MOAb) to tumor associated antigens were used in the autoradiography. MoAb B72.3 (IgG1), specific for TAG-72 glycoprotein,15 was supplied by Dr. Schlom of National Cancer Institute (Bethesda, MD). MoAb CEA-79.1 (IgG2.), was made at Seoul National University.16 This antibody was specific for CEA, and did not bind nonspecific cross-reacting antigens.17 Both antigens are located on the cell surface. These monoclonal antibodies were labeled with 1251 by the chloramine T method. One mg of antibody was reacted with 37 MBq (1 mCi) of 1251 (New England Nuclear, Boston, MA) by adding 12.5 ug of chloramine T and the reaction was stopped by adding 43.8 ug of sodium thiosulfate. The reaction time was 2 minutes. A PD- 10 column (Pharmacia, Piscataway, NJ) was used to separate the radiolabeled antibody from free 1251. The specific activity was around 1.0 mCi/mg. More than 90% of the purified radioactivity was precipitable with 10% trichloroacetic acid. 

The immunoreactivity of 125I-labeled B72.3 was determined by solid-phase radioimmunoassay with extracts of LS 174T cell, which expressed TAG-72.18 The immunoreactivity of the 125I-1abeled CEA-79. I was determined by a serial cell binding assay. SNU-C4 human colon cancer cells, which expressed CEA, were reacted with radiolabeled antibody, and immunoreactivity was calculated by the double inverse plot. 19 The immunoreactivity of 125I-CEA-79.1 ranged from 44.6% to 74.4%. In the case of 125I-B72.3, the immunoreactivity ranged from 51.2% to 73.6%. 

In vitro quantitative autoradiography 
The tumor specimens were cut into 20 micron frozen sections with a cryomicrotome. The tissue sections of each tumor were divided into two groups:1 ) total binding of specific antibody, and 2) the nonsaturable binding of specific antibody. Sections were fixed in 0.25% glutaraldehyde for 20 minutes. For the nonsaturable binding group, sections were incubated with unlabeled antibody (B72.3 or CEA-79.1) for 30 minutes. All sections were then incubated for 30 minutes in a solution containing 2% bovine serum albumin (BSA) and 10% chicken serum (CS) in phosphate buffer saline (PBS) to reduce the nonspecific binding of the radiolabeled antibody. The individual sections were then incubated for 60 minutes in a solution of 125I-labeled antibody at a concentration ranging from 1.3 nmol/liter to 83.3 nmol/liter. After this incubation, the slides were washed in PBS and dehydrated with ethanol. 

The autoradiographic standard was prepared with 125I-human serum albumin, as described by Mies.20 The 2 sets of tissue sections were exposed to Kodak SB5 film with standard for two days before they were developed. The autoradiographic films were digitized with a scanning microdensitometer (Amersham, Arlington Heights, IL). The optical density measured from the 125I-labeled human serum albumin standard was plotted against the specific activity of each standard. A polynomial fitting of these data provided a standard curve. We compared the autoradiographic image to a hematoxylin and eosin (H & E) stained adjacent tissue section in order to define the region of interest in the autoradiographic image. The mean optical density of the selected region was obtained and the concentration of radiolabeled antibody in the tumor (uCi/g) was determined from the standard curve. This specific activity was corrected for its immunoreactivity. We calculated the concentration of binding antibody as moles of antibody per grams of tumor (mol/g) from the specific activity value for each 125I-labeled antibody preparation. Overlying the autoradiographic and H & E section images also allowed us to correlate the distribution of bound radioactivity, i.e., antigen, with tumor cell distribution . 

Analysis of binding activity data 
The data from the saturation studies were analyzed by the Nonlinear Least Square Fitting method. We used two equations to analyze the saturation curve and obtained values for the dissociation constant (Kd) and the maximum binding of the antibody (Bmax) (Fig.1). 
The first equation was a saturation plot. 
Total Binding = Bmax x [Ab] /Kd + [Ab] + a x [Ab]  
Nonsaturable Binding = a x [Ab]

The Bmax is the maximal amount of 125I-labeled antibody which binds specifically to the tissue; [Ab] is the concentration of 125I-labeled antibody in the incubation media; Kd is the antigen/antibody dissociation constant, and a is the slope of the nonsaturable curve. Specific binding was obtained by subtracting the nonsaturable binding curve from the total binding curve. 

