clin cytokeratin intermediate filament expression benign ...heatley, maxwell, whiteside, toner...

2J Clin Pathol 1995;48:26-32

Cytokeratin intermediate filament expression inbenign and malignant breast disease

M Heatley, P Maxwell, C Whiteside, P Toner

AbstractAim-To carry out a comprehensivestudy of cytokeratin expression in benignand malignant breast epithelium andbreast myoepithelial cells; to examinechanges in the cytokeratin profile inmalignant and benign epithelium and incarcinomas of increasing histologicalgrade.Methods-Frozen sections from fibro-adenomas (19 cases), fibrocystic disease(19 cases), and infiltrating ductal (68cases), lobular (seven cases), and muci-nous carcinomas (three cases) wereexamined using a panel of monoclonalantibodies.Results-The luminal epithelium in allfibroadenomas and all cases of fibrocysticdisease, as well as tumour cells in mostcarcinomas, reacted with the specificantibodies to cytokeratins 7, 8, 18, and 19and to antibodies which included thesecytokeratins in their specificities (Cam5 2, AEI, AE3, RCK102, and LP34). In afew ductal carcinomas none of thetumour cells reacted for cytokeratins 7, 8,or 18. Three ductal carcinomas expressedcytokeratin 14. Only occasional casesexpressed cytokeratins 3, 4, 10, and 13.Antibodies which included cytokeratins 5and 14 in their specificities detectedmyoepithelial cells less efficiently thanantiactin antibodies.Conclusion-The cytokeratin profiles inthe luminal epithelium in benign breastdisease and in tumour cells in most carci-nomas are similar in most cases. Somecarcinomas, however, are negative forcytokeratins 7, 8, or 18. This may providea means of predicting the biologicalbehaviour of a histologically borderlinelesion.(7 Clin Pathol 1995;48:26-32)

Keywords: Breast, cytoskeletal proteins, cytokeratins.

Department ofPathology, TheRoyal Group ofHospitals Trust,Grosvenor Road,Belfast BT12 6BLM HeatleyP MaxwellC WhitesideP TonerCorrespondence to:Mr P Maxwell.Accepted for publicationJuly 1994

The cytoskeleton of eukaryotic cells containsintermediate filaments, which measure 7-11nm in diameter as well as microtubules andmicrofilaments.1 Depending on their relativemolecular mass and protein structure, cyto-plasmic intermediate filament proteins havebeen categorised into five classes. The twentycytokeratin intermediate filament proteins are

typically expressed in epithelial cells and incarcinomas.2-' Between two and 10 cytoker-atins are typically expressed by epithelial cells

depending on the cell type, growth environ-ment in the tissue, embryonic development,state of tissue development, and the presenceof disease.6

Whilst carcinomas usually express thecytokeratin profile of their tissue of origin,' 6exceptions may occur. Tumours may eitherbegin to express a novel cytokeratin or, alter-natively, may cease to express cytokeratinswhich are normally present in their tissue oforigin.7 Furthermore, changes in the profile ofcytokeratin expression have been found withincreasing tumour grade. These changes mayalso be caused by alterations in the epitopeconfiguration of some cytokeratins.8

This study was carried out to determinewhether cytokeratin expression can be used todistinguish malignant tumours of the breastfrom benign conditions. The value of cytoker-atm expression in predicting patient prognosiswas also studied, by seeking an associationbetween the expression of different cytoker-atins and tumour grade, in ductal carcinomas.The expression of cytokeratins in breastmyoepithelium and in a group of unusualtumours is also described.

As routine tissue processing with formalinbased fixatives results in denaturing of anti-gens, this study was performed on frozensection material. Although resulting in someloss in the quality of morphology, the useof frozen sections provides the optimalmeans for preserving immunoreactivity.Alcohol fixation and freeze drying followedby embedding the material in paraffin waxresult in only mild antigenetic distortion andgive good morphology.3 1' 12

