[CANCER RESEARCH 46, 6475-6480, December 1986]

Histochemical Studies with an Estrogen Receptor-related Protein in Human

Breast TumorsAntonio Cano,1 Arnold I. Coffer, Remy Adatia, Rosemary R. Millis, Robert D. Rubens, and Roger J. B. King2

Hormone Biochemistry Department, Imperial Cancer Research Fund, P. O. Box 123, Lincoln's Inn Fields, London WC2A 3PX[A. C., A. I. C., R. A., K. J. B. K.J, andClinical Oncology Unit, Guy's Hospital, London SEI 9RT[R. R. M., R. D. R.J, England

ABSTRACT

The histochemical characteristics of a M, 29,000 phosphoproteii.related to estradiol receptor are described in a large series of humanbreast tumors. The antigen was detected with a monoclonal antibody(D5) raised against partially purified human myometrial estradiol receptor. An indirect immunoperoxidase method was used with methacarn-fixed, wax-embedded sections. Quant ¡tation of staining and its reproduc-ibility are described. Results with trucut biopsies agree with those obtained with larger tumor sections. Normal breast is infrequently positive.Histochemical staining is higher in invasive carcinoma than in normalbreast with ductal carcinoma in situ adjacent to infiltrating tumorsexhibiting intermediate values. Furthermore, most in situ carcinomashave a heterogeneous staining pattern. About 20% of invasive tumorsalso exhibit heterogeneity. No simple correlation is seen between stainingand histológica! grade. There are more low-staining tumors in young(<50 yr old) patients than in older women.

Staining correlates with levels of cytosol estradiol receptor but notcytosol progesterone receptor. However, cytosol estradiol receptor-negative, cytosol progesterone receptor-positive tumors tend to have positiveM, 29,000 phosphoprotein levels. Positive staining is associated with ahigher response rate to hormone therapy (50%). None of the negativetumors responded to hormone treatment. With these patients, comparisonof histochemical assay for M, 29,000 phosphoprotein and (3H]estradiol

binding assays indicated that the former was at least as good as the latterassay in predicting hormone response. About 20% of cytosol estradiolreceptor-positive tumors have low M, 29,000 phosphoprotein, and suchtumors have poor response to hormone treatment.

INTRODUCTION

In experiments designed to raise monoclonal antibodiesagainst RE3 from human uterus, we obtained an IgGl mono

clonal antibody that recognized p29 and not the estradiol binding unit (1, 2).The immunogen was myometrial RE partiallypurified by ammonium sulfate precipitation and affinity diromatography on an estradiol hemisuccinate column (3). Antibody screening was by ability to precipitate I25l-labeled estradiolspecifically bound to ammonium sulfate-purified, human myometrial RE. We now know that p29 will complex with REunder certain conditions, one of which is treatment with ammonium sulfate (1,4). It was therefore possible that p29 wasan artefactual contaminant of the RE preparations used in theseexperiments and that it had no biological importance in thecontext of estrogen action. However, several points militatedagainst this view: p29 was only found in RE-positive tissues (4,5); it was quantitatively related to RE but not RP levels (2, 4),and the conditions provoking RE p29 interaction (high salt ortemperature were stimulatory, molybdate inhibitory) were those

Received 3/31/86; revised 6/24/86, 8/15/86; accepted 8/22/86.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1British Council Fellow. Present address: Department of Obstetrics and Gyne-

cology. Facultad de Medicine. Espinardo, Murcia. Spain.2To whom requests for reprints should be addressed.3The abbreviations used are: RE, cytosol estradiol receptor; RP, cytosol

progesterone receptor; PBS, phosphate-buffered saline (153 mM NACI-3 mMKCI-10 mM NajHPO.,-2 HIM KH2PO4, pH 7.4); ELISA, enzyme-linked immu-nosorbent assay; p29, M, 29.000 phosphoprotein.

that activate RE for nuclear binding (1, 4). It is thereforepossible that p29 is either a previously undetected componentof the estradiol receptor machinery or a specific product ofestrogen action.

We have previously reported a pilot study on the histochemical localization of this antigen in human tissues (5). These dataindicated the cytoplasmic location of the antigen in ethanol-fixed, wax-embedded samples and supported the idea of aqualitative and quantitative relationship between p29 and RE.This paper extends those data on 393 tumors with respect to(a) using a different fixative to improve tissue preservation, (b)validation of methodology, (e) quantitative relationship to RE,and (d) clinical usefulness of histochemical assessment of p29in human breast tumors.

MATERIALS AND METHODS

Unless stated otherwise, all procedures were carried out at roomtemperature.

Patients. Tumors were obtained from patients with operable breastcancer treated either by modified radical mastectomy or excision oftumor followed by radiotherapy. The clinical data are derived from asubset of patients who have relapsed and received endocrine therapyalone (either ovarian ablation or tamoxifen either alone or in combination with prednisolone).

