Gut, 1991, 32, 1465-1469

Gastric cancer associated structure in mucusglycoproteins shown as a clinically useful marker

I Hakkinen, T Nevalainen, R Paasivuo, P Partanen, K Seppala, P Sipponen

AbstractA new monoclonal antibody has beendeveloped which is capable of detectingstructures in gastric mucus glycoproteinsexpressed in the fetus and in adult gastricmucosa in conditions such as gastric carci-noma. Cancer associated monoclonal anti-bodies were selected by testing them againstvarious mucous glycoprotein samples from thealimentary tract, including salivary glyco-proteins from both secretory and non-secretory subjects, and cancerous and normalgastric juice glycoproteins. They were testedagainst 1000 samples of gastric juice froman unselected population. Immunochemicalcharacterisation suggested that the glyco-proteins picked up by P4 and i1l include one ofthe compounds reacting with rabbit anti-fetalsulphoglycoprotein antigen serum. On thebasis of a clinical trial and immunohistologicalevaluation further evidence was obtained of P4as the most promising antibody for furtherexperimentation. A total of 302 gastric juicespecimens from patients with various gastricsymptoms were analysed using the enzymelinked immunosorbent assay technique and P4antibody. Of 10 gastric cancers, nine had P4 inthe gastric juice. A positive correlation wasfound between gastric ulcer and the appear-ance of P4. Duodenal ulcers were not corre-lated to P4. Atrophic gastritis and P4 coincidedless frequently. Raised P4 values were found inbetween 3% and 9%/o of subjects, depending onthe population. Cancer cases showed high P4values, which aliows adjustment of the lowerlimit of a positive result to high level whereby aconsiderable number of non-cancerous P4positives are omitted.

Department ofPathology, University ofTurku, Department ofGastroenterology,Medical Clinic,University of Helsinki,Department ofPathology, JorviHospital, Espoo, andLabsystems ResearchLaboratories, Helsinki,FinlandI HakkinenT NevalainenR PaasivuoP PartanenK SeppalaP SipponenCorrespondence to:Dr I Hakkinen, Department ofPathology, University ofTurku, Kiinamyllynk 8-10,20520 Turku, Finland.

Accepted for publication3 December 1990

Gastric cancer cells and their precursors, beingof epithelial origin, have characteristics ofepithelial mucous cells such as the capacity tosecrete mucus glycoproteins. Consequently, we

find these secretory products in the gastric juice.Earlier studies on fetal sulphoglycoproteinantigen (FSA), where a rabbit antiserum was

used to detect qualitatively these gastric cancer

associated blood group type glycoproteins, were

based on this idea.'2 It was stated that theantibodies detected structures other than knownblood group determinants in these blood grouptype glycoproteins.3 Immunohistological datasuggested that FSA structures were expressed innormal ontogenic fetal development of the gut.4

Detection of the FSA for the diagnosis ofgastric cancer among rural populations with a

high incidence of gastric cancer was promising inthe early 1970s but became less important later,with a decline in incidence of this geographically

associated type of cancer. More precisely,although the method was satisfactory in showingearly cases of cancer, the FSA test was plaguedby a high rate of non-cancerous FSA positives,about 8% in an unselected population.5

In experimental studies gastric carcinomas canbe produced by administration of a carcinogen.6When the dose of the carcinogen was reduced astage was found where the fetal glycoproteinmarkers were expressed but a morphologicalcancer did not develop.' A relatively shortexposure to carcinogens, such as chemicals inour surroundings, may cause an analogoussituation, and thus explain the recently observedtendency in our population towards an increasein non-cancerous FSA positives.The FSA glycoprotein, characterised by poly-

clonal rabbit antiserum, necessarily contains amultitude of determinants which in individualsamples can be present in incomplete combina-tions. The possibility exists that some of the FSAdeterminants are more strongly cancer associ-ated - that is, expressions triggered later incarcinogenesis. To investigate this, monoclonalantibodies were produced, then characterisedand selected for use in a clinical trial. The presentstudy gives immunochemical and histoimmuno-logical data on the most promising of thesemonoclonal antibodies, P4. A clinical study usingP4 was included to support our concept of itsassociation with cancer.