The second equation was a double inverse plot between the specific binding data and the concentration of labeled antibody. 

l/Specific binding = l/Bmax + Kd/(Bmax x [Ab]) 

A plot of 1/specific binding versus 1/[Ab] will yield a straight line with the value at the y-intercept being equal to l/Bmax A computer fitting of the curve gave the Bmax, which is the maximal concentration of 125I-labeled antibody specifically bound to the tumor. The specific activity of the 125I-labeled antibody was used to calculate the maximal pmol/g of the antibody bound to the tumor. The concentration of bound antibody was considered to be equivalent to the concentration of the antigenic epitope. The curve fit and binding calculations were made by means of the RS-1 curve modeling program in this equation. 

Statistical analysis 
Differences between pathologic types in the concentration of antigen were tested for their significance by Student's-t test. The rates of coincidence between antigen expression and cellular distribution were compared by x2-test. 

RESULTS 
Table 1 shows the results of the quantitative autoradiographic analysis of the tumor sections from stomach cancer specimens. The concentration of TAG-72 could be measured in 25 specimens (75.7%) and was undetectable in 8 specimens (24.3%). CEA was detectable in all specimens except one (96.9%). Out of 40 pathologic lesions, TAG-72 was expressed in 29 (72.5%), and CEA was expressed in 90% of the lesions. The median value of TAG-72 was about one third the concentration (27.9 pmol/g) of that of CEA ( 101.7 pmol/g). The concentration of TAG-72 ranged from 8.4 pmol/g to 562.9 pmol/g (Fig. 2A), and the concentration of CEA ranged from 8.8 pmol/g to 525.3 pmol/g (Fig. 2B). 

The concentration of TAG-72 and CEA in the tumor sections was compared to the tumor's pathologic type (Table 2). The concentration of TAG-72 was higher in mucinous adenocarcinoma than in other types, but the concentration of CEA did not vary greatly between the various types of tumors. 

We compared the antigen levels of the various tumors to the presence or absence of mucin, which was determined by H & E stain. Carcinomas with mucinous components had significantly (p < 0.001) higher TAG-72 concentrations than carcinomas without mucinous components (Table 3). However, the CEA containing tumors showed no difference between mucin containing and noncontaining adenocarcinomas. In our study, the poorly differentiated carcinoma without a mucinous component was the most frequent type (50%). In this type of tumor, the concentration of CEA was significantly (p < 0.005) greater than the concentration of TAG-72. 

The pattern of the tumor cell distribution, as found in H & E sections, was compared to the pattern of antigen expression in the autoradiographic images . Figure 3 shows a typical example from our study of a case of mucinous adenocarcinoma with diffuse cellular distribution. CEA expression was homogeneous through the tumor cells, but TAG-72 expression was heterogeneous. The pattern of TAG-72 antigen expression coincided with the tumor cell distribution in 41.4% of the cases. There were also significant numbers of tumor cells that expressed a low or negligible amount of TAG-72 antigen. The patterns of CEA antigen expression coincided with tumor cellular distribution in 80.5% of the cases; in only 19.5% CEA expression was heterogeneous (Table 4). 

DISCUSSION 
Radioimmunoscintigraphy and radioimmunotherapy have been studied as new methods to diagnose and treat malignancies. Gastrointestinal carcinomas express a variety of tumor associated antigens including CEA, TAG-72, GICA and DuPan-2.7,9,10 In this study, we measured the concentration and distribution of CEA and TAG-72 in stomach cancer. CEA has been described as a Mr. 180,000 glycoprotein complex which is expressed by embryonic colonic mucosa and carcinomas of the gastrointestinal tract. Monoclonal antibody CEA-79. I was made from CEA, purified from the culture supernatant of a human colon cancer cell line, LS174T.16 Antibody B72.3 is a murine lgG1, which was prepared from a membrane-enriched extract of human metastatic breast carcinoma as the antigen, and B72.3 recognizes the Mr > 10^6 novel tumor-associated glycoprotein, TAG-72. 15 

