J Clin Pathol 1994;47:117-121

Immunohistochemical demonstration of c-erbB-2oncoprotein in gastric adenocarcinoma:

Comparison of cryostat and paraffin wax sectionsand effect of fixation

K Y Chiu, S L Loke, F C S Ho

AbstractAims-To investigate the effects offixation on the immunohistochemicaldemonstration of c-erbB-2 oncoproteinusing paraffin wax and cryostat sections;to compare c-erbB-2 expression in non-neoplastic and neoplastic gastric tissues.Methods-Adjacent blocks oftumour andnon-neoplastic tissue from four gastrec-tomy specimens were put into a panel of10 fixatives including acetone, B5,Bouin's fluid, Carnoy's fluid, bufferedformalin, formol dichromate, zinc for-malin, 4%/O paraformaldehyde, periodate-lysine-paraformaldehyde (PLP) andperiodate-lysine-paraformaldehyde-dichromate (PLPD) before embedding inparaffin wax for sectioning. Similar tis-sue blocks were snap frozen and cryostatsections were postfixed in these fixatives,either alone or in combination, beforeimmunostaining.Results-In paraffin wax embeddedsections the best fixative was PLP, andin frozen tissues the best results wereobtained after fixation of cryostat sec-tions in buffered formalin followed bycold methanol and acetone. Applyingthese fixatives to samples from a further16 gastrectomy specimens, strong mem-brane staining of c-erbB-2 protein wasfound in the tumour in eight of 16 cases(50%) using paraffin wax sections, andstaining was stronger in the better differ-entiated carcinomas. For frozen tissues,positive membrane staining was found inall gastric adenocarcinomas, but differ-ential staining intensity associated withtumour differentiation could not bedetected.Conclusions-These results indicate thatfixation and paraffin wax embeddingaffect the results of immunohisto-chemical demonstration of c-erbB-2 ingastric cancer. The choice of fixative iscritical in the demonstration and eval-uation of c-erbB-2 protein expression byimmunohistochemistry in gastric carci-nomas. Staining results also varydepending on whether frozen or paraffinwax embedded tissues are studied.

(J Clin Pathol 1994;47:117-121)

The human proto-oncogene, c-erbB-2 (alsocalled HER-2 and neu gene) encodes a cellsurface glycoprotein that is similar in struc-ture to the epidermal growth factor receptor(EGF-R),'-5 and has been mapped to bandq21 of chromosome 17.6 It is the human ana-logue of the transforming gene neu, originallyfound in rat neuroblastoma cell lines derivedfrom tumours induced by ethylnitrosourea.'The product/ of the c-erbB-2 gene is a 185kilodalton transmembrane glycoprotein, withan extracellular ligand binding domain andan intracellular domain with tyrosine kinaseactivity, which is involved in signal transduc-tion.

Because of these structural similarities tothe EGF-R, it has been suggested that the c-erbB-2 gene product may function as a recep-tor for some as yet unidentified growth factor.In normal tissues c-erbB-2 expression hasbeen demonstrated predominantly in epithe-lial cells,7 but the physiological function of c-erbB-2, like that of other known peptidegrowth factor receptors, has not been clari-fied. A possible role for the c-erbB-2 gene inthe pathogenesis of human cancer has beenimplicated by its relatively frequent amplifica-tion in human carcinomas arising frombreast,l'0 stomach," ovary,9 salivary gland,'2kidney,'3 14 colon,'4 and lung. 5

Immunohistochemical staining of oncopro-teins has been used as a semiquantitative andrelatively simple method of assessing onco-gene expression. The results reported byvarious authors are often conflicting. Onepossible explanation is the use of differentfixation procedures by various workers. Wetherefore tested the effects of various fixativeson immunohistochemical staining intensityand cellular localisation of c-erbB-2 protein,using paraffin wax embedded and cryostatsections. After selecting an optimal fixativewe compared the expression of c-erbB-2 inadenocarcinomas of the stomach with adja-cent non-neoplastic stomach mucosal epithe-lium.