Methods

PREPARATION OF GLYCOPROTEIN ANTIGENS FORIMMUNISATIONThe cancerous gastric juice sample was takenfrom a patient with gastric cancer (age 67 years,blood group 0+, histological type: diffusecarcinoma). The specimen was taken by oraltube after an overnight fast, the patient havingreceived 100 ml of phosphate buffered saline pH6-5. The antigen was prepared following theprocedure described earlier.8 Polyanions fromthe gastric juice were separated using cetylpyridinium chloride (CPC) (Merck, Germany)and ethanol precipitations. The supernatant ofthe CPC precipitation containing the neutralglycoproteins was further treated with fourvolumes of ethanol, and the precipitate wasdissolved in saline, to be used in immunisation asfraction 1. CPC precipitated polyanions weredivided into two fractions by a further CPC-saltprecipitation procedure, and they were used inimmunisations as fractions 2 and 3.

PRODUCTION OF MONOCLONAL ANTIBODIESPreparation of monoclonal cell fusions and the

1465

Hakkinen, Nevalainen, Paasivuo, Partanen, Seppdla, Sipponen

cloning procedure was as described elsewhere.9To measure the specific reactivity of the mouse

antibodies to the antigen, the enzyme linkedimmunoassay technique was used, with anti-mouse immunoglobulin conjugated with alkalinephosphatase. The antigenic preparation used inimmunisations was coated in microplate wells bytreatment with 0.2% glutaraldehyde in 0 1 mol/lphosphate buffer pH 5, at +4°C overnight.'`The quantitative assay for FSA using the

monoclonal antibodies produced was developedby detecting the antigen on the solid phase withthe monoclonal antibody. Attempts to coatnative gastric juice samples met with difficulties.This was probably caused by competition for thefixing sites by too many kinds of unrelatedproteins. Half an hour's incubation at 65°C in0.1% papain-cysteine solution buffered in 0-2mol/l acetate to pH 5 6 split most of the proteinsand solubilised the FSA antigen molecules with-out destroying the antigenic properties. Fourvolumes of ethanol added subsequently pre-

cipitated the glycoproteins, which in aqueous

solution could then be coated onto the solidphase.

TESTS FOR SPECIFICITY OF THE CLONESThe specificity of clones was tested by measuringtheir reactivity with (i) non-secretory saliva, (ii)secretory saliva, (iii) 10 samples of gastric juicefrom normal control subjects, (iv) 10 samples ofcancerous gastric juice, and (v) blood groupABH, Lewisab systems, performed as in routineblood group serology. In addition, selected anti-bodies were further used to test 1000 specimensof gastric juice from an unselected populationaged 49-74 years which had been preservedfrozen from an earlier study.

Selected specific monoclonal antibodies were

immobilised using CNBr2 activated Sepharose(Pharmacia-LKB Biotechnology, Uppsala,Sweden). A small antibody column was preparedfor affinity purification of the glycoproteinantigen. Gas chromatography for sugar analysisof the affinity purified glycoprotein was per-formed at the Department ofMedical Chemistry,University of Helsinki. A gel filtration profile ofthe glycoproteins in various preparations was

obtained on a Sephacryl S-200 column, detectedby the specific monoclonal antibody.Sodium dodecylsulphate-polyacrylamide gel

electrophoresis" was performed using 6% or 8%vertical slab gels either under reducing or non-

reducing conditions. For immunoblotting,'2 thepolypeptides were transferred onto nitro-cellulose sheets (Trans-Blot Transfer Medium,Bio-RAD, Richmond, CA). Immunoblottingwas performed and detected by using peroxi-dase coupled antimouse immunoglobulins(Dakopatts, Glostrup, Denmark).

IMMUNOHISTOLOGICAL STUDY

The immunochemically selected monoclonalantibody P4 (see Results) was tested. A standardimmunoperoxidase procedure was used on

formaldehyde fixed tissue sections. Parallelnegative controls were included in all assays. Thefollowing organs and conditions were studied:

non-cancerous stomach, gastric carcinoma,pancreatic carcinoma, small intestine, normalcolon, colonic carcinoma, gall bladder, and sub-mandibular and parotid glands.

CLINICAL STUDYThe clinical study included 302 patients referredfor gastroscopy at the gastroenterology unit ofthe University Hospital, Medical Department,Helsinki. The cases were deliberately notselected in any respect, the aim being to get anoverall picture of specimens from the differentgroups of patients. To standardise the gastricjuice specimens each patient drank 100 ml of 0 1mol/l phosphate buffered saline pH 6.5 immedi-ately before the gastroscope was introduced. Thestomach was then emptied by suction throughthe gastroscope. The samples were preserved insealed tubes at -20°C.