It is well known that CEA is expressed frequently in gastric cancer.21-24 TAG-72 is also expressed in a vast majority of gastric adenocarcinomas.21,25 Double histochemical staining with antibodies to CEA and TAG-72 revealed that the two antibodies, in combination, reacted with at least 90% of the carcinoma cells in 13 out of 17 tumors.21 However, Gero et al. reported a poor correlation between the CEA and TAG-72 antigen values in sera obtained from gastric cancer patients, suggesting the complementarity of CEA and TAG-72 measurements in the analysis of gastric cancer.26 In addition, Ohuchi et al. confirmed that when monoclonal antibodies B72.3 and COL-6 (anti-CEA antibody) were simultaneously reacted with the gastric carcinoma tissue, more carcinomas were detected than with each individual antibody alone.21 We found that CEA was expressed in 90% of the gastric tumors and TAG-72 was in 72.5% of lesions. In 95% of lesions, at least one of the two antibodies reacted with the stomach cancer specimen. 

Previously, Del Vecchio reported the use of in vitro quantitative autoradiography for quantitating and evaluating melanoma tumor antigens.27 In contrast to immunohistochemical techniques, this method yields numerical data which are not subject to the observer's judgment. And different antigen/antibody systems can be directly compared. In this autoradiographic system, we measured, strictly speaking the antigen recognized by the monoclonal antibody we used rather than the antigen itself. By using a standard stick of melanoma cultured cells (H2669), we examined the effects of variable incubation time and temperature with the radiolabeled antibody and found that l hour's incubation at room temperature was enough to obtain adequate data.13- We also compared these autoradiographic data to those obtained by Scatchard analysis and found similar values.13 Del Vecchio reported the coefficient of variation in quantitative autoradiography was 20%.27 

In this study we found that the amount of CEA and TAG-72 varied greatly among stomach cancers. This suggests that these two antibodies cannot be used uniformly for radioimmunoscintigraphy or radioimmunotherapy of stomach cancer. The maximal values of both antigens in stomach cancer were similar (around 500-600 pmol/g), but the median value for CEA was higher than that for TAG-72. Our autoradiographic data showed that the TAG-72 concentration was higher in mucin containing adenocarcinoma. This finding is in accord with previous reports, in which TAG-72 seems to be present on the mucinous protein of adenocarcinoma.21 Clinical radioimmunoscintigraphy with the B72.3 antibody revealed a greater accumulation of B72.3 in mucinous adenocarcinoma.10 In contrast, the concentration of CEA expression was similar in both groups. In addition, there was no difference in the concentration of CEA among well differentiated, moderately differentiated and poorly differentiated adenocarcinomas. This finding was previously observed when using an immunoperoxidase staining method.28 Santeusanio et al. reported results which differ from ours where they found a higher percentage of CEA positivity in well differentiated tumors as compared to moderately differentiated and undifferentiated tumors;29 this report was not quantitative. The present study demonstrated both CEA and TAG-72 in areas of intestinal metaplasia. Nagura et al. found that the intensity of the CEA immunochemial reaction was related to the severity of metaplasia.28 Ohuchi detected TAG-72 in benign lesions with intestinal metaplasia.25 Generally, intestinal metaplasia of the stomach mucosa is believed to be a precancerous lesion. Our findings suggested a relationship between these antigens and carcinogenesis in the stomach. 

Here we have shown that a significant number of stomach cancers expressed cellular CEA and TAG-72 heterogeneously . Heterogeneous expression of tumor antigens has been observed previously in gastric carcinomas.21,30 There is not only antigenic heterogeneity among carcinoma cell populations, but also a temporal modulation of tumor antigen.21 The CEA antigen was more frequently homogeneous than was TAG-72. However, in light of the degree of antigenic heterogeneity, the use of mixtures of monoclonal antibodies reactive with different antigens may be useful for the application of monoclonal antibodies as diagnostic adjuncts to detect or treat human carcinomas. 