MethodsTISSUE PREPARATIONSA total of 20 fresh gastrectomy specimensreceived in the Department of Pathology,University of Hong Kong, were studied; thespecimens were all resections performed for

Department ofPathology UniversityofHong Kong QueenMary HospitalCompound HongKongK Y ChiuS L LokeF C S HoCorrespondence to:Kwan-Yi ChiuAccepted for publication7 September 1993

117

on April 22, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.47.2.117 on 1 February 1994. D

ownloaded from

Chiu, Loke, Ho

Table 1 Duration offixation ofparaffin wax andfrozensections in 10fixatives

Duration offixation

Paraffin wax Frozensection section

Fixative (hours) (minutes)

Acetone 3 10B5 2 15Bouin 3 15Camoy 3 10Neutral buffered formalin 20 15Formol dichromate 20 15Zinc formalin 20 15Paraformaldehyde 20 15PLP 20 1 5PLPD 20 15

gastric cancer. Tissue blocks (5 mm in diam-eter) were obtained from both the tumourand adjacent non-neoplastic mucosa and putin the following fixatives before embedding inparaffin wax for sectioning: (a) acetone; (b)B5; (c) Bouin's fluid (Bouin); (d) Carnoy'sfluid (Carnoy); (e) 10% neutral bufferedformalin; (f) formol dichromate; (g) zinc for-malin; (h) 4% paraformaldehyde; (i) perio-date-lysine-paraformaldehyde (PLP); and (j)periodate-lysine-paraformaldehyde-dichro-mate (PLPD). Paraffin wax sections were cutat 3 ,um, mounted, and dried at 37°Covernight before immunostaining. Similar tis-sue blocks were also snap frozen and cryostatsections were cut at 6 ,um, air-dried at roomtemperature for two hours, and then fixed inthe above fixatives, either alone or in combi-nation, before immunostaining. The durationof fixation in different fixatives for paraffinwax and frozen sections is shown in table 1.Using the first four cases, the fixative thatgave the highest sensitivity in paraffin waxsections compared with frozen sections, bestmembrane localisation, high signal to back-ground ratio, and best preservation of mor-phology, was chosen for the detailed study ofthe remaining cases.

HISTOLOGICAL EVALUATIONSThe selected blocks were assessed histologi-cally and the presence or absence of adeno-carcinomatous tissue was confirmed. Therewere 10 cases of intestinal type (of which onewas also mucinous), four cases of diffuse type(of which one was also mucinous), and twocases of mixed intestinal and diffuse type.The carcinomatous tissue present in the

Table 2 The effect of differentfixatives on immunohistochemical staining of c-erbB-2oncoprotein in four cases ofgastric carcinoma

Paraffin wax Frozen

Fixative Normal Tumour Normal Tumour

Acetone - f+ + ++B5 -

Bouin -

Camoy - - +Neutral buffered formalin - f+ f+ +Formol dichromate -Zinc formalin - f+Paraformaldehyde - f+ f+ +PLP f+ ++ f+ +PLPD + + - +

- = negative or weak staining of <5% tumour cells; f+ = focal moderate staining of 5-25%tumour cells; f++ = focal strong staining of 5-25% tumour cells; + = moderate staining of>25% tumour cells; ++ = strong staining of >25% tumour cells.

chosen blocks was graded using a modifiedBroder's classification. A predominance ofsolid area was regarded as poorly differenti-ated, and a predominance of well formedglands or a papillary pattern was regarded aswell differentiated. Moderately differentiatedareas were those with poorly formed glandu-lar lumina, or predominantly cribriform inappearance.