MethodsFresh specimens of breast tissue, includingthose from lumpectomies and mastectomiesfor malignant conditions and local excisionsfor benign breast disease, were transported onice from the operating theatre to thehistopathology laboratory. In most cases aprevious fine needle aspiration biopsy hadbeen performed and a diagnosis of benignor malignant breast disease had beenestablished. On arrival in the laboratory,the resection margins were marked withindian ink. Impalpable lesions were fixedin 10% unbuffered formol saline before sec-tioning and were excluded from the study.All other specimens were sectioned at 1 cmintervals.A small portion of tissue measuring 1 x 1

x 0-5 cm was excised and immediately

26

on June 24, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.48.1.26 on 1 January 1995. D

ownloaded from

http://jcp.bmj.com/

Expression of cytokeratins in breast disease

Table 1 Specificities of antibodies used

Dilution onAntibody Specificity Supplier cryostat sections

RCK105 Cytokeratin 7 Euro- 1 in 20Diagnostica bv

BA17 Cytokeratin 19 Dakopatts 1 in 10M20 Cytokeratin 8 ICN 1 in 20RGE53 Cytokeratin 18 ICN 1 in 20Cam 5-2 Cytokeratins Beckton 1 in 10

8, 18, 19 DickensonAnticytokeratin 14 Specific antibody to Cymbus 1 in 10

cytokeratin 14RCK102 Cytokeratins Euro- 1 in 20

5, 8 Diagnostica bvAE1 Cytokeratins ICN 1 in 50

10, 14, 15, 16, 19AE3 Cytokeratins ICN 1 in 50

1-8AE2 Cytokeratins ICN 1 in 50

1, 2, 10, 11LP34 Cytokeratins Dakopatts 1 in 10

6, 18AE5 Cytokeratin 3 ICN 1 in 50AE8 Cytokeratin 13 ICN 1 in 50RSKE60 Cytokeratin 10 Euro- 1 in 20

Diagnostica bv6B10 Cytokeratin 4 Euro- 1 in 20

Diagnostica bv

Table 2 Total number of cases studied in each group

Type of disease No. ofpatients

Infiltrating ductal carcinoma 68Lobular carcinoma 7Mucinous carcinoma 3Phyllodes tumour (benign) 2Phyllodes tumour (malignant) 1Fibroadenoma 19Fibrocystic disease 19

plunged into pre-cooled isopentane (Prolabo,Paris, France). Sections were cut at 5 ,umintervals on a Bright cryostat. Cytokeratinswere detected on cryostat sections using mono-clonal antibodies at the dilutions shown intable 1, using the indirect peroxidase method.

Following fixation, gross dissection of thespecimen was completed the following morn-

ing, when representative blocks were taken forhistological examination. The definitive diag-nosis was made on 4 um paraffin sectionsstained with haematoxylin and eosin and thetumours typed as infiltrating ductal, lobular,

1J

A~ , o 4

Figure 1 Staining of the luminal epithelium in thisfibroadenoma is seen with an

antibody specific to cytokeratin 8. The myoepithelium does not react with this antibody(frozen section x 240).

or mucinous carcinomas. Ductal carcinomaswere graded according to the system devisedby Bloom and Richardson"3 and modified byElston et al.'4 For statistical analysis, 12 casesof fibroadenosis, four of fibrocystic disease,one of sclerosing adenosis, one of duct ecta-sia, and one histologically normal breast weregrouped together under the heading fibrocysticdisease. The haematoxylin and eosin stainedfrozen sections in each of these cases wereexamined and compared with the paraffin sec-tions to ensure that the appearances of thefrozen sections were representative of those inthe breast as a whole.

In each case slides were reviewed on a LeitzLaborlux K microscope. A positive reactionwith each of the antibodies was noted whenthere was distinct brown staining in theepithelial cells in breast sections. The propor-tion of cells reacting with the antibody wasestimated on a semiquantitative scale. Caseswere graded between one and five, dependingwhether under 25, 26-50, 51-75, 76-99, or100% of the epithelial or tumour cells reactedwith the antibody under study. If no stainingwas present in the epithelium, the case wasgraded as zero.The presence of a reaction in myoepithelial

cells in cases of fibrocystic disease, fibroade-nomas, and surrounding foci of ductal carci-noma in situ, adjacent to 22 of the invasivetumours, was noted and the proportion ofthese cells reacting with each antibodyestimated. In keeping with the protocol forexamining breast specimens described aboveno case of ductal carcinoma in situ was sam-pled without associated invasion.The proportion of tumour cells reacting