Tissues and Staining. Tissues were fixed in methacarn (metha-nol:chloroform:acetic acid; 60:30:10) and processed as described previously (5). This method gave the same staining as the previouslydescribed ethanol fixation (5), but tissue preservation was superior withmethacarn. For routine purposes the indirect immunoperoxidasemethod was used as described previously (5). Sections were stained for1 h with antibody D5 (8 jig of protein per ml) that had been purifiedby ion exchange and Protein A chromatography (1). This preparationhas been stored at 4°Cin solution (400 Mgprotein per ml) for 3 yr

without deterioration. After washing, the sections were incubated with1:50 dilution of peroxidase-conjugated sheep anti-mouse IgG (Amer-sham International, Amersham, United Kingdom) containing 1:25dilution of human serum for 45 min followed by a 10-min incubationwith diaminobenzidine-hydrogen peroxide reagent (5).

In some cases, alkaline phosphatase/antialkaline phosphatase staining was used as described by Cordell et al. (6). The avidin-biotin methodwas also tested as described in the manufacturer's leaflet (Vectastain).

When imidazole was tested as an enhancing agent for the peroxidasereaction, it was added at a final concentration of 10 mM as recommended (7).

Quantitäten of Staining in Breast Tumor Sections. Tumor cellularity,staining intensity, and proportion of positive tumor cells were assessedby eye as previously described (5). The proportion of tumor cells persection was assessed by eye and allocated on a score of 0-6. Stainingintensity was estimated on a 0-3+ scale. The staining index was asimple multiple of the cellularity and staining scores. Two sectionswere stained, and values were calculated as a mean of the two individualvalues. Tumors were defined as being heterogeneous if more than 10%of tumor cells differed in positivity or negativity from the remainingtumor cells. All assessments were made without knowledge of receptorstatus.

Comparison of Open Biopsy and Trucut Samples. Trucut biopsies (8)were obtained from excised tumor, and sections of trucut sample andexcised tumor were both processed. The trucut samples were not takendirectly from patients.

6475

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ESTROGEN RECEPTOR-RELATED PROTEIN IN HUMAN BREAST TUMORS

Assessment of Cellularity by Microscope Grid. The eyepiece of themicroscope was fitted with a graticule that divided each microscopefield into 100 squares. The proportion of tumor cells in each of thesesquares was scored as low (1 point), medium (2 points), or high (3points), depending on whether <30%, 30-60%, or >60% of the areawas occupied by tumor cells. Ten fields were chosen at random for eachsection. Finally, the total score/10 fields was divided by the totalnumber of squares assessed (1000).

Histology. Histological grade of ductal tumors was assessed according to the method of Bloom and Richardson (9).

Receptor Assays. Ligand binding assays were performed as describedpreviously (10). Immunoassay of RE by ELISA was as described in themanufacturer's instructions (Abbott Diagnostics, Ltd., Wokingham,

United Kingdom). Positive tumors had >10 fmol of RE or RP per mgof protein. Protein was measured in the cytosols by the Coomassie bluemethod as described in the manufacturer's leaflet (Biorad, Ltd., Wat

ford, United Kingdom) using bovine serum albumin as standard.ELISA for p29. Aliquots (25 /¿I)of the same cytosols as were used

for receptor assays were incubated for 90 min in 200 n\ of PBS inImmulon 1 microtiter wells (Dynatech Limited, Billingshurst, UnitedKingdom). After washing 3 times with distilled water, 200 ^1 of D5(0.75 Mg/m') were added to each well and incubated at 37°Cfor 2 h.

After three distilled water washes, 200 fi\ of peroxidase-labeled sheepanti-mouse immunoglobulin (Amersham International, United Kingdom) were added to each well [1:500 dilution in PBS:0.05% (v/v)Tween 20] and incubated for l h at 37°C.Wells were washed 3 times

with distilled water and incubated with 200 n\ of peroxidase substrate(50 mg of o-phenylene diamine hydrochloride (Sigma Chemicals Limited, Poole, United Kingdom) in 100 ml of 0.1 M sodium phosphate-citric acid buffer (pH 5.0) containing 20 n\ of hydrogen peroxide (100volume %; Fisons Limited, Loughborough, United Kingdom)]. After20 min the reaction was stopped with 50 nl of 0.5 M sull'urie acid, and

plates were read at 450 nM. Control samples were as described aboveexcept that D5 was omitted from the appropriate incubation medium.Calibration curves were prepared from a myometrial cytosol standardized to an arbitrary number of units of p29.

Clinical Assessment. This was by internationally agreed criteria (11).A minimum period of 6 mo was used before assessment was made.Responders had either disappearance of all known disease (completeresponse) or >50% decrease in measurable lesions with the appearanceof no new lesions (partial response). The no-change category includedtumors in which there was a <50% decrease or <25% increase in size.Response failure defined patients in which (a) some lesions regressedwhile others progressed, (b) new lesions appeared, or (c) all lesionsprogressed.

RESULTS

Staining Method

Thirteen samples were stained both by avidin-biotin andindirect immunoperoxidase methods. The former gave less thana 2-fold increase in staining intensity and, in view of theincreased complexity of the method, was not further used. Theaddition of imidazole to the peroxidase step of the indirectmethod increased staining by about half, but background colorwas also enhanced. Therefore, ¡midazolewas not used routinely.The alkaline phosphatase bridge method gave comparable results to those obtained by indirect immunoperoxidase, and thelatter was used for routine assays.

