J Clin Pathol 1987;40:734-740

Immunohistology of oestrogen receptor and D5antigen in breast cancer: correlation with oestrogenreceptor content of adjacent cryostat sections assayedby radioligand binding and enzyme immunoassayD D GIRI, V J M DANGERFIELD, R LONSDALE, K ROGERS,*J C E UNDERWOOD

From the Departments ofPathology and *Surgery, University ofSheffield Medical School, Sheffield

SUMMARY Two monoclonal antibodies recognising epitopes associated with oestrogen receptorprotein were evaluated against the assayable soluble oestrogen receptor concentration in a series of149 breast carcinomas. One antibody (anti-ER) recognises the hormone binding unit of oestrogenreceptor and gives nuclear staining; the other antibody (anti-D5) was raised to a component ofsoluble oestrogen receptor and gives cytoplasmic staining. To minimise variations attributable totumour heterogeneity and sampling error immunohistology using the two monoclonal antibodies,radioligand binding assays, enzyme immunoassays, and quantitative histology were done on adja-cent frozen sections. Thirty nine per cent, 48%, 54%, and 43% of the tumours were found to beoestrogen receptor positive by radioligand binding assay, anti-ER and anti-D5 immunohistology,and enzyme immunoassay, respectively. Strong correlations (p < 00005) were found betweenanti-ER immunohistology and the radioligand binding assay. Only weak correlations were foundbetween anti-D5 immunohistology and the results of other assay methods for oestrogen receptor.Nuclear staining of human breast cancers with the anti-ER monoclonal antibody thus seems to bean acceptable alternative to biochemical assays, with the additional advantage of showing inter-cellular and regional heterogeneity for oestrogen receptor content.

Clinical decisions on endocrine manipulation in themanagement of breast cancer are often based on theoestrogen receptor content determined by radioligandbinding assays of tumour homogenates. Althoughthese assays satisfy theoretical criteria for the bio-chemical characterisation of oestrogen receptor, vari-ous defects have become apparent. In addition totheir inability to elucidate the intercellular andregional heterogeneity in oestrogen receptor contentwithin a tumour, these assays preclude simultaneoushistological confirmation that viable tumour ispresent in the assayed sample and an assessment of itsmorphological features.1 Radioligand binding assaysare also affected by factors such as protein estimationmethod,2 quality of radioactive tracers, and methodsof tissue storage and homogenisation. Inter-laboratory and intralaboratory variations in the bio-

Accepted for publication 25 February 1987

chemical measurement of oestrogen receptor contenton the same tumour sample have been reported andtheir possible therapeutic implications discussed.3

There is therefore a need for an effective alterna-tive. Histochemical methods based on fluoresceinatedligands or antiligand antibodies have questionablevalidity.1 Serious theoretical objections have beenraised against these histochemical methods, and theconsensus is that while they may have as yetunsubstantiated clinical relevance they do not appar-ently detect high affinity oestrogen receptor, which,on current evidence, is the main determinant of theendocrine responsiveness of breast cancers.

This study is a comparative evaluation of twomonoclonal antibodies against biochemical data. Theoestrogen receptor monoclonal antibody recognisesthe hormone binding unit of oestrogen receptor butreacts with an epitope remote from, and apparentlyundisturbed -by-, ligand interactions at the hormonebinding site.4 The D5 monoclonal antibody was

734

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from

Immunohistology of oestrogen receptor and D5 antigen in breast cancer

raised to a component associated with solubleoestrogen receptor and, under certain conditions,forms complexes with it.5To ensure meaningful comparisons between meth-

ods and reagents and to assess the influence oftumour cellularity on assay results all discrepant fac-tors including intratumoural oestrogen receptor vari-ations were virtually eliminated by using adjacentfrozen sections cut from the same tissue block.

Material and methods

TUMOUR SAMPLESTissue blocks from 149 breast carcinomas weretrimmed of obvious fat, necrosis, and sclerosis, andsnap frozen at - 50°C in isopentane. Tissue wasstored at - 70°C for not more than three months forradioligand binding assay and for up to two years forimmunohistology and enzyme immunoassay. Allimmunohistology and assay procedures were per-formed on cryostat sections cut adjacently from thesame face of these tissue blocks.