with each antibody in invasive ductal carcino-mas was compared with the proportion ofepithelial cells in the benign conditions todetermine whether any difference was statisti-cally significant. A comparison was also madebetween the proportion of cells reacting infibroadenomas and in fibrocystic disease.Both of these comparisons were carried outusing the Mann-Whitney U test. If either ofthe analyses resulted in a statistically signifi-cant difference, the proportion of cells react-ing in ductal carcinomas was compared,firstly, with the proportion reacting in thefibroadenomas and, secondly, with the pro-portion in fibrocystic disease, using theMann-Whitney U test. This latter analysiswas performed to determine which of thebenign conditions resulted in the differencebetween the benign and malignant groups.The proportions of cells reacting with anti-

bodies in each of the three grades of infiltratingductal carcinoma were also compared todetermine whether cytokeratin expressionchanged with differences in tumour grade.This statistical analysis was performed usingthe Kruskal-Wallis test.The reaction pattern of the various anti-

bodies with the myoepithelial cells surround-ing areas of in situ ductal carcinoma and thebenign breast diseases were compared usingthe Mann-Whitney U test. A p value of < 0-05was considered significant.

27


http://jcp.bmj.com

/J C


ownloaded from

http://jcp.bmj.com/

Heatley, Maxwell, Whiteside, Toner

Figure 2 In a section adjacent to that in fig 1 complementary staining of myoepithelialcells is seen with the antibody to cytokeratin 14. No reaction is evident in the luminalepithelium (frozen section x 240).

Figure 3 The antibody to cytokeratin 8 reacts with the luminal tumour cells in this focusof in situ ductal carcinoma (frozen section x 240).

Figure 4 In a section adjacent to that infig 3 the antibody to cytokeratin 14 highlightsthe myoepithelial cells surrounding the foci of in situ carcinoma (frozen section x 240).

ResultsCytokeratins were detected in the luminalepithelium of the 19 fibroadenoma cases, the19 cases of fibrocystic disease, the three phyl-

lodes tumour cases, and in the tumour cells inthe 68 cases infiltrating ductal, the seven cases

of lobular, and the three cases of mucinouscarcinoma (table 2 and figs 1-5).The luminal epithelial cells in most cases of

benign breast disease expressed cytokeratins7, 8, 18, and 19 and reacted with the anti-bodies which included these molecules intheir spectrum of reactivity (RCK105, BA17,M20, RGE53, Cam 5-2, RCK102, AEI,AE3, and LP34). In some cases this stainingwas heterogeneous (tables 3 and 4). A similarpattern of reaction was seen in the tumourcells of the infiltrating ductal carcinomas,although a subgroup of tumours was identi-fied in which one or more cytokeratins were

absent from all of the cancer cells (table 5).In cases of benign breast disease there was

a statistically significant difference in the pro-portion of tumour cells and luminal epithelialcells which reacted with the antibodies tocytokeratins 7 and 19 (p = 0 0044 and p =

0-0032, respectively; table 6). There were sta-tistically significant differences in the propor-tions of tumour cells and luminal epithelial

Table 3 Number of cases of the 19fibroadenoma casesreacting, in different proportions, to various anticytokeratinantibodies

Grade of reactivity

0 1 2 3 4 5

Antibody Number of cases per grade

RCK105 0 0 0 0 2 17BA17 1 0 0 2 8 8M20 0 0 0 3 5 10RGE53 0 0 1 6 9 1Cam52 0 0 1 1 1 16Anticytokeratin 14 17 1 0 0 0 0RCK102 0 1 0 0 7 11AEI 0 1 0 0 4 14AE3 1 0 0 1 6 11AE2 17 2 0 0 0 0LP34 1 0 1 0 6 11AE5 5 1 2 2 2 6AE8 18 1 0 0 0 0RSKE60 17 0 0 1 0 06B10 17 1 0 1 0 0

0 = No reactive cells; 1 = < 25% reactive cells; 2 = 26-50%reactive cells; 3 = 51-75% reactive cells; 4 = 76-99% reactivecells; 5 = 100% reactive cells.