Validation of Assay

Cellularity. The subjective assessment by eye was comparedwith that obtained by counting cell number per microscope fieldwith the aid of a grid in the eyepiece. A highly significantcorrelation was obtained (P < 0.001) as were the correlationsbetween yield of soluble protein per g of wet weight tumor andCellularity judged either by grid (P = 0.001) or by subjective

assessment (P < 0.01). Reproducibility was reasonable as determined by analysis of samples stained on three differentoccasions. Only one of the ten tumors that were stained onthree different occasions showed a greater than 50% day-to-dayvariation in stain index.

Comparison with P29 Content. A highly significant (P <0.001) correlation was obtained between the histochemical andELISA assay of p29 (Fig. 1).

Comparison of Results Obtained with Open Biopsy and I mailSamples. A good correlation was obtained with the two typesof sample (Fig. 2). One sample pair not included in these datacontained no tumor cells in the trucut sample. Using an arbitrary cut-off value of staining index 3.5 to distinguish positiveand negatives samples, 32 of 36 samples were in agreement. Ofthe 4 discordant results, 2 were due to differences in cellularity,and 2 to stain intensity between the trucut and open biopsysamples.

Analysis of Multicentric Tumors Obtained Concurrently fromthe Same Patient. Ten of 13 patients had separate tumors thatgave less than 2-fold differences in each of cellularity, stainingintensity, and index (Patients 1 to 10; Table 1). Of the threepatients whose tumors showed greater divergence (Patients 11to 13; Table 1), only one (No. 13) would have changed frompositive to negative using a cut-off value for index of 3.5. Withthis particular patient, the discrepancy was mainly due to col

2 12-

10-

8-XUIoz

4z

"<>' 1 3 5 7 9 11 13.i..»...LE p29 CONCENTRATION (ELISA) U/25/*L

Fig. I. Correlation of histochemical and biochemical analysis of p29 in breasttumors. I . units of p29 determined from a myometrial cytosol standard (see"Materials and Methods").

TRUCUT

OPEN BIOPSY

Fig. 2. Comparison of p29 stain index of open biopsy samples and trucutbiopsies of the same sample. A line with a slope = 1 has been included. Linearcorrelation, P< 0.001.

6476

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ESTROGEN RECEPTOR-RELATED PROTEIN IN HUMAN BREAST TUMORS

Table 1 Comparison of separate tumors obtained concurrently from the samepatient

Patient1234S678910111213Histológica!grade"32»222Lobular*23*23*23*12»22*32»221"22»2233*HistochemistryCellularity3.53.53.54333.533.532.543243.5443.552.82.533.5252Intensity0.5000.31.211.51.32.322.53.233.22.33.52.533.21.83.51.53.552.321.5Stainindex1.8001.23.635.33.98.166.212.896.49.212.3101211910.63.810.510.54.6103

" Ductal unless stated otherwise.' Second primary.

% WITH HIGHINDEX (»3.5)

80

60

40

20

STAIN INDEX8

180 123 35 24

PHENOTYPE (RE.RP)

Fig. 3. Influence of receptor phenotype on p29 staining. Positive tumorscontained > 10 fmol of either RE (first designation) or RP (second designation)per mg of protein. Stain indices are given as mean ±SE with the number ofsamples given in each block. Stain index comparisons by Student's t test. ++versus H—.not significant; ++ versus —,P< 0.001; ++ versus-+, not significant;— versus —+,P < 0.05. High staining comparisons by x! analysis. ++ versusH—,P< 0.01; ++ versus —, P< 0.001; ++ versus —Knot significant; — versus-+,P< 0.001. Not significant. P > 0.05.

hilarity rather than stain intensity. The within-patient variations were not related to histológica! grade.

Comparison of Stain Index with RE and RP

Most tumors that contained both RE and RP had a highstaining index, although 20% had low p29 (Fig. 3). The RE+p29 low tumors were all positive when tested by the AbbottELISA kit for RE, which rules out the possibility of a defect inour RE assay. Tumors having little, if any, RE and RP had alow index, and 85% had an index below the arbitrarily set levelof 3.5. Of the five RE—RP—tumors with an index greater than

3.5, only one (3%) had an index greater than 5.The RE+RP— tumors had a lower mean index (difference

not significant) and a smaller proportion of high index tumors(P < 0.01) than the RE+RP+ counterparts (Fig. 3). As p29correlates with RE but not RP levels (see below), the lowerstaining obtained with RE+RP— samples probably reflects thelower RE levels of these tumors as compared to RE+RP+samples. The latter had a mean RE value of 269 fmol/mg ofprotein as compared with 187 fmol/mg of protein for theRE+RP- samples (P < 0.01).

Of particular interest were the RE-RP+ tumors which hada significantly higher index (P < 0.05) and a higher proportionof high value samples (P < 0.001) than the RE-RP- samples.Of 17 RE-RP+ tumors assayed both by histochemistry and

Abbott RE ELISA, 7 were positive by ELISA and 12 byhistochemistry. These data suggest that such tumors are amixture of RE+ and RE—samples.