ANTIBODIESThe mouse monoclonal antibody recognisingthe extracellular domain of c-erbB-2, devel-oped from mice immunised with NIH/3T3cells transfected with a full length cDNAclone of human c-erbB-2, was obtained fromTriton Biosciences Inc (Alameda, CaliforniaUSA). It specifically precipitates the p185protein in metabolically labelled cell lysatesand does not cross react with the EGF recep-tor protein (170 kilodaltons) or any proteinsin non-transfected NIH/3T3 cells.'6

IMMUNOHISTOCHEMISTRYImmunostaining was performed using thealkaline phosphatase-anti-alkaline phos-phatase (APAAP) method.'7 Briefly, the sec-tions were sequentially incubated with 10%normal rabbit serum for 10 minutes, mono-clonal antisera (diluted 1 in 20) overnight at4°C, rabbit antimouse immunoglobulin(diluted 1 in 50) for 30 minutes, and mouseAPAAP complex (diluted 1 in 100) for 30minutes. The intensity of the reaction prod-uct was enhanced by repeating rabbit anti-mouse immunoglobulin and the APAAPsteps for 10 minutes each. The sites of alka-line phosphatase reactivity were visualised byincubation with naphthol AS-BI-phosphateand hexazotized new fuchsin in TRIS-buffered saline, pH 8-7, for 10 minutes.Finally, the sections were washed and coun-terstained with Mayer's haematoxylin, dehy-drated, cleared, and mounted in Permount.An unrelated mouse monoclonal antibody,

UCHL1, was used in place of the primaryantibody as a negative control. Paraffin waxsections of breast carcinoma were used aspositive controls.Normal rabbit serum, UCHL1, rabbit

antimouse immunoglobulin and APAAPcomplex were purchased from DakopattsA/S, Copenhagen, Denmark.

EVALUATION OF THE STAINING RESULTSA semiquantitative analysis was carried out.The staining reactions were recorded asfollows: - = negative or weak staining of lessthan 5% tumour cells; f + = focal moderatestaining of 5-25% tumour cells; f ++ = focalstrong staining of 5-25% tumour cells;+ = moderate staining of more than 25%tumour cells; ++ = strong staining of morethan 25% tumour cells. The staining wasconsidered optimal when: (i) background waslow; (ii) localisation to membrane was sharp;(iii) signal was strong.

CHOICE OF FIXATIVEThe results of immunohistochemical staining

118

on April 22, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.47.2.117 on 1 February 1994. D

ownloaded from

c-erbB-2 in gastricadenocarcinoma19

formol dichromate were unsuitable and theyshowed negative staining.

In frozen sections the best results wereobtained after fixation in neutral buffered for-malin followed by cold methanol and acetoneand the reaction was localised to the cellmembrane (fig 1). Acceptable results werealso obtained after fixation in neutral bufferedformalin, paraformaldehyde and PLP, andthe reaction was localised to cell membrane.Fixation in Gamoy and PLPD showed mod-erate diffuse cytoplasmic staining. Althoughstrong membrane staining was shown afterfixation in acetone, the result was unaccept-able due to poor tissue morphology. Otherfixatives such as B5, Bouin, formol dichro-mate and zinc formnalin provided negativeresults.

In both methods focal membrane stainingof normal gastric epithelia, mainly located atthe surface, was also noticed on some of thesections (fig 2). Negative staining withUCHL1 confirmed the specificity of this result.

Figure 1 Frozen section of moderately differentiated gastric adenocarci2

neutral buffered formalin followed by cold methanol and acetone, showin,membrane stainingfor c-erbB-2 protein.

of c-erbB-2 oncoprotein in parfrozen sections are summarised

paraffin wax embedded secti,

result was found after fixationito 20 hours and the reactioi

localised to the cell membrane

areas. Focal moderate memt

was also obtained after fixatic

neutral buffered formalin, zinc

paraformnaldehyde. Moderate

also found after fixation in

fixatives such as B5, Bouin,

x 1:

X X

W 4pA.:

4. A!,

M

X

D

0.4AW

40.

AW

Figure 2 Normal gastric epithelium fixed in PLP (two hours), paraffiishowing negative staining for c-erbB-2 protein.