P4 analysis was performed using the enzymelinked immunosorbent assay (ELISA) tech-nique, the papain treated samples being coatedonto solid phase as described above. Since in thepreliminary tests i11 (see Results) gave negativeresults in some of the cancer cases, P4 wasselected as the more promising antibody, andpatient analyses are reported for P4 only. Tocontrol the daily variation in the ELISA results,a standard antigen prepared from a P4 positivesample was included in every run. The resultsare given as absorbance figures. The lower limitof a positive result was set at 0.40 absorbance,nearly the cut off of the background value.

ResultsAll three antigen preparations caused an immuneresponse in mice. Cloning of the fusions gave 20active clones in fusion 1 (antigen preparation 1),23 in fusion 2, and 18 in fusion 3. The testsperformed are listed in Table I. Two individualclones, il1 in fusion 1 and P4 in fusion 3, wereselected to be the most promising as regards theircancer association. Unlike i1 1 and P4, most of theother clones reacted to salivary and normalgastric glycoproteins. The two selected cloneswere further tested against 1000 specimens ofgastric juice derived from an unselected ruralpopulation. P4 antibody gave a positive reactionin 30 samples, whereas i11 reacted in only 14 ofthe same 30 P4-positive samples. All i11 positivesamples were also P4 positive.

Affinity purified glycoprotein from a P4 anti-

TABLE I Tests for characterisation ofthe most promisingmonoclonal antibodies, P4 (from immunisation with acidcancerous gastric glycoproteins) and il 1 (from immunisationwith neutral cancerous gastric glycoproteins)

Tests performed P4 ill

Enzyme linked immunoassay technique:Reactivity with non-secretory saliva -Reactivity with secretory salivaReactivity with 10 cancerous gastric juice

samples 10/10 10/10Reactivity with 10 non-cancerous gastric

juice samples - -

Reactivity with 1000 samples fromunselected population (aged 49-74 years) 30/1000 14/1000

Reactivity with blood group antigens:ABH - -

Lewisab

1466

Gastric cancer associated structure in mucus glycoproteins shown as a clinically useful marker

TABLE II Gaschromatography analysis ofsugars ofP4 affinity-purifiedcancerous gastricglycoprotein

Percentage

Mannose 18.7Galactose 38-8Fucose 11*7Glucosamine 20-1Sialic acid 3-7Galactosamine 14 9

body column indicated that besides the P4epitope, the i1 1 epitope could also be situated inthe same macromolecule, but the picture variedfrom one sample to another so that macromole-cules with P4 epitopes but not i11 appeared. Onthe other hand, i11 alone without the P4 epitopewas not found. In the clinical study, ill waslacking in some of the P4 positive cancer cases.On these grounds, P4 was considered to associatewith amaximum number ofthe cancer-associatedglycoproteins.

Table II gives the results of sugar analyses ofthe P4 antibody affinity purified glycoprotein. Asimilar preparation from a subject with bloodgroup A gave a positive reaction with monoclonalanti-A, indicating that P4 positive macromole-cules contain a blood group determinant. This isconsistent with earlier studies on FSA.3 Rabbitanti-FSA also reacted with P4 immunopurifiedFSA in immunodiffusion. P4 reacted withsamples of meconium.

Figure 1 shows the gel filtration profile of a P4positive sample of gastric juice glycoprotein,both untreated and after papain treatment. Noappreciable alteration in the molecular size of theP4 molecules can be observed. The profileindicates some polydispersity of the P4 activemacromolecules, the molecular size rangingfrom 40000 to 150 000. It further suggests thatthe papain treatment solubilises some otherglycoproteins.The immunoplotting pattern of a P4-iI 1

positive sample after sodium dodecylsulphate-polyacrylamide gel electrophoresis is given inFigure 2. Compared with the profile from theSephacryl S-200 run (Fig 1), no discrepancyexists since 6% electrophoresis gel apparentlyobstructs the free movement of the largest mole-cules. Figure 3 shows the standard dilution curveof the P4 antibody enzyme linked immunoassay,the papain treated antigen on the solid phase.