In summary, stomach cancer biopsies exhibit a wide variation in TAG-72 and CEA expression with heterogeneity in cellular antigen expression. The CEA expression is more frequent and hemogeneous than TAG-72. These findings suggest that the anti-CEA antibody may be more suitable than the B72.3 antibody for diagnostic immunoscintigraphy and immunotherapy of stomach cancer. 

ACKNOWLEDGMENTS 
We thank Ms. Mee Kyoung Hong and Ms. Seok Rye Choi of the Department of Nuclear Medicine, Seoul National University Hospital for their excellent technical support. This study was supported in part by grants from the International Atomic Energy Agency (6037/RB) and from the Cancer Research Center of the Korea Scientific and Engineering Foundation (KOSEF SRC-56-CRC94-6-2) . 

REFERENCES 
l. Parkin DM. Stjernsward J, MuirCS. Estimates of the worldwide frequency of twelve major cancers. Bull WHO 62: 163-182, 1984. 
2. Weed TE, Neessle W, Ochsner A. Carcinoma of the stomach: why are we failing to improve survival?Ann Surg 193: 407-4l3, 1981. 
3. Diel JT, Hermann RE. Cooperman AM, Hoerr SO. Gastric carcinoma: a ten-year review. Ann Surg 198: 9-12, 1983. 
4. Goldenberg DM, DeLand F, Kim E, Bennett S, Primus FJ, van Nagell JR, et al. Use of radiolabelled antibodies to carcinoembryonic antigen for the detection and localization of diverse cancers by external photoscanning. N Engl J Med 298: 1384-1388, 1978.  
5. Goldenberg DM, Larson SM. Radioimmunodetection in cancer identification. J Nucl Med 33: 803-814, 1992. 
6. Armitage NC,Perkins AC,Pimm MV,Farrands PA, Baldwin RW, Hardcastle JD, et al. The localization of an anti-tumor monoclonal antibody 791T/361 in gastrointestinal tumors. Br J Surg 71: 407-412, 1984. 
7. Epenetos AA, Britton KE. Mather S, Shepherd J. Granowska M, Talyor-Papadimitriou J, et al. Targeting of iodine-123labelled tumor-associated monoclonal antibodies to ovarian, breast and gastrointestinal tumors. Lancet 2: 999- 1008, 1982. 
8. Carrasquillo JA, Sugarbaker P. Colcher D, Reynolds JC, Esteban J, Bryant G, et al. Radioimmunoscintigraphy of colon carcinoma with iodine-131-labelled B72.3 monoclonal antibody. J Nucl Med 29: 1022-l030, 1988. 
9. Wagner C, Muller-Wallraf R, Nisson S. Groener J, Breurer H. Localization and concentration of carcinoembryonic antigen (CEA) in gastrointestinal tumors: correlation with CEA levels in plasma. J Natl Cancer Inst 67: 539-547, 1981. 
10. Estaban JM, Colcher D, Sugarbaker P, Schlom J. Quantitative and qualitative aspects of radiolocalization in colon cancer patients of intravenously administered MAb B72.3 . Int J Cancer 39: 50-59, 1987. 
11. Brummendorf T. Anderer FA, Staab HJ, Hornung A, Kieninger G. Carcinoembryonateantigen: diagnose der tumorprogression bei testinalen tumoren. Dtsch Med Wschr 110: 1963-1968, 1985. 
12. Larson SM, Carrasquillo JA, Reynolds JC. Radioimmunodetection and radioimmunotherapy. Cancer Invest 2: 363-381, 1984. 
13. Chung J-K, Graham K, Carrasquillo JA, Reynolds JC. Distribution and concentration of different epitopes on melanoma-associated antigens in human malignant melanoma. JNucl Med 32: 1056, 1989.
14. Chung J-K, Lee DS, Lee MC, Koh-C-S, Reynolds JC. Quantitative measurement of tumor-associated antigen by autoradiography. J Korean Cancer Associat 23: 740-748, 1991.