CRYOSTAT MICROTOMYEach tissue block was mounted frozen on a metalchuck using a small volume of ET buffer for micro-tomy in a cryostat running at -20°C.

Subject to the availability of reagents, the followingsections were cut at 4-6 pm from each tumour blockand picked up on glass slides: one test and one controlsection for oestrogen receptor monoclonal antibodyimmunohistology; one test and one control sectionfor D5 monoclonal antibody immunohistology; onesection for quantitative estimation of cellularity.The following sections were cut at 40 pm, picked up

on small glass coverslips, and transferred to plastictest tubes containing buffer: sections to give a totalsection area of about 1 cm2 into 200 p1 ET buffer foroestrogen receptor by assay radioligand binding; asimilar number of sections into 400 pI Tris buffer forenzyme immunoassay using the oestrogen receptormonoclonal antibody.

ER AND D5 IMMUNOHISTOLOGYThe oestrogen receptor monoclonal antibody (AbbottLaboratories Ltd) and the D5 monoclonal antibody(Amersham International plc) were both used inaccordance with the manufacturers' protocols. Insummary, cryostat sections were treated sequentiallyat room temperature, unless stated otherwise, as fol-lows:Oestrogen receptor monoclonal immunohistologySections were fixed in 3-7% neutral phosphatebuffered saline-buffered formalin for 10-15 minutes;phosphate buffered saline wash; methanol at -20°Cfor three to five minutes; acetone at - 20°C for one to

three minutes; phosphate buffered saline washes; nor-mal goat serum for 15 minutes; oestrogen receptormonoclonal antibody (rat) for 30 minutes; phosphatebuffered saline washes; goat antirat immunoglobulinantibody for 30 minutes; phosphate buffered salinewashes; rat antiperoxidase-peroxidase complex for 30minutes; phosphate buffered saline washes;diaminobenzidine (DAB) in hydrogen peroxide forsix to eight minutes; water wash; haematoxylin coun-terstain, dehydrate, cleared, and mounted.DS monoclonal immunohistology Sections werebriefly air dried; 0 3% hydrogen peroxide in methanolfor seven to 10 minutes; phosphate buffered salinewashes; normal goat serum for 20 minutes; D5 mono-clonal antibody (mouse) for 30 minutes; phosphatebuffered saline washes; goat antimouse immu-noglobulin antibody for 20 minutes; phosphatebuffered saline washes; mouse antiperoxidase-peroxidase complex for 20 minutes; phosphatebuffered saline washes; DAB in hydrogen peroxidefor six to eight minutes; phosphate buffered salinewashes; haematoxylin counterstain, dehydrate,cleared and mounted.

Control sections were treated with normal rat ormouse serum substituted for the oestrogen receptorand D5 monoclonal antibodies, respectively. Positivecontrol sections from known oestrogen receptor andD5 positive tissues were included in each batch; inaddition, known oestrogen receptor positive controlsprovided by Abbott Laboratories for use with theirreagent were included in each batch.

Staining intensity was scored and recorded on a 4point scale by one observer throughout (DDG) withindependent verification (JCEU): 0 = no staining; += weak staining; + + = intermediate; + + + =strong staining. An estimate of the proportion oftumour cells positive for oestrogen receptor and D5was also recorded in each case. For comparison withthe results of the radioligand binding assay and theenzyme immunoassay, tumour sections showing + +or + + + staining intensity in more than 25% of thetumour epithelial population were regarded as posi-tive; the rest of the sections were regarded as negative.A staining index was calculated and used to exam-

ine linear correlations with quantitative assays:

staining index = staining intensity x % of cells positive% cellularityQUANTITATIVE HISTOLOGYTumour epithelial celluarity was estimated by pointcounting the haematoxylin and eosin stained cryostatsections.