Immunohistologically, P4 was localised ingastric cancer cells in the cytoplasm and in thesecretion (Fig 4A). In none of the sections did allcancer cells show the P4 structure. The stainingpicture varied also in its intensity. As a rule the

A B C

140.l

94.Cl

67r

S

cm

_1

_.FA'A

Figure 2: Immunoblotting pattern ofa P4-111 positive sampleafter sodium dodecylsulphate-polyacrylamide gelelectrophoresis. Lines A and B: molecular size standards; lineC: protein staining; line D: P4 antibody; line E: i,1 antibody;line F: an unrelated monoclonal antibody.

normal gastric mucosa of the same section - thatis, adjacent to a tumour - was negative as wasthe mucosa from non-cancerous stomachs.Immunohistologically, metaplasia and atrophicgastritis were always negative, both in the gastriccancer cases and in non-cancerous cases. Excep-tionally, some single glands stained positively(Fig 4B), less frequently in non-cancerousstomachs than in cancer cases. Some dysplasticglands adjacent to a cancer were positivelystained in about two thirds of the cases (Fig 4C).Table III summarises the immunohistologicalresults for various gastrointestinal materials.

Table IV gives the results of the P4 analysis of

10 01(

D0OD

1:1

Figure 1: Gelfiltration onSephacryl S-200 ofa pooledsample ofgastric juice mucusglycoproteins. (A) Aftertreatment with papain; (B)no enzymatic treatment.Detection at 210 nm (0);detection at 405 nm with P4monoclonal antibody usingenzyme immunoassay (O).

0.

(A) After papain treatment

0o oo o00 0aO mn

00rls,

-EcDO0'I

O No enzymatic treatment

o oo oo oCO C)

ml

1467

D E f

A*i.

C)0_-

.1

11n1

Hakkinen, Nevalainen, Paasivuo, Partanen, Seppald, Sipponen

1-000-

0900-

0-800-

Figure 3: Dilution curve ofpapain treated glycoproteinantigen ofgastric juice. P4antibodyenzyme linkedimmunoassay.Antigen on the solid phase.

E

,£ 0-7000

@ 0-600cm

0. 0-500-

0.400-

0.300O

hand, P4 and duodenal ulcer showed no positivecorrelation.

1:5 1:10 1:20 1:30 1: 50Antigen

clinical material, according to tfigures; 83-1% of the samples v

(0.00-0.39 absorbance). Slight0-79 absorbance) was found in 7high positives (absorbance 0.88.9%. Out of this group, a su

values over 1-0 absorbance is sh(Table V correlates the P4 re

gastric diagnoses. Nine of the ]cases showed P4 in the gastric jithe 30 cases of atrophic gastritis i

Among benign gastric disordewas definitely associated with FI

Figure 4: (A) Gastric cancercells stained positively usingthe P4 monoclonal antibodyand the peroxidase technique(original magnificationxS50). (B)Morphologicallynormal gastric mucosaadjacent to a gastriccarcinoma. Some glandsstained weakly positively.P4 antibody peroxidasetechnique (originalmagnification x400). (C)Positively stained dysplasticgland in the mucosa adjacentto a gastric carcinoma. P4antibody peroxidasetechnique (originalmagnification x400).

.*',J ~ ~ ~ ~ ~ i; *,

.*~ _ =, * # _ .

!I0-!5;s? | - 4L_

DiscussionAll active clones reacted with the glycoproteinstructures. Immunisation with a restrictedselection of glycoproteins possibly leaves morereactive capacity in mouse in the immune systemto identify different components compared withimmunisation where whole cancer cells areused. `3

1:70 11.0 Monoclonal antibodies P4 and i1 1 were1:70 1:100 selected using reactions to glycoproteins of theupper alimentary canal. A positive reaction to all10 samples of cancerous gastric juice and a

their absorbance negative reaction to controls (Table I) was avere P4 negative promising result. Among 1000 gastric juiceelevation (0 40- specimens collected earlier, 30 were positive to*9%. The rate of P4 and 14 to il1. These incidence rates only30 or over) was illustrate the rate of marker-positive samples inbgroup of high an unselected population of high cancer risk ageown separately. groups. In the absence of clinical data no correla-~sults to clinical tion to cancerous or precancerous states can bel0 gastric cancer made from them.uice. Only six of P4 and ill reacted with meconium, whichwere P4 positive. indicates that the antigen structures belong tors, peptic ulcer normal fetal development, unlike the mutation)4. On the other based aberrations such as blood group A-like

structures.'4 5 Blood group specificity wasabsent, which points to the assumption that

* other antigen structures are involved in P4-i,1expressions in the blood group-like gastric glyco-proteins of FSA. The P4 immunopurified glyco-

"> F- i; protein reacted in immunodiffusion with thepolyclonal anti-FSA, thus connecting the P4glycoprotein to the FSA.8 However, it is not yetclear whether true P4 specificity is also present inrabbit anti-FSA.