Table 4 Number of cases of 19 cases offibrocystic diseasereacting, in different proportions, to various anticytokeratinantibodies

Grade of reactivity0 1 2 3 4 5


RCK105 0 0 1 1 5 12BA17 0 0 2 3 7 7M20 0 0 0 2 7 9RGE53 0 1 3 3 7 5Cam52 0 1 1 0 0 17Anticytokeratin 14 15 2 0 0 0 0RCK102 0 1 0 0 7 11AEI 0 0 0 0 1 18AE3 0 0 0 1 3 15AE2 18 1 0 0 0 0LP34 0 0 0 0 3 16AE5 3 4 2 0 3 7AE8 19 0 0 0 0 0RSKE60 18 0 0 0 1 06B10 19 0 0 0 0 0


28


http://jcp.bmj.com

/J C


ownloaded from

http://jcp.bmj.com/

Expression of cytokeratins in breast disease

cells in fibroadenomas reacting with bothantibodies (cytokeratin 7 p = 0-0022 andcytokeratin 19 p = 0-00312). The proportionsof tumour cells and luminal epithelial cells infibrocystic disease which reacted with theantibody to cytokeratin 19 (p = 0006) were

Figure S A positive reaction is present in the myoepithelial cells surrounding this focusof ductal carcinoma in situ with the specific antibody to cytokeratin 14 (frozen sectionx 240).

Table S Number of cases out of 64 infiltrating ductalcarcinoma cases reacting, in different proportions, tovarious anticytokeratin antibodies

Grade of reactivity

0 1 2 3 4 S


RCK105 7 4 1 8 12 35BA17 2 3 0 3 1 1 49M20 2 3 3 2 14 42RGE53 4 3 1 1 3 16 27Cam52 1 2 3 2 6 54Anticytokeratinl14 61 2 0 0 0 1RCK102 4 2 1 3 9 45AEI 1 0 3 5 8 51AE3 7 0 2 6 1 1 40AE2 59 2 0 0 0 0LP34 1 2 4 4 9 48AE5 18 8 5 4 14 1 1AE8 60 1 1 1 0 0RSKE60 61 0 1 0 0 06B10 55 5 0 0 0 3

0 = No reactive cells; 1 = <25% reactive cells; 2 =26-50%reactive cells; 3 = 51-75% reactive cells; 4 = 76-99% reactivecells; 5 = 100% reactive cells.

Table 6 Summary of the reaction of antibodies to intermediatefilament proteins in cellsfrom benign and malignant conditions of the breast

The position of cells reacting to antibody

Fibroadenoma Fibrocystic disease Adenocarcinoma

Antibody Positive Negative Positive Negative Positive Negative

RCK105 19* 0 19t 0 60 7BA17 18* 1 19* 0 66 2M20 18t 0 18t 0 64 2RGE53 17t 0 19t 0 60 4Cam 5-2 19t 0 19t 0 67 1Anticytokeratin 14 it 17 2t 15 3 61RCK102 19t 0 19t 0 60 4AEI 19t 0 19t 0 67 1AE3 18t 1 19t 0 59 7AE2 2t 17 it 18 2 59LP34 18t 1 19t 0 67 1AE5 13t 5 16t 3 42 18AE8 it 18 Ot 19 3 60RSKE60 it 17 it 18 1 616B10 2t 17 Ot 19 8 55* p < 0-05 Mann-Whitney U test compared with infiltrating ductal carcinoma.t p > 0 05 Mann-Whimney U test compared with infiltrating ductal carcinoma.

significantly different, whilst the proportionsreacting with the antibody to cytokeratinseven were not. No statistically significant dif-ferences were identified when these compar-isons were made for the other antibodiesstudied.

Comparison of the proportion of tumourcells in different grades of infiltrating ductalcarcinoma which reacted with the antibodiesshowed no evidence of a statistically signifi-cant difference (table 7).

Cytokeratin expression in tumour cells inlobular carcinomas (seven cases), mucinouscarcinomas (three cases), and in the luminalepithelium of phyllodes tumours (three cases)reflected that noted for the more commonbreast lesions (table 8).