~ioafxedin COMPARISON OF CARCINOMA WITH ADJACENTigstrong NON-NEOPLASTIC MUCOSA

The results of c-erbB-2 oncoprotein expres-sion in paraffin wax and frozen sections of 16cases gastric adenocarcinomas are shown intable 3. In frozen tissues positive membrane

raffin wax and staining was found in all gastric carcinomas.Iin table 2. In Enhanced oncoprotein expression (comparedions, the best with the adjacent non-neoplastic mucosa)inPLP fortwo was identified in 13 out of 16 cases (81%).n was mainly There was no correlation between the inten-of the tumour sity of staining and tumour differentiation, asbrane staining assessed on frozen sections.in in acetone, In paraffin wax embedded tissues the stain-formalin and ing results differed from those in frozen sec-staining was tions. Positive membrane staining was foundPLPD. Other in only eight of 16 (50%) of gastric carcino-Camoy and mas, of which seven cases expressed more

than the adjacent non-neoplastic mucosa. Inmost of the positive staining tumours t-hemembrane staining was heterogeneous in itsintensity, wth some areas more intense thanothers. Sometimes the staining was demon-strated intracytoplasmically in addition toobvious membrane staining. Most positivecases with prominent staining (+ +) were

mas (fig 3). In contrast to this, poorly differ-

entiated carcinomas showed weaker stainingor none at all (fig 4).

Discussion4~~~~~~~~1Porter et al' reported that there was no dif-

ference in the detection of c-erbB-2 oncogeneproduct in frozen and paraffin wax sections ofbreast lesions. However, there was a poorcorrelation in c-erbB-2 immunoreactivitybetween the paraffin wax and frozen sectionsin our Study; agreement between the twotypes of sections was found in only 50% ofgastric cancers. Eight negative cases werefound in paraffin wax sections, but not infrozen sections of gastric cancers. The locali-

~zwxsetion sation of staining in tumour tissues was simi-lar in both paraffin wax and frozen sections.

119

on April 22, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.47.2.117 on 1 February 1994. D

ownloaded from

Chiu, Loke, Ho

Table 3 Comparison of c-erbB-2 oncoprotein staining in paraffin wax and frozensections ofgastric adenocarcinoma

Paraffin wax FrozenDegree of

Case differentiation Normal Tumour Normal Tumour

GI Well f+ + + + f+ + + +G2 Moderate f+ + + f+ + + +G3 Moderate f+ f+ + + +G4 Moderate - f+ + +G5 Moderate f+ + +G6 Moderate f+ + + - ++G7 Moderate f+ + + + f+ + + +G8 Moderate f + + + f+ + +G9 Moderate + + f+ + + +GIO Poor f+ - + + + +GIl Poor - + + +G12 Poor + + + + + +G13 Poor + + + +G14 Poor f+ +G15 Poor f+ - + + + +G16 Poor f+ + + + + +

(For key and abbreviation -to + +, please see table 2).

Because Porter et al used routine formalinfixed paraffin wax embedded sections in theirstudy, and PLP fixed paraffin wax sectionswere used in our study, the discrepancybetween our results and Porter's may havebeen due to the different fixatives and theantibody used. The different detection of c-

erbB-2 protein in paraffin wax and frozen sec-

tions may due to the fragile antigenic sites forthe protein, as these may be destroyed or

diminished when exposed to fixation and heatduring paraffin wax processing.

Although neutral buffered formalin was a

poor fixative for demonstrating c-erbB-2 pro-tein in gastric carcinoma, interestingly, strongmembrane staining was demonstrated in our

neutral buffered formalin fixed control(breast carcinoma) paraffin wax sections. Thedifferent effect of the same fixative on the twotumour tissues is not readily explained. Itmay have been due to the quantitative or

qualitative differences in the c-erbB-2 protein

Figure 3 Paraffin wax section of moderately differentiated gastric adenocarcinoma fixedin periodate-lysine-paraformaldehyde (two hours), showing strong membrane staining forc-erbB-2 protein.

in gastric and breast tissues or, as suggestedby Falck et al, breast cancers in a positivetumour uniformly overexpress the c-erbB-2proteins, while in gastric cancers, the expres-sion can be patchy or very focal.'9