Immunohistologically P4-il1 antibodies|.s detected only structures of gastric origin among

all alimentary organs and diseases studied. Ourclinical data indicate that non-cancerous P4-il1

-0**L *'production also exists, although it is not easy toi-, E show this immunohistologically.

P4 analysis from gastric juice seems to detect80% to 90% of gastric carcinomas. In this study

A'*Zi.w . ~the cancer cases were not selected but were from,t a group of gastroenterology patients and in

various clinical states. Whereas the rate of P4positivity in the unselected population was about

V 3%, both low and high positives included, therate in selected cases from a gastroenterologyunit would be expected to be higher. It actuallyvaried between 9% and 16%, depending on howhigh the lower limit of a positive test was set.Gastric carcinomas gave high values of P4, which

41 * allows a rise of the lower limit of positivity, andconsequently the cutting off of an appreciablenumber of non-malignant P4 positives. A larger

.tr-g group of cancer patients is, however, neededto fix this limit definitely. For instance, raisingthe limit from 0-40 to 0.80 in our series cuts offone cancer case but at the same time 23 non-cancerous P4 positive cases.

In our cases gastric ulcer showed the highestrate of non-cancerous P4 positivity. On the otherhand, atrophic gastritis, which is known to beassociated with gastric cancer, showed less fre-quent correlation to P4 positivity. In addition, we

n.gnn-J.

1468

Gastric cancer associated structure in mucus glycoproteins shown as a clinically useful marker 1469

TABLE III Material studied immunohistologically using the P4 antibody

No ofSpecimen specimens Immunohistologicalfindings P4

Non-cancerous stomach 5 1 of 5 cases: single positive glands +4 or 5 cases: totally negative

Gastric carcinoma* 22 14 of 22 cases: part of cancer cells positive +16 of 22 cases: a few positive dysplastic glands +3 of 22 cases: totally negative

Pancreatic carcinoma 2Small intestine 2Normal colon 10Colonic carcinoma 10Gall bladder 4Submandibular gland 2Parotid gland 2

* 12 were of intestinal and 10 of diffuse type.

TABLE IV Results ofP4 found no immunohistological verification of P4analyses grouped by being localised in atrophic gastritis. In fact thereabsorbancefigures

____ was a P4 positive rate of about 10% in acuteAbsorbance P4 gastritis (Table V). It thus seems plausible that0-00-0-39 251(83.1%) the origin ofbiochemical alteration is to be found0-40-0 79 24 (7 9%) outside a morphological atrophic gastritis. Some100+ 28 (860%)* benign conditions, gastric ulcer, and to a lessTotal 302 extent gastritis apparently can include parallel

factors, being linked to carcinogenesis at thecolumn. biochemical level. The positive results associated

with benign gastric ulcer and gastritis could limitits use as a marker for gastric cancer.

Gastric emptying is a routine part of gastro-scopy, and sampling gastric juice only requires aminute to remove the yield to a preservingcontainer. However, to guarantee a proper speci-men for quantitative P4 analysis, it is essentialthat the buffer is swallowed. By natural contrac-tion movements it will properly mix.

Patients' attitudes to gastric examinationdepend greatly on motivation. The inconveni-ence caused by gastroscopy is widely accepted,thanks to the availability of information on itssuperior accuracy. In the 1970s, acceptance ofgastric intubation in our population study of40000 patients was 75%.5 We have to bear inmind that at present there is no serum markerwhich is capable of detecting gastric malignancyreasonably early. A gastric juice marker has afar better potential for this. In mass screeningor clinics, P4 could be considered as a signal ofthe existence of a dysplastic condition or amalignancy.

Supported by a grant from the Finnish Cancer Society.