RADIOLIGAND BINDING ASSAYThe radioligand binding assay of the oestrogen recep-tor content of cryostat sections has been described

735

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from

736elsewhere.6 Briefly, 2 x 10-9M 3H-oestradiol in200p1 of ET buffer (15 mmol disodium edetic acid10mmol Tris-hydrochloric acid, 0-5% bovine serumalbumin, 0-02% sodium azide; pH 7 4) was added tothe tubes containing cryostat sections immersed in200 p1 ET buffer. After one hour at room temperaturethe tubes were centrifuged and the clear supernatant,enriched with diffusible oestrogen receptor nowoccupied by 3H-oestradiol, aspirated into clean tubes.Fifty M1 was withdrawn for protein assay by Brad-ford's method7 and calculated volumes (about 10 p1)of trypsin (1 mg/ml) and calcium chloride (to givefinal concentration of 2 mmol) added to encouragelimited proteolysis over a further 30 minutes' incu-bation at 10°C. Two hundredul was then withdrawnand subjected to isoelectric focusing in poly-acrylamide gel. Radioactivity in gel slices was mea-sured by liquid scintillation counting and the receptorcontent calculated by measuring the area of the peakof radioactivity attributable to specific binding abovethe diagrammatic baseline. Results were expressed asfmol/mg protein and values > 10 were regarded aspositive.

ENZYME IMMUNOASSAYThe coverslips bearing the 40pm sections immersedin 400p1 Tris buffer were gently crushed to facilitate

_*b

.XU-s

AP

Fig I Invasive ductal adenocarcinoma oJ breast stained withanti-oestrogen receptor monoclonal antibody. Tumour cellnuclei are uniformly stained. Stromal cell nuclei are

unstained. (Immunoperoxidase with weak haematoxylinnuclear counterstain.)

Giri, Dangerfield, Lonsdale, Rogers, Underwoodpelleting of the section debris by centrifugation atlOOOg for 10 minutes at 0°C. Two 150pl aliquots ofthe supernatant, now enriched with diffusibleoestrogen receptor, were placed in separate tubes anda further 50pl used for protein assay by Bradford'smethod.7 The oestrogen receptor content of thesupernatants was assayed by enzyme immunoassayusing the oestrogen receptor monoclonal antibodyand other reagents supplied in kit form for this pur-pose (Abbott Laboratories Ltd). Diluent (100 p1) and100 p1 supernatant were added to each well and fol-lowed by a bead coated with the oestrogen receptorrat monoclonal antibody. After incubation for 18hours at 0°C the bead was washed twice in distilledwater and then incubated with 200 p1 of oestrogenreceptor conjugate solution for one hour at 37°C.After two further washes in distilled water the beadwas transferred to a test tube and 300 p1 OPD (o-phenylenediamine. 2 HCI) substrate added and incu-bated for 30 minutes. The reaction was stopped by theaddition of I ml IN H2SO4 to each tube. The reactionproduct was read colorimetrically within two hours at492 nm.The x2 test and Spearman's rank correlation test

were used for all the comparative analyses.

Results

OESTROGEN RECEPTOR MONOCLONAL ANTIBODYImmunohistology with the oestrogen receptor mono-clonal antibody produced exclusively nuclear staining(fig 1): 71 (47 6%) of the tumours were assessed aspositive (graded as + + to + + + staining); 78(52.4%) were assessed as negative (graded as 0 to +staining). Cytoplasmic staining was noted in threetumours only; it was ascribed to endogenous per-oxidase activity, which was not blocked by the stain-ing procedure followed routinely, because it was alsopresent in negative control sections. No staining ofstroma or necrotic areas was observed. Intratumouralheterogeneity with respect to staining was noted asfollows, but not quantified: cell to cell, different stain-ing intensities in adjacent cells within the same region(fig 2); regional, different staining intensities indifferent cell groups (fig 3); benign v malignant,different staining intensities for benign and malignantepithelial cells.