l5. Colcher D. Horan Hand P, Nuti M, Schlom J. A spectrum of monoclonal antibodies reactive with human mammary tumor cells. Proc Nat Acad Sci (Wash) 78: 3199-3203,1981. 
16. Lee J-H, Chung H-K, Kimm S-W. Purification of carcinoembryonic antigen from culture supernatant of human colon cancer cell lines LS174T using monoclonal immunoaffinity chromatography. Korean J Biochem 20: 23-31, 1988. 
17. Oh YS, Lee JH, Lee YH. Kim SR, Chung H-K, Tchai BS. Production and characterization of monoclonal antibodies to carcinoembryonic antigen. Korean J Biochem 20: 4348, 1988. 
18. Colcher D, Keenan AM. Larson SM, Schlom J. Prolonged binding of a radiolabeled monoclonal antibody (B72.3) used for the in situ radioimmunodetection of human colon carcinoma xenografts. Cancer Res 44: 5744-5751, 1984. 
19. Chung J-K, Choi CW, LeeMC, Chung H-K, KimNK. Choi KW, et al. Radioimmunoscintigraphy of advanced gastrointestinal carcinomas employing 1-131 labeled CEA-79 monoclonal antibody. Ann Nucl Med 7: 65-70, 1993. 
20. Mies G, Nievuhr I, Hossman KA. Simultaneous measurement of blood flow and glucose metabolism by autoradiographic technique. Stroke 12: 581-588, 1981. 
21. Ohuchi N, Simpson JF, Colcher D, Schlom J. Complementation of anti-CEA and anti-TAG-72 monoclonal antibodies in reactivity to human gastric adenocarcinomas. Int J Cancer 40: 726-733, 1987. 
22. Hockey MS, Stoker HJ. Thompson H. Woodhouse CS, MacDonald F, Fielding JWC, et al. Carcinoembryonic antigen expression and heterogeneity in primary and autologous metastatic gastric tumours demonstrated by a monoclonal antibody. Br J Cancer 49: 129-133, 1984. 
23. Ejeckam GC, Huang SN, McCaughey WTE, Hold P. Immunohistopathologic study on carcinoembryonic antigen (CEA)-like material and immunoglobulin A in gastric malignancies. Cancer 44: 1606-1614, 1979. 
24. Ohuchi N, Wunderlich D, Fujita J, Colcher D, Muraro R, Nose M, et al. Differential expression of carcinoembryonic antigen in early gastric adenocarcinomas versus benign gastric lesions defined by monoclonal antibodies reactive with restricted antigen epitopes. Cancer Res 47: 35653571, 1987. 
25. Ohuchi N, Thor A, Nose M, Fujita J, Kyogoku M, Scholm J. Tumor-associated glycoprotein (TAG-72) detected in adenocarcinomas and benign lesions of the stomach. Int J Cancer 38: 643-650, 1986. 
26. Gero EJ, Colcher D, Ferroni P, Melsheimer R, Giani S, Schlom J, et al. CA 72-4 radioimmunoassay for the detection of the TAG-72 carcinoma-associated antigen in serum of patients. J Clin Lab Anal 3: 360-369, 1989. 
27. Del Vecchio S, Reynolds JC, Blasberg RG, Neumann RD, Carrasquillo JA, Hellstroem I, et al. Measurement of local Mr 97,000 and 250,000 protein antigen concentration in sections of human melanoma tumors using in vitro quantitative autoradiography. Cancer Res 48: 5475-5481 , 1988. 
28. Nagura H, Asai T, Watanabe K. Immunocytochemical localization of secretory component (SC) and carcinoembryonic antigen (CEA) in human gastric mucosa. Acta Histochem Cytochem 14: 76, 1981. 
29. Santeusanio G, Peronace L, Castagna G, De Muro G, Santi D, D'orazio A, et al. Immunohistochemical study of carcinoembryonic antigen (CEA) in gastric tumors: Correlation with preparative serum levels, histologic type and grade of anaplasia of the tumor. J Surg Oncol 37: 13-19, 1988. 
30. Nielsen K,TeglbjaergPS. Carcinoembryonicantigen (CEA) in gastric adenocarcinomas. Acta Path Microbiol Immunol Scand 90: 393-396, 1982.