TABLE V P4 measurements from gastric juice using theELISA technique and correlated to clinical diagnoses

Absorbance

00- 040-Diagnosis 0-39 0-79 0 8+ 10+*

Acute gastritis 75 5 3 3Atrophic gastritis 30 2 4 1Erosion 6 4 0 0Gastric ulcer 15 5 6 4Duodenal ulcer 17 1 0 0Hyperplastic polyp 5 3 1 1Gastric carcinoma 1 1 8 7Oesophageal carcinoma 0 0 1 1Gastric lymphoma 0 0 1 1Diagnosis total 149 21 24 18No disorder verified 102 3 3 0Total 251 24 27 18

*Included in the 0-80+ column.

1 Hakkinen I. An immunochemical method for detectingcarcinomatous secretion from human gastric juice. ScandJ3Gastroenterol 1966; 1: 28-32.

2 Hakkinen I, Viikari S. Occurrence of fetal sulphoglycoproteinantigen in the gastric juice of patients with gastric diseases.Ann Surg 1969; 169: 277-81.

3 Hakkinen IPT. Gastric fetal sulfoglycoprotein antigen (FSA)and blood group antigens A and B. Int Arch Allergy 1974; 47:380-7.

4 Hakkinen I, Korhonen LK, Saxen L. The time of appearanceand distribution of sulphoglycoprotein antigens in thehuman foetal alimentary canal. IntJ3 Cancer 1968; 3: 582-92.

5 Hakkinen IPT, Heinonen R, Inberg MV, Jarvi 0, VaajalahtiP, Viikari S. Clinicopathological study of gastric cancers andpre-cancerous states detected by fetal sulfoglycoproteinantigen screening. Cancer Res 1980; 40: 4308-12.

6 Sasajima K, Kawachi T, Sano T, Sugimura T, Shimosato Y,Shirota A. Esophageal and gastric cancers with metastasesinduced in dogs by N-ethyl-N-nitro-N nitrosoguanidine.J Natl Cancer Inst 1977; 58: 1789-94.

7 Hakkinen IPT, Heinonen R, Isberg U, Jarvi 0. Canine gastricglycoprotein antigens in early carcinogenesis. Cancer 1984;53: 928-34.

8 Hakkinen IPT. The purification procedure for human gastricjuice FSA and its chemical composition. Clin Exp Immunol1980; 42: 57-62.

9 Kohler G, Milstein C. Derivation of specific antibody produc-ing tissue culture and tumor lines by cell fusion. Eur JImmunol 1976; 6: 511-9.

10 Holt PG, Cameron KJ, Stewart GA, Sedgwick J, Turner K.Enumeration of human immunoglobulin-secreting cells bythe ELISA-plague method: IgE and IgG isotypes. ClinImmunol Immunopathol 1984; 30: 159-64.

11 Laemmli UK. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature 1970; 227:680-3.

12 Towbin H, Staehelin T, Gordon J. Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. Proc Natl Acad Sci USA1979; 76: 4350-4.

13 Eisen HN, Siskind GW. Variations in affinities of antibodiesduring the immune response. Biochemistry 1964; 3: 996-1008.

14 Hakkinen I. A-like blood group antigen in gastric cancer cellsof patients in blood groups 0 or B. J Natl Cancer Inst 1970;44:1183-93.

15 Hirohashi S, Clausen H, Yamada T, Shimosato Y, HakomoriS. Blood group A cross-reacting epitope defined by mono-clonal antibodies NCC-LU-35 and -81 expressed in cancer ofblood group 0 or B individuals: its identification as Tnantigen. Proc Natl Acad Sci USA 1985; 82: 7039-43.

found no immunohistological verification of P4being localised in atrophic gastritis. In fact therewas a P4 positive rate of about 10% in acutegastritis (Table V). It thus seems plausible thatthe origin ofbiochemical alteration is to be foundoutside a morphological atrophic gastritis. Somebenign conditions, gastric ulcer, and to a lessextent gastritis apparently can include parallelfactors, being linked to carcinogenesis at thebiochemical level. The positive results associatedwith benign gastric ulcer and gastritis could limitits use as a marker for gastric cancer.

Gastric emptying is a routine part of gastro-scopy, and sampling gastric juice only requires aminute to remove the yield to a preservingcontainer. However, to guarantee a proper speci-men for quantitative P4 analysis, it is essentialthat the buffer is swallowed. By natural contrac-tion movements it will properly mix.