Strong correlation was obtained between staininggrades and radioligand binding oestrogen receptorassay results (p < 0.0005, x2 test) (table 1). Similarly,a good linear correlation was obtained between stain-ing indices and radioligand binding assay results (r =0-801, Spearman's rank correlation test) (fig4). Athreshold staining index of 10 units segregated mostoestrogen receptor positive and negative cases withlittle overlap. Strong correlations were also found

4w,:!..

0:

;.f

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from

Immunohistology of oestrogen receptor and DS antigen in breast cancerbetween the oestrogen receptor monoclonal antibodystaining grades (table 1; p < 0-005) and index (r =0 803), and the results of the enzyme immunoassayusing the same oestrogen receptor monoclonal anti-body. No significant correlation was observedbetween the immunohistology with oestrogen recep-tor and D5 monoclonal antibodies (table 2; p > 0 1).Of the 16 cases with grades + +/+ + + staining

grades that were negative by radioligand binding* oestrogen receptor assay, nine were tumours from

premenopausal women and three had unusually lowcellularity (< 10% compared with average cellularityof about 20%). Two of the four cases exhibiting onlyweak staining but positive on radioligand bindingoestrogen receptor assay had unusually high cellu-larity (38% and 48%).

Cryostat sections from 93 carcinomas were assayedusing the oestrogen receptor monoclonal antibody inthe enzyme immunoassay and the results comparedwith those obtained from the radioligand bindingassay. A high degree of correlation was found (p <0-0005, r = 0-895). Of the 10 tumours that were posi-tive on the enzvme immunoassav hut negative on the

Fig 2 Heterogeneity ofnuclear oestrogen receptor

expression shown by anti-oestrogen receptor monoclonalantibody. Stained and unstained nuclei are evident in thisgroup of invasive carcinoma cells. (Immunoperoxidase withweak haematoxylin nuclear counterstain.)

*}rON,

?'~~~A

Fig 3 Regional heterogeneity ofoestrogen receptor

expression shown by anti-oestrogen receptor monoclonal

antibody. Nuclei ofcribriform intraduct component of this

carcinoma are uniformly stained; surrounding invasive

component is unstained. (Immunoperoxidase with weak

haematoxylin nuclear counterstain.)

.,V,_ 1,, ,, _, ,,,_ ,,,,,,,,, aaCL, vLs ,,_b.,r V,_ l,, 1*,,

radioligand binding assay, seven were from pre-menopausal patients.

D5 MONOCLONAL ANTIBODYOf the 128 tumours examined with the D5 mono-clonal antibody, 68 (53 9%) and 56 (46 1 %) exhibited+ +/+ + + and 0/+ staining grades, respectively.Staining was always cytoplasmic (fig 5). Unexpectedstaining attributable to the D5 monoclonal antibodywas seen occasionally in stromal fibroblasts,inflammatory cells, and vascular endothelium (fig 6).While cell to cell heterogeneity of staining intensitywas less than that observed with the oestrogen recep-tor monoclonal antibody, regional differences anddifferences between benign and malignant epitheliumwithin the same tissue sample were noted.

There were no significant correlations (p > 0 05)between staining grades or indices and oestrogenreceptor concentrations determined by either radio-ligand binding assay or enzyme immunoassay (table3).

Discussion

The highly significant correlation between the resultsof the radioligand binding assay and immuno-histology with the oestrogen receptor monoclonalantibody verifies the suitability of this reagent for theroutine determination of the oestrogen receptor con-tent of breast carcinomas. Although the result is notquantitative to the same extent as that obtained froma radioligand binding assay, immunohistology per-mits confirmation of the presence of viable tumour

.4P.S.S...u *..i.. .. . . .. z |* . W...... _ _

.... } v s*. _.s:w.Z i.. B. .w .......__K ........

a 1 _ -a-FF }F. r w.* - * .wA t ':tv _' to *.e

jlE ML }.!_ ::

w# * ':'.i 44 = . .A

WB# _. r = W}

.w: v ' '_jlS, ib | _:b-_- j w w*_ X

s. *

...: .S '.:. *t:??.