Heterogeneous staining patterns were obtained in about aquarter of RE+RP+ and RE+RP- tumors and in about one-

third of the cases of the other two phenotypes.There was no significant relationship between the amount of

RP (fmol/mg of protein) and stain index (P> 0.1), whereas theindex was linearly correlated with RE (P< 0.05). The statisticalsignificance of this correlation was greater if limited to tumorswith RE below 100 fmol/mg of protein (P< 0.001). These dataare presented in histogram form in Fig. 4. Because of thediscrepant behavior of the RE—RP+ tumors (Fig. 3), these

samples have been omitted from the data in Fig. 4. The relationship is most evident in the RE range of 0-50 fmol/mg ofprotein; above this level the stain index tends to plateau, although the percentage of tumors with high index increases upto the 101-200 fmol/RE mg of protein group (Fig. 4). Statistically significant relationships between RE/mg of protein and

% WITH HIGH (13 5)INDEX (•-•)

10080«04020

0

% WITH

HETEROGENEOUSPATTERN (o-o)

6040200

STAIN INDEX

12

10

8

6420

1

50

50021-50 101-20011-20 51-100 201-500

RE (fmol/mg protein)

Fig. 4. Correlation of p29 staining with RE levels. Stain indices are given asmean ±SE with the number of samples given in each block.

6477

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ESTROGEN RECEPTOR-RELATED PROTEIN IN HUMAN BREAST TUMORS

p29 were not obtained if stain intensity rather than index wasused.

Heterogeneous staining patterns are most common in thelow RE samples but are seen in all the RE categories shown inFig. 4. The lower percentage of heterogeneous tumors in thevery high RE tumors (>500 fmol/mg of protein) than in the101-200 or 201-500 fmol groups did not attain statisticalsignificance.

Influence of Clinical Features on Staining Patterns

Patient Age. The staining index was lower in the <50 yr ascompared to the 50-80 yr age group (P < 0.05), and the formercontained more low staining samples than the latter, althoughthis only approached statistical significance (0.1 > P > 0.05).The decrease in index and percentage of tumors with low stainin the very old patients (>80 yr) did not attain statisticalsignificance.

The percentage of tumors with heterogeneous staining pattern did not change with patient age (Table 2).

Histology. There were no differences between ductal andlobular tumors, whereas the mucoid tumors tended to havelower staining than either ductal (P < 0.01) or lobular (0.1 > P> 0.05) tumors, although the very small number of mucoidtumors in this analysis should be stressed (Table 2). Grade 2tumors had a higher stain index than Grade 3 tumors (P <0.01), whereas Grade 1 tumors differed from neither of thepoorer differentiated groups.

Of special interest is the lower staining intensity of normalbreast epithelium (both ductal and lobular) as compared to thatseen in invasive tumors, with adjacent carcinoma in situ exhibiting intermediate levels (Fig. 5). These data for normal epithelium were obtained from tissue adjacent to or interspersed withthe tumor, but the same negative results have been obtainedwith normal breast adjacent to fibroadenomas in premenopau-sal women (data not shown).

Clinical Data

The overall picture on the limited number of patients indicates that the histochemical method predicts response to hormone therapy in a worthwhile way (Table 3). Of the positivepatients, 50% had either a complete or partial response and12% failing treatment. The predictive value of RE assays carriedout on the same samples was less precise with 18% failing and41% responding to treatment. RP assays gave similar results tothose for p29. The histochemical negative patients had a high

Table 2 Effect of age and histological type on staining

No. ofPercentage

Age(yr)<5050-80>80HistologyDuctalGlG2G3TotalLobularMucoidsamples1002741921141137299313Stain

index"5.4

±0.5*6.2±0.35.2±1.05.4

±I.I6.7±0.4'5.3±0.4'6.4±0.574.9

±0.62.0±l.S7Low

index*43e3253*383140363966Heterogeneous36323233303332420

" Mean ±SE.* Less than 3.5.c x2 comparison with 50-80 yr-old group: 0.1 > P > 0.05.* x2 comparison with 50-80 yr-old group: P > 0.1.' Student's t test, G2 versus G3, P < 0.01.1 Ductal versus mucoid, P < 0.01 (Student's I test).* Age < 50 yr versus 50-80 yr P < 0.05 (Student's t test).

STAIN INDEX

7654321

0Normal

Carcinomain-situ

121MlJL1371

2 3

Grade Carcinoma

Fig. 5. p29 staining of normal and neoplastic ductal epithelium. Results aregiven as mean ±SE with the number of samples given in each block. Forstatistical evaluations, see text.

Table 3 Response to hormone therapyComparison of predictive value of histochemical assessment of p29 and bio

chemical assay of RE and RP is shown.

Phenotype0 Assay

No. of patients/total no.