Previous studies by different investigatorshave shown discrepant results for the expres-sion of c-erbB-2 oncoprotein in gastrictumours. Houldsworth et al,20 Yonemura etal,2' Falck et al,'9 and Lemoine et al,22 usingformalin fixed paraffin wax sections, showedthat positive staining was found in 10%, 12%,19%, and 26% of gastric carcinomas, respec-tively. This low level of positivity, comparedwith our results, indicates that using routineformalin fixed paraffin wax embedded mater-ial may underestimate the percentage of posi-tivity in tumours studied. Southern blottinghas shown c-erbB-2 amplification in 6-22% ofgastric carcinomas in various studies (Park etal 23; Yokota et al 1113; Ranzani et al24). Inanother study, using western blotting,Kameda et al 5 reported that 55% of casesshowed higher expression of c-erbB-2 proteinin extracts of tumours compared with normaltissue. These studies support our finding thatc-erbB-2 protein is frequently overexpressedin gastric cancer.

Focal to membrane staining was found in"normal" gastric epithelium in paraffin waxsections, even when the tumour was negative,but this staining pattern was not observed infrozen sections in our study.Our results demonstrate that the sensitivity

of immunostaining is dependent on fixationprocedure and the detection of c-erbB-2 in50% of our cases using PLP fixed paraffinwax sections is more in keeping with thefinding by Kameda et al using westernblotting.25 Using frozen sections, we not onlydetected the presence of c-erbB-2 in all 16cases of gastric carcinomas, but we have alsodemonstrated the consistent enhanced ex-pression of the oncoprotein in tumour tissues.A higher prevalence of membrane staining

was found in moderate to well differentiatedcarcinomas than in poorly differentiated ade-nocarcinoma in PLP fixed paraffin wax sec-tions in our study. This is similar to the resultreported by Jain et al.26 There is no correla-tion, however, between the differentiation ofthe tumour and the intensity of staining infrozen sections, in contrast to the resultsusing paraffin wax sections. This suggeststhat there may be a difference in resistance toloss of antigenicity of c-erbB-2 protein fol-lowing fixation in tumours with a differentdegree of differentiation. The reason for this isnot known and deserves further investigation.We conclude that the choice of fixative is

critical in the demonstration and evaluationof c-erbB-2 protein expression shown byimmunohistochemistry in gastric carcinomas.Staining results also vary, depending onwhether frozen or paraffin wax embedded tis-sues are studied. Correlation of stainingintensity with tumour differentiation is alsoinfluenced by such factors. This must betaken into consideration when collecting andcomparing data.

120

on April 22, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.47.2.117 on 1 February 1994. D

ownloaded from

c-erbB-2 in gastric adenocarcinoma

r~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~41

n~ ~J70

work.was...

5.~ ~ ~ ~ ~ ~ ~ ~ ~~WO

Aril

o0esi0x

*qI'K ~ ~

J:~~ ~ ~ ~ ~ ~ b.-el -'8MrtmuJnign1aue1943253

Y~angFnLMa a..Tyrosine

reetrwtJxesvhmlgroatobihneknoee

YY.:.-'Sciece

Oft ~ ~ ~ ~

W~~~~~~mN,S

.0.9~ ~ ~

6. Fuuhgg-,

Loaiaino5oe0-rb-eae

mpiictoniV

A~~ ~ ~ ~ ~

specific expression of the neu oncogene in rat develop-ment. Pr-oc NadAcad Sci USA 1987;84:8498-501.

8 Slamon DJ, Clark GM, Wong SG, et at. Human breast

cancer: correlation of relapse and survival with amplifi-cation of the HER-2/neu oncogene. Science 1987;

235:177-82.

9 Slamon DJ, Godolphin W, Jones L-A, et at. Studies of the

HER-2/neu proto-oncogene in human breast and ovar-

ian cancer. Science 1989;244:707 12.