*W :?1|' IE

737

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from

Giri, Dangerfield, Lonsdale, Rogers, UnderwoodTable 1 Correlation between immunohistology staining grades obtained with oestrogen receptor monoclonal antibody andoestrogen receptor content ofadjacent cryostat sections assayed by radioligand binding and enzyme immunoassay

No of tumours

Oestrogen receptor Oestrogen receptor content by:monoclonal Radioligand assay Enzyme immunoassayimmunohistologygrades Negative Positive Negative Positive

O/ + 70 4 49 6++ +++ 16 47 4 33

p < 0 0005 p < 0-0005

220

200

- 180

° 160

CYE 1400

Ec

0

20

<10 i

340 530 70044 4

0 0

0

0

0

0

0* 00*

0

. *

* 69'. - 0 0

<10 20 30 4.0 50 60 70ER monoclonol antibody stainng nf

Fig 4 Correlation between oestrogen recep,radioligand binding and anti-oestrogen recepantibodi' staining index on adjacent cryostat+ 0801). Sixty nine tumours had both oestrconcentrations and staining indices less thanprotein and units, respectively. Of 14 tumourreceptor concentrations of < 1Ofmol/mg prostaining indices > 10, most were either oflovfrom premenopausal Itwomen.

252 450 tissue in the assessed sample and an estimate of the4

* 0-112 degree of cellular heterogeneity with respect tooestrogen receptor content.The use of adjacent cryostat sections for simulta-

neous immunohistology and radioligand bindingassay enabled us to determine the extent to which tis-sue sampling and variations in tumour cellularitycontributed to apparent discrepancies between thetwo methods. Thus we were able to show that some ofthose tumours, which were positive on immu-nohistology but negative (<1O fmol/mg) by radio-ligand binding assay, gave these discrepant resultsbecause they were of unusually low cellularity, a fac-tor which does not influence the immunohistologicalstaining grade. Other tumours that were oestrogenreceptor negative on assay but positive by immu-nohistology were from premenopausal women; endo-genous oestrogens occupy the radioligand bindingsite in these tumours, but this does not renderunavailable or disturb the conformation of the epit-

80 ope recognised by the oestrogen receptor monoclonaldex antibody.

Immunohistology with the D5 monoclonal anti-tor assay by body, which reacts with an epitope on a component oftor monoclonal soluble oestrogen receptor not associated with thesections (r = hormone binding site, correlated poorly with the

ogen receptor oestrogen receptor content of adjacent sections

rs with oestrogen assayed either by enzyme immunoassay with thetein but with oestrogen receptor monoclonal antibody or by radio-4 cellularity or ligand binding assay. Although the protocol we fol-

lowed for DS immunohistology (Amersham

Table 2 Correlation between immunohistology staining grades obtained with oestrogen receptor and D5 monoclonalantibodies on adjacent cryostat sections

Oes.trogen receptor No of tumoursnionoclonalinimunohistology D5 monoclonal immunohistology gradesgradles 0/+ + +/+ + + Total

0 + 35 32 67++,+++ 21 36 57

Total 56 68 124p > 0.1

738

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from

Immunohistology of oestrogen receptor and DS antigen i

m:x;.tz w

Fig 5 Invasive ductal adenocarcinoma of breast stained withanti-D5 monoclonal antibody. There is strong uniformc!vtoplasmic staining oftumour cells. Stromal cells areunstained. (Immunoperoxidase.)

. V r; V

Fig 6 Stromnal staining by anti-D5 monoclonal antibody.Tunmour cells are unstained. (Immunoperoxidase.)

in breast cancer 739

International plc) gave investigators the option ofusing either frozen sections or paraffin wax sections, itis reported that false positive staining may beobtained with the former (King RJB, personal com-munication 1986). This is unexpected; artefactualresults in immunohistology are more likely with sec-tions of processed fixed tissue than with frozen sec-tions, though paraffin sections of tissue fixed inmethacarn are now advocated for D5 immuno-histology.8 Loss of D5 antigen by diffusion from thefrozen sections is unlikely, and the range of positiveand negative staining within our series of tumoursseems more likely to be attributable to inherentdifferences between tumours rather than to meth-odological problems.