Respond No change FailPositive Histochemistry 8/16(50)* 6/16(38) 2/16(12)

[3H]Estradiol 9/22(41) 9/22(41) 4/22(18)[3H]Progesterone 9/16(56) 6/16(38) 1/16(6)

Intermediate Histochemistry 1/5 (20)[3H]Estradiol['HJProgesterone 0/1

Negative Histochemistry 0/6 (0)['HJEstradiol 0/3 (0)[3H]Progesterone 0/8 (0)

1/5(20)

0/1

3/5(60)

1/1

2/6(33) 4/6(67)0/3(0) 3/3(100)3/8(38) 5/8(63)

°Positive: histochemistry index, >6; [3H]estradiol and [3H]progesterone, >19fmol/mg of protein. Intermediate: histochemistry index, 4-6; [3H]estradiol and[3H]progesterone, 10-19 fmol/mg of protein. Negative: histochemistry index, <4;[3H]estradiol and [3H]progesterone, <10 fmol/mg of protein.

* Numbers in parentheses, percentage.

Table 4 Response to hormone therapy according to RE or RP and p29 phenotype

Phenotype(no. of patients/total no. of patients)"

ResponseNochangeFailResponseNo

changeFailRE+p29+8/14

(57)*6/14

(43)0/14(0)RP+p29+8/13(61)4/13(31)1/13(8)RE+p29-1/8(13)3/8(37)4/8(50)RP+p29-1/3(33)2/3

(67)0/3(0)RE-p29-0/2

(0)0/2(0)2/2(100)RP-p29-0/7

(0)1/7(14)6/7

(86)RE-p29+0/10/11/1RP-p29+0/2

(0)2/2(100)0/2

(0)" Positive: p29 stain index >6; RE, RP >19 fmol/mg of protein. Negative: p29

stain index <6; RE, RP <20 fmol/mg of protein.* Numbers in parentheses, percentage.

failure rate (67%) and none responded. The three RE-negativetumors all failed treatment. RP assays gave similar results tothose for p29. The tumors with intermediate staining levelspredominantly (60%) failed treatment. The cut-off values usedhere were set to optimize the separation of responders andfailures. In fact, few of the hormone-insensitive tumors werecompletely negative for p29.

Some RE-positive tumors contain low or undetectable levelsof p29 (Fig. 4). Such tumors tend to be insensitive to hormonetreatment (Table 4). None of the RE+ p29+ tumors failed torespond. No obvious pattern is discernible with RP and p29phenotypes apart from the RP—p29—tumors being unrespon

sive in contrast to the good response rate for RP+ p29+ tumors(Table 4). Of particular interest would be the behavior of theRE+ RP—samples. Five of the seven tumors on which we have

clinical data were p29 negative and did not respond to treatment. The two p29 positive, RE+ RP- samples had stabilized

6478

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ESTROGEN RECEPTOR-RELATED PROTEIN IN HUMAN BREAST TUMORS

disease. We have no clinical data on RE—RP+ tumors with

respect to p29.There are too few tumors with heterogeneous staining pat

terns to warrant comment on their clinical behavior.

DISCUSSION

M, 29,000 phosphoprotein recognized by D5 is clearly relatedto RE but in what way is unknown. The two possibilities areeither that it is a previously undetected component of theestradiol receptor machinery or that it is a product of estrogenaction. The former view is compatible with the purificationmethod used to generate the antigen and with p29's facility of

interacting with activated RE (1, 4). However, p29 cannot be aprecursor or processed product of the estradiol binding unit ofRE as p29 and RE are translated from separate mRNAs.4 Thep29 could be a non-hormone-binding component of the receptormechanism. The existence of such components has gained somecredence from recent work with a M, 90,000 protein associatedwith 8S receptors for all steroid classes (12-15) and a A/r108,000 protein having similarities to the B subunit of chickoviduct RP ( 16). We feel that comparisons between our antigenand either the A/r 90,000 or M, 108,000 antigens are unrewarding, as those antigens are bigger than p29 and are also presentin hormone-insensitive cells.

Our preliminary data do not support the idea that p29 is anestrogen-induced protein. The p29 content of breast tumorsdoes not correlate with RP, although p29 is markedly lower inglandular epithelium of luteal phase endometrium than in proliferative endometrium, suggesting an inhibitory effect of progesterone.5 Overall, we feel that the present data are too incom

plete to ascribe a biological function to p29, but both thequalitative and quantitative relationship to RE and the clinicalresults on response to hormone therapy convince us of itsrelevance to estrogen action. The histochemical results reportedhere support that view; the stain index correlates with RE butnot RP and increases with increasing RE content. The lattercorrelation is strongest in the range of 0-100 fmol of RE permg of protein which probably reflects the insensitivity of thehistochemical assay at high p29 levels. The same observationhas been made with the histochemical assessment of RE (17).Clinically, p29 positive tumors show a good response rate tohormone treatment. Of additional interest are the RE-positivetumors that have low p29. These tumors were positive whentested with the Abbott ELISA kit for RE, so they cannot be anartefact of poor RE assays. It is tempting to speculate that suchtumors have a defect in their estrogen response mechanism.The very limited data on hormone sensitivity of such tumorsare compatible with that view. Furthermore, we have not yetseen any such phenotypes in normal endometrial cells (data notshown). A better comparison would have been between normaland neoplastic breast, but this has not been possible as we onlyrarely see positive staining in normal epithelium and haveinsufficient material to perform RE assays. However, the published biochemical (18, 19) and autoradiographic (20) data onlow RE in normal human breast epithelium are in agreementwith our p29 data.