10 Venter DJ, Tuzi NL, Kumar 5, et at. Overexpression of

the c-erbB-2 oncoprotein in human breast carcinomas:

immunohistological assessment correlates with gene

amplification. Lancet 1 987;ii:69-72.1 Yokota J, Yamamoto T, Miyajimna N, et at. Genetic alter-

ations of the c-erbB-2 oncogene occur frequently in

tubular adenocarcinoma of the stomach and are often

accompanied by amplification of the v-erbA homologue.Oncogene 1988;2:283-7.

12 Semba K, Kamata N, Toyoshima K, et at. A v-erbB

related proto oncogene, c-erbB-2, is distinct from the c-

erbB-1/epidermal growth factor receptor gene and is

amplified in a human salivary gland adenocarcinoma.

Pr-oc NadAcad Sci USA 1985;82:6497-501.

13 Yokota J, Yamamoto T, Toyoshima K, et at.

Amplification of c-erbB-2 oncogene in human adeno-

carcinomas in vivo. Lancet 1986;i:765-7.14 Tal M, Wetzler M, Josefberg Z, et at. Sporadic amplifica-

tion of the HER-2/neu protooncogene in adenocarcino-

mas of various tissues. Cancer Res 1988;48:1517 20.

15 Chine MJ, Battifora H. Abnormalities of proto-oncogenes

in non-small cell lung cancer: correlations with tumour

type and clinical characteristics. Cancer 1 987;60:2669-74.

16 Langton BC, Tran K, Jackson J, er at. The developmentand characterization of antibodies to different regions of

the c-erbB-2 protein on breast tumour tissue and cell

lines. Pr-oceedings of the American Association for Cancer

Research 1989;30:352.17 Cordell IL, Falini B, Erber WIN, et at. Immunoenzymatic

labeling of monoclonal antibodies using immune com-

plexes of alkaline phosphatase and monoclonal anti-

alkaline phosphatase (APAAP complexes). J7 Histochem

Cytochem 1984;32:219 29.

18 Porter PL, Garcia R, Moe R, et at. c-erbB-2 oncogene

protein in rn-situ and invasive lobular breast neoplasia.Cancer 1991;68:331-4.

19 Falck VG, Gullick WJ. c-erbB-2 oncogene product stain-

ing in gastric adenocarcinoma. An immunohisto-

chemical study. J7 Pathot 1989;159:107111.20 Houldsworth J, Cordon-Cardo C, Ladanyi M, et at. Gene

amplification in gastric and esophageal adenocarcino-

mas. Cancer Res 1990;50:6417-22.

21 Yonemura Y, Ninomiya I, Yamaguchi A, et at. Evaluation

of immunoreactivity for erbB-2 protein as a marker of

poor short termn prognosis in gastric cancer. Cancer Res

1991;51: 1034-8.

22 Lemoine NR, Jain 5, Silvestre F, et at. Amplification and

overexpression of the EGF receptor and c-erbB-2 proto-

oncogenes in human stomach cancer. Br J7 Cancer

1991;64:79-83.

23 Park J-B, Rhim JS, Park S-C, et at. Amplification, over-

expression, and rearrangement of the erbB-2 proto-

oncogene in primary human stomach carcinomas.

Cancer Res 1 989;49:6605-9.24 Ranzani GN, Pellegata NS, Previdere C, et at.

Heterogenous proto-oncogene amplification correlates

with tumour progression and presence of metastases in

gastric cancer patients. Cancer Res 1990;50:781 1-4.

25 Kameda T, Yasui W, Yoshida K, et at. Expression of

erbB-2 in human gastric carcinomas: relationshipbetween pl85",rB-2 expression and the gene amplifica-tion. Cancer Res 1990;50:8002-9.

26 Jain 5, Isabel Filipe M, Gullick WJ, et at. c-erbB-2 proto-

oncogene expression and its relationship to survival in

gastric carcinoma: an immunohistochemical study on

archival material. Int Cancer 1991;48:668-7 1.

121

on April 22, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.47.2.117 on 1 February 1994. D

ownloaded from