Stromal staining with the D5 antibody has not beenreported previously; we observed it in a small numberof tumours even when the carcinoma cells were nega-tive. Although this may denote oestrogenresponsiveness of the stromal cells in these lesions,nuclear staining of stromal cells with the oestrogenreceptor antibody was never seen.Most of the unoccupied oestrogen receptor within

oestrogen responsive cells is now thought to residewithin the nucleus,9 10 and the oestrogen receptormonoclonal antibody gave consistent nuclear stain-ing. Consistent cytoplasmic staining by the D5 anti-body contributes to the suspicion that the D5 antigenis not associated with the hormone binding activity ofoestrogen receptor, though there is evidence that thepresence of D5 antigen, a 29000 dalton phos-phoprotein, may be a useful indicator of oestrogenresponsiveness.8 Unlike oestrogen receptor, D5immunostaining shows poor correlation with histo-logical differentiation and tumour ploidy (Walker R,personal communication 1986).Our results show a strong positive correlation

between immunohistology with the oestrogen recep-tor monoclonal antibody and the radioligand bindingassay data; this is consistent with other reports.' 1 - 17Our methodology, using cryostat sections for all theimmunohistology and assays, permitted us to exam-ine some of the reasons for discrepant results, particu-larly in tumours where the cellularity is unusually lowor high. Tumours consisting of sparsely distributedoestrogen receptor positive cells on immunohistologymay give misleadingly low or negative radioligandbinding assay results because of dilution by receptornegative stromal elements.'8We conclude that the oestrogen receptor mono-

clonal antibody used in this study is the only reagentcurrently available that permits the histological deter-mination of the oestrogen receptor content of breastcarcinomas. Although the information obtained fromits use is not quantitative, it has the advantages ofgiving simultaneous confirmation that viable tumour

gplpxr- -dP--

Is.

.MRIV,V.11 i

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from

740 Giri, Dangerfield, Lonsdale, Rogers, UnderwoodTable 3 Correlation between immunohistology staining grades obtained with D5 monoclonal antibody and oestrogen receptorcontent ofadjacent cryostat sections assayed by radioligand binding and enzyme immunoassay

No of tumours

Oestrogen receptor content by:DS monoclonal Radioligand assay Enzyme immunoassayimmunohistologygrades Negative Positive Negative Positive

0/+ 40 16 21 12++ +++ 36 32 24 21

Oo5< p < 0o p > 02

is present in the assessed sample; of enabling anassessment to be made of tumour cell heterogeneitywith respect to receptor content; and of avoiding theproblem of false negative results when using homoge-nates due to dilution by stromal elements. Immu-nohistology and enzyme immunoassay with theoestrogen receptor monoclonal antibody also permitthe detection and quantification of oestrogen receptorin tumours from premenopausal women. Endo-genous oestrogens may saturate all the availableoestrogen receptor in these lesions, thus oftenresulting in misleadingly low or negative radioligandbinding assay results.The D5 monoclonal antibody showed poor

correlation with oestrogen receptor content, butthis does not exclude the possibility that positiveimmunostaining for D5 antigen may have clinical andtherapeutic importance.8

We thank Abbott Laboratories Ltd for providingthe oestrogen receptor monoclonal antibody forimmunohistology and enzyme immunoassays, and DrRoger King (Imperial Cancer Research Fund, Lon-don) and Amersham International plc for providingthe D5 monoclonal antibody. This work was sup-ported by the Yorkshire Cancer Research Campaign.DD Giri is a Commonwealth Medical ResearchScholar for whom the Association of CommonwealthUniversities generously provided a research supportgrant.References

I Underwood JCE. Oestrogen receptors in human breast cancer:review of histopathological correlations and critique ofhistochemical methods. Diagn Hisiopaihol 1983;6:1-22.