The positive staining of many RE—RP+ tumors is also ofinterest as it correlates with the good clinical response of suchtumors to hormone therapy (21, 22) and to the idea that suchsamples may have occupied RE (23).

Heterogeneous staining patterns were most common in tu-

' S. Green, P. Chambón, and R. J. B. King, unpublished observations.5 M. Padwick el al., manuscript in preparation.

mors with very low RE, but about a third of all tumors exhibitedsuch a pattern irrespective of RE content. The heterogeneousnature of human breast tumors has been noted for many features(24-28) including RE (17, 20, 29-32). Identification of heterogeneous tumors by histochemical means may provide additional information of clinical use in deciding types of treatmentto be used with individual patients. The proportion of tumorswith heterogeneous staining pattern was not influenced bypatient age or histological type, whereas young patients (<50yr) tended to have both a lower mean stain index and a lowerpercentage of high staining tumors than the 50-80 yr age group.This parallels the experience with RE assays (33) as does thesimilarity of ductal and lobular tumors in our series (34). Adiscrepancy between histochemistry and RE occurs with respectto histological Grade 1 tumors. While Grade 3 tumors have alower stain index than the Grade 2 tumors (which agrees withthe receptor data), Grade 1 tumors have p29 values similar toGrade 3 samples (which differs from the experience with receptors) (35).

Normal breast epithelium (ductal and alveolar) is only rarelypositive and never very strongly staining, whereas the invasivetumors are often strongly positive. This parallels the patternseen with the histochemistry of RE with the Abbott antibody.6

Clearly, there must be an intermediate stage, and our limiteddata suggest that carcinoma /// situ might represent such astage. The cause of this amplified staining is not known, and itwould be of interest to study the behavior of various hyperplasticbreast epithelia and pure in situ carcinoma.

Despite the somewhat subjective nature of the histochemicalassay, it is reproducible and correlates with p29 content ofcytosols determined by ELISA. Furthermore, there is reasonable agreement in most, but not all, multiple assays from thesame patient. Of practical importance is the observation that,provided an assessable number of tumor cells are present,analyses on trucut biopsies agree with those obtained on largersections. This should facilitate data accumulation on small orinaccessible tumors or cytological samples.

There appears to be an important clinical correlation withp29 levels. The low failure and high response rate of p29+tumors and high failure rate of p29- tumors to hormone

manipulation suggests a clinial role for the assay of this antigen.The cut-off values used in this study are determined empiricallyin order to optimize the distinction between responders andfailures. In fact, none of the hormone-insensitive tumors wascompletely negative for p29. A similar point has been made byPertschuk et al. (31) for RE histochemistry. They only found12 completely negative tumors of the 25 that failed treatment.

The clinical results reported here for histochemical assay ofp29 are similar to those reported for histochemical RE assay(17, 31). Pertschuk et al. (31) found 5 responders and 3 no-change patients in their 9 positive samples (89%) as comparedwith 8 responders and 6 no-change patients of 16 samples(88%) in our analysis. Like us, they also identified a group oftumors with intermediate staining levels that exhibited a heterogeneous response pattern. Their 21 negative samples contained one responder compared with 0 of 6 patients in ourseries.

In our hands, the histochemical assay of p29 is at least asgood as the RE assay in identifying the responders in thepositive patients. The small number of negative samples makescomparison difficult. Neither assay detected any responders inthe negative group, whereas histochemistry picked up 2 no-

6 K. S. McCarty, personal communication.

6479

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ESTROGEN RECEPTOR-RELATED PROTEIN IN HUMAN BREAST TUMORS

change patients compared to none in the RE group. Histochem-istry identifies more negative tumors than the RE assay. Fewof the RE+ p29—tumors so identified are hormone responsive

(13%). On the other hand, none of the RE+ p29+ tumors failedto regress. Of particular interest is the clinical behavior ofRE+RP- and RE-RP+ tumors. We have no clinical data for

the latter phenotype, and the results on the former are toosparse to warrant comment. The biochemical data clearly showno relationship between p29 and RP (Fig. 3; Ref. 2), whereasthe clinical results indicate some similarity in the predictivevalue of the two analyses. However, given the limited amountof clinical data it is not possible to say if the histochemicalassay of p29 is an alternative to RE and RP or whether itsimply provides additional clinical information.

Thus far, we have no clinical data on the responses of heter-ogeneously staining tumors. Such data will be awaited withinterest as they may provide evidence as to whether such tumorsbehave differently to the more homogeneous tumors. The possibility exists that grossly heterogeneous tumors would be bettertreated with combined hormone and chemotherapy.

ACKNOWLEDGMENTS

We thank Christine Joyce, Steven Edmeades, and Paul Coffer forhelp with patient identification and Cheryl Gillett, Cus Kawenga, andMarian Egan for the preparation and sectioning of tumors. AmershamInternational pic are thanked for their help and advice.

REFERENCES

1. Coffer, A. I., Lewis, K. M., Brockas, A. J., and King, R. 3. B. Monoclonalantibodies against a component related to soluble estrogen receptor. CancerRes.. 45: 3686-3693, 1985.