Patients' attitudes to gastric examinationdepend greatly on motivation. The inconveni-ence caused by gastroscopy is widely accepted,thanks to the availability of information on itssuperior accuracy. In the 1970s, acceptance ofgastric intubation in our population study of40000 patients was 75%.5 We have to bear inmind that at present there is no serum markerwhich is capable of detecting gastric malignancyreasonably early. A gastric juice marker has afar better potential for this. In mass screeningor clinics, P4 could be considered as a signal ofthe existence of a dysplastic condition or amalignancy.

Supported by a grant from the Finnish Cancer Society.

TABLE V P4 measurements from gastric juice using theELISA technique and correlated to clinical diagnoses

Absorbance

00- 040-Diagnosis 0-39 0-79 0 8+ 10+*

Acute gastritis 75 5 3 3Atrophic gastritis 30 2 4 1Erosion 6 4 0 0Gastric ulcer 15 5 6 4Duodenal ulcer 17 1 0 0Hyperplastic polyp 5 3 1 1Gastric carcinoma 1 1 8 7Oesophageal carcinoma 0 0 1 1Gastric lymphoma 0 0 1 1Diagnosis total 149 21 24 18No disorder verified 102 3 3 0Total 251 24 27 18

*Included in the 0-80+ column.

1 Hakkinen I. An immunochemical method for detectingcarcinomatous secretion from human gastric juice. ScandJ3Gastroenterol 1966; 1: 28-32.

2 Hakkinen I, Viikari S. Occurrence of fetal sulphoglycoproteinantigen in the gastric juice of patients with gastric diseases.Ann Surg 1969; 169: 277-81.

3 Hakkinen IPT. Gastric fetal sulfoglycoprotein antigen (FSA)and blood group antigens A and B. Int Arch Allergy 1974; 47:380-7.

4 Hakkinen I, Korhonen LK, Saxen L. The time of appearanceand distribution of sulphoglycoprotein antigens in thehuman foetal alimentary canal. IntJ3 Cancer 1968; 3: 582-92.

5 Hakkinen IPT, Heinonen R, Inberg MV, Jarvi 0, VaajalahtiP, Viikari S. Clinicopathological study of gastric cancers andpre-cancerous states detected by fetal sulfoglycoproteinantigen screening. Cancer Res 1980; 40: 4308-12.

6 Sasajima K, Kawachi T, Sano T, Sugimura T, Shimosato Y,Shirota A. Esophageal and gastric cancers with metastasesinduced in dogs by N-ethyl-N-nitro-N nitrosoguanidine.J Natl Cancer Inst 1977; 58: 1789-94.

7 Hakkinen IPT, Heinonen R, Isberg U, Jarvi 0. Canine gastricglycoprotein antigens in early carcinogenesis. Cancer 1984;53: 928-34.

8 Hakkinen IPT. The purification procedure for human gastricjuice FSA and its chemical composition. Clin Exp Immunol1980; 42: 57-62.

9 Kohler G, Milstein C. Derivation of specific antibody produc-ing tissue culture and tumor lines by cell fusion. Eur JImmunol 1976; 6: 511-9.

10 Holt PG, Cameron KJ, Stewart GA, Sedgwick J, Turner K.Enumeration of human immunoglobulin-secreting cells bythe ELISA-plague method: IgE and IgG isotypes. ClinImmunol Immunopathol 1984; 30: 159-64.

11 Laemmli UK. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature 1970; 227:680-3.

12 Towbin H, Staehelin T, Gordon J. Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets:procedure and some applications. Proc Natl Acad Sci USA1979; 76: 4350-4.

13 Eisen HN, Siskind GW. Variations in affinities of antibodiesduring the immune response. Biochemistry 1964; 3: 996-1008.

14 Hakkinen I. A-like blood group antigen in gastric cancer cellsof patients in blood groups 0 or B. J Natl Cancer Inst 1970;44:1183-93.

15 Hirohashi S, Clausen H, Yamada T, Shimosato Y, HakomoriS. Blood group A cross-reacting epitope defined by mono-clonal antibodies NCC-LU-35 and -81 expressed in cancer ofblood group 0 or B individuals: its identification as Tnantigen. Proc Natl Acad Sci USA 1985; 82: 7039-43.