2 Howanitz PJ, Howanitz JH, Skrodzki CA, Woloszyn T. Proteinmethod influences on calculation of tissue receptorconcentration. Am J Clin Pailhol 1986;85:37-42.

3 Raam SR, Gelman R, Cohen JL, et al. Oestrogen receptor assay:interlaboratory and intralaboratory variations in themeasurement of receptors using the dextran-coated charcoaltechniques: a study sponsored by ECOG. Eur J Cancer198 1;17:643-9.

4 Greene GL, Nolan C, Engler JP, Jensen EV. Monoclonalantibodies to human oestrogen receptor. Proc Nail Acad SciUSA 1980;77:5115-9.

5 Coffer AI, Lewis KM, Brockas AJ, King RJB. Monoclonalantibodies against a component related to soluble oestradiolreceptor. Cancer Res 1985;45:3686-93.

6 Underwood JCE, Dangerfield VJM, Parsons MA. Oestrogenreceptor assay of cryostat sections of human breast carcinomaswith simultaneous quantitative histology. J Clin Pathol1983;36:399-405.

7 Bradford MM. A rapid and sensitive method for the quantitationof microgram quantities of protein utilizing the principle ofprotein-dye binding. Anal Biochem 1976;72:248-54.

8 Cano A, Coffer AI, Adatia R, Millis RR, Rubens RD, King RJB.Histochemical studies with an oestrogen receptor relatedprotein in human breast tumours. Cancer Res 1986;46:6475-80.

9 King WJ, Greene GL. Monoclonal antibodies localize oestrogenreceptor in the nuclei of target cells. Nature 1984;307:745-7.

10 Welshons WV, Lieberman ME, Gorski J. Nuclear localization ofunoccupied oestrogen receptors. Nature 1984;307:747-9.

11 Hawkins RA, Sangster K, Krajewski A. Histochemical detectionof oestrogen receptors in breast carcinoma: a successfultechnique. Br J Cancer 1986;S3:407-10.

12 McCarty KS, Szabo E, Flowers JL, etal. Use of a monoclonalanti-oestrogen receptor antibody in the immunohistochemicalevaluation of human tumours. Cancer Res (Suppl)1986;46:4244-8.

13 DeSombre ER, Thorpe SM, Rose C, et al. Prognostic usefulnessof oestrogen receptor immunocytochemical assays for humanbreast cancer. Cancer Res (Suppl) 1986;46:4256-64.

14 Charpin C, Martin P-M, Jacquenier J, Lavaut MN,Pourreau-Schneider N, Toga M. Oestrogen receptorimmunocytochemical assay (ER-ICA): computerized imageanalysis system, immunoelectron microscopy, and comparisonswith oestradiol binding assays in 1 15 breast carcinomas. CancerRes (Suppl) 1986;46:4271-7.

15 Heubner A, Bech T, Grill H-J, Pollow K. Comparison ofimmunocytochemical oestrogen receptor assay, oestrogenreceptor enzyme immunoassay, and radioligand-labelledoestrogen receptor assay in human breast cancer and uterinetissue. Cancer Res (Suppi) 1986;46:4291-6.

16 Jonat W, Maass H, Stegner HE. Immunohistochemicalmeasurement of oestrogen receptors in breast cancer tissuesamples. Cancer Res (Suppl) 1986;46:4296-8.

17 Ozzello L, De Rosa CM, Konrath JG, Yeager JL, Miller LS.Detection of oestrophilin in frozen sections of breast cancersusing an oestrogen receptor immunocytochemical assay.Cancer Res (Suppi) 1986;46:4303-7.

18 King WJ, DeSombre ER, Jensen EV, Greene GL. Comparison ofimmunocytochemical and steroid-binding assays for oestrogenreceptor in human breast tumours. Cancer Res1 985;45:293-304.

Requests for reprints to: Professor JCE Underwood,Department of Pathology, University of Sheffield MedicalSchool, Beech Hill Road, Sheffield SlO 2RX, England.

on 3 May 2018 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.40.7.734 on 1 July 1987. Dow

nloaded from