2. Coffer, A. I., Spiller, G. H., Lewis, K. M., and King, R. J. B. Immunoradi-ometric studies with monoclonal antibody against a component related tohuman estrogen receptor. Cancer Res.. 45: 3694-3698, 1985.

3. Coffer, A. I., Milton, P. J. D., Pryse-Davies, J., and King, R. J. B. Purificationof oestradiol receptor from human uterus by affinity chromatography. Mol.Cell. Endocrino!., 6: 231-246, 1977.

4. King, R. J. B., and Coffer, A. I. The generation of antibodies against partiallypurified oestradiol receptor from human myometrium. In: V. C. Jordan (ed.),Estrogen/Amiestrogen Action in Breast Cancer Therapy. Madison, WI:University of Wisconsin Press, pp 375-394, 1986.

5. King. R. J. B., Coffer, A. I., Gilbert, J., Lewis, K.. Nash, R., Millis, R., Raju,S., and Taylor, R. W. Histochemical studies with a monoclonal antibodyraised against a partially purified soluble estradici receptor preparation fromhuman myometrium. Cancer Res., 45: 5728-5733, 1985.

6. Cordeil, J. L., Falini, B., Erber, W. N., Ghosh, A. K., Abdulaziz, Z.,MacOonald, S., Pulford, K. A. F., Stein, H., and Mason, D. Y. Immunoen-zymatic labeling of monoclonal antibodies using immune complexes ofalkaline phosphatase and monoclonal anti-alkaline phosphatase (APAAPcomplexes). J. Histochem. Cytochem., 32: 219-229, 1984.

7. Trojanowski, J. Q., Obrocka, M. A., and Lee, V. M.-Y. A comparison ofeight different chromogen protocols for the demonstration of immunoreac-tive neurofilaments or glial filaments in rat cerebellum using the peroxidase-antiperoxidase method and monoclonal antibodies. J. Histochem. Cytochem., 31: 1217-1223, 1983.

8. Millis, R. R. Needle biopsy of the breast. In: R. W. McDivitt, L. Ozello, andN. Kaufman (eds.). The Breast, pp. 186-203. Baltimore: Williams & Wilkins,1984.

9. Bloom, H. J. G., and Richardson, W. W. Histológica! grading and prognosisin breast cancer. A study of 1409 cases of which 359 have been followed for15 years. Br. J. Cancer, //: 359-377, 1957.

10. King, R. J. B., Hayward, J. L., Kumaoka, S., and Yamamoto, H. Comparisonof soluble oestrogen and progestin receptor content of primary breast tumoursfrom Japan and Britain. Eur. J. Cancer, 13:967-970, 1977.

11. Hayward, J. L., Carbone, P. P., Heuson, J.-C, Kumaoka, S., Segaloff, A.,and Rubens, R. D. Assessment of response to therapy in advanced breastcancer. Eur. J. Cancer, 13: 89-94, 1977.

12. Catelli, M. G., Binart, N., Jung-Testas, L, Renoir, J. M., Baulieu, E.-E.,Feramisco, J. R., and Welch, W. J. The common 90-kd protein component

of non-transformed '8S' steroid receptors is a heat-shock protein. EMBO J.,¿•3131-3135,1985.

13. Housley, P. R., Sanchez, E. R., Westphal, H. M., Beato, M., and Pratt, W.B. The molybdate-stabilized L-cell glucocorticoid receptor isolated by affinitychromatography or with a monoclonal antibody is associated with a 90-92-kDa nonsteroid-binding phosphoprotein. J. Biol. Chem., 260:13810-13817,1985.

14. Joab, I., Radanyi, C., Renoir, M., Buchou, T., Catelli, M.-G., Binart, N.,Mester, J., and Baulieu, E.-E. Common non-hormone binding component innon-transformed chick oviduct receptors of four steroid hormones. Nature(Lond.), 308: 850-853, 1984.

15. Schuh, S., Yonemoto, W., Brugge, J., Bauer, V. J., Riehl, R. M., Sullivan,W. P., and Toft, D. O. A 90,000-dalton binding protein common to bothsteroid receptors and the Rous sarcoma virus transforming protein, pp60v *".J. Biol. Chem., 26: 14292-14296, 1985.

16. Peleg, S., Schrader, W. T., Edwards, D. P., McGuire, W. L., and O'Malley,B. W. Immunologie detection of a protein homologous to chicken progesterone receptor B subunit. J. Biol. Chem., 260:8492-8501, 1985.

17. McCarty, K. S., Miller, L. S., Cox, E. B., Konrath, J., and McCarry, K. S.Estrogen receptor analyses. Correlation of biochemical and immunohisto-chemical methods using monoclonal antireceptor antibodies. Arch. Pathol.Lab. Med., 109: 716-721, 1985.

18. Hahnel, R., and Vivian, A. B. Biochemical experience with the estimation ofestrogen receptors in human breast carcinoma. In: W. L. McGuire, P. P.Carbone, and E. P. Vollmer (eds.), Estrogen Receptors in Human BreastCancer, pp. 205-235. New York: Raven Press, 1975.

19. Leung, B. S., Moseley, H. S., Davenport, C. E., Krippaehne, W. W., andFletcher, W. S. Estrogen receptor in prediction of clinical responses toendocrine ablation. In: W. L. McGuire, P. P. Carbone, and E P. Vollmer(eds.), Estrogen Receptors in Human Breast Cancer, pp. 107-129. NewYork: Raven Press, 1975.

20. Buell, R. H., and Tremblay, G. The localization of 3H-estradiol in estrogenreceptor-positive human mammary carcinoma as visualized by thaw-mountautoradiography. Cancer (Phila.), 51: 1625-1630, 1983.

21. Hawkins, R. A., Roberts, M. M., and Forrest, A. P. M. Oestrogen receptorsand breast cancer: current status. Br. J. Surg., 67: 153-169, 1980.

22. Sedlacek, S. M., and Horwitz, K. B. The role of progestins and progesteronereceptors in the treatment of breast cancer. Steroids, 44: 467-484, 1984.

23. Sarrif, A. M., and Durant, J. R. Evidence that estrogen-receptor-negative,progesterone-receptor-positive breast and ovarian carcinomas contain estrogen receptor. Cancer (Phila.), 48: 1215-1220, 1981.

24. Bocker, W., Klaubert, A., Bahnsen, J., Schweikhart, G., Pollow, K., Mitze,M., Kreienberg, R., Beck, T., and Stegner, H.-E. Peanut lectin histochemistryof 120 mammary carcinomas and its relation to tumor type, grading, staging,and receptor status. Virchows Arch. Pathol. Anat., 403:149-161, 1984.

25. Ellis, I. O., Hinten, C. P., MacNay, J., Elston, C. W., Robins, A., Owainati,A. A., Blarney, R. W., Baldwin, R. W., and Ferry, B. Immunocytochemicalstaining of breast carcinoma with the monoclonal antibody NCRC 11: a newprognostic indicator. Br. Med. J., 290: 881-883, 1985.

26. Garcia, M., Salazar-Retana, S., Richer, G., Domerque, J., Capony, F., Pujol,H., Laffargue, F., Pau, B., and Rochefort, H. Immunohistochemical detectionof the estrogen-regulated 52,000 mol. wt. protein in primary breast cancersbut not in normal breast and uterus. J. Clin. Endocrinol. Metab., 59: 564-566, 1984.

27. Hand, P. H., Nuti, M., Colcher, D., and Schlom, J. Definition of antigenicheterogeneity and modulation among human mammary carcinoma cell populations using monoclonal antibodies to tumor-associated antigens. CancerRes., «:728-735, 1983.

28. Wilkinson, M. J., Howell, A., Harris, M., Taylor-Papadimitriou, J., Swindell,R., and Sellwood, R. A. The prognostic significance of two epithelial membrane antigens expressed by human mammary carcinomas. Int. J. Cancer,33: 299-304, 1984.

29. King, W. J., DeSombre, E. R., Jensen, E. V., and Greene, G. L. Comparisonof immunocytochemical and steroid-binding assays for estrogen receptor inhuman breast tumors. Cancer Res., 45: 293-304, 1985.

30. Kodama, F., Greene, G. L., and Salmon, S. E. Relation of estrogen receptorexpression to clonal growth and antiestrogen effects on human breast cancercells. Cancer Res., 45: 2720-2724, 1985.

31. Pertschuk, L. P., Eisenberg, K. B., Carter, A. C., and Feldman, J. G.Immunohistologic localization of estrogen receptors in breast cancer withmonoclonal antibodies. Cancer (Phila.), 55: 1513-1518, 1985.

32. Shimada, A., Kimura, S., Abe, K., Nagasaki, K., Adachi, L, Yamaguchi, K.,Suzuki, M., Nakajima, T., and Miller, L. S. Immunocytochemical stainingof estrogen receptor in paraffin sections of human breast cancer by use ofmonoclonal antibody: comparison with that in frozen sections. Proc. Nati.Acad. Sci. USA, 82: 4803-4807, 1985.

33. Braunsberg, H., James, V. H. T., Jamieson, C. W., Desai, S., Carter, A. E.,and Hulbert, M. Effect of age and menopausa! status on estimates ofoestrogen binding by human malignant breast tumours. Br. Med. J., 4: 745-747, 1974.

34. Millis, R. R. Correlation of hormone receptors with pathological features inhuman breast cancer. Cancer (Phila.), 46: 2869-2871, 1980.

35. King, R. J. B. Analysis of estradiol and progesterone receptors in early andadvanced breast tumors. Cancer (Phila.), 46: 2818-2821, 1980.

6480

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1986;46:6475-6480. Cancer Res   Antonio Cano, Arnold I. Coffer, Remy Adatia, et al.   Protein in Human Breast TumorsHistochemical Studies with an Estrogen Receptor-related

  Updated version

  http://cancerres.aacrjournals.org/content/46/12_Part_1/6475

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/46/12_Part_1/6475To request permission to re-use all or part of this article, use this link

on May 24, 2020. © 1986 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from