Cytokeratin expression in myoepithelialcells in the 19 fibroadenomas and 19 cases offibrocystic disease was also studied (figs 1 and2; tables 9 and 10). Although no case of ductalcarcinoma in situ without invasion was exam-ined, foci of ductal carcinoma in situ wereidentified in the breast tissue adjacent to inva-sive tumours in 22 cases (figs 3, 4, and 5).These results are presented in table 11.The reaction pattern identified in the

myoepithelial cells was less predictable thanthat observed in the luminal epithelium andtumour cells. For example, the specific anti-body to cytokeratin 14, a protein which isusually widely distributed within the myoepi-thelium, did not detect these cells in 10 of thecases of benign breast disease and in situ car-cinoma examined. Furthermore, each of theother antibodies to myoepithelial cytokeratins,RCK102 (detects cytokeratin 5), AE1(detects cytokeratins 14 and 19), and AE3(detects cytokeratins 5 and 7), reacted withthe myoepithelium in less than half of thecases. These antibodies were less efficientthan a specific antiactin antibody, whichdetected myoepithelial cells in all cases exam-ined (figs 5 and 6).

Cytokeratin 14 was expressed in the myo-epithelium in a greater proportion of cases offibrocystic disease than in fibroadenoma cases(p = 0 008). There was no evidence of astatistically significant difference with any ofthe other antibodies examined (table 12).

DiscussionLuminal epithelial cells in the resting mam-mary gland express cytokeratins 7, 8, 16, 18,and 19, whilst the myoepithelial cells expresscytokeratins 5, 7, 14, and 17.'5 In addition,the expression of cytokeratins 18 and 19 bymyoepithelial cells has been described byWetzels et alP6 and Ferrero et al,17 respectively.As expected, the luminal epithelial cells in

most cases of benign breast disease reactedwith the specific antibodies to cytokeratin 7,8, 18, 19, and to those antibodies whichincluded these cytokeratins and cytokeratin16 in their spectrum of reactivity (thatis, RCK102, Cam 5-2, LP34, AE1, and AE3).

In a number of cases some epithelial cellsdid not express these cytokeratins, resulting ina heterogeneous reaction pattern in the luminal

29


http://jcp.bmj.com

/J C


ownloaded from

http://jcp.bmj.com/

Heatley, MaxweUl, Whiteside, Toner

Table 7 Breast carcinomas reacting to anticytokeratin antibodies by tumour gradeThe proportion of cells reacting to antibody

Not Kruskal-WaMlisAntibodies reacting Grade I Grade II Grade III test (p value)

RCK105 7 18 30 12 043 NSBA17 2 22 32 12 0-61 NSM20 2 20 31 13 0 54 NSRGE53 4 21 29 10 0-26 NSCam 5-2 1 22 32 13 0-93 NSAnticytokeratin 14 61 0 1 2 0 90 NSRCK102 4 18 29 13 0 34 NSAE1 1 23 31 13 0-56 NSAE3 7 21 27 11 0-20 NSAE2 59 0 2 0 0 30 NSLP34 1 23 31 13 0-76 NSAE5 18 12 18 12 007 NSAE8 60 0 2 1 0 50NSRSKE60 61 0 1 0 050 NS6B10 55 2 4 2 0-92 NS

Table 8 Number of cases of lobular carcinoma, mucinous carcinoma and benign andmalignant phyUodes tumours in which the epithelial cells reacted to a panel ofanticytokeratin antibodies

Tumour type

Phyllodes PhyllodesLobular Mucinous tumour tumour

Antibody carcinoma carcinoma (benign) (malignant)

RCK105 5/7 2/3 2/2 1/1BA17 6/7 3/3 2/2 1/1M20 7/7 2/2 2/2 1/1RGE53 7/7 2/2 2/2 1/1Cam 5-2 7/7 3/3 2/2 1/1Anticytokeratin 14 0/7 0/2 1/2 0/1RCK102 5/6 2/2 2/2 1/1AEI 7/7 3/3 2/2 1/1AE3 7/7 3/3 2/2 1/1AE2 0/6 0/2 1/2 0/1LP34 7/7 3/3 2/2 1/1AE5 3/6 1/2 2/2 1/1AE8 0/6 0/2 0/2 0/1RSKE60 0/6 0/2 0/2 0/16B10 0/6 1/2 0/2 0/1

Table 9 Number ofcases offibroadenoma where themyoepithelial cells reacted, in different proportions, tovarious anticytokeratin antibodies

Grade of reactivity

0 1 2 3 4 5


RCK105 15 0 1 0 0 3BA17 15 1 1 1 1 0M20 13 0 2 0 1 0RGE53 17 0 0 0 0 0Cam52 12 2 0 0 0 5Anticytokeratin 14 4 5 5 0 1 3RCK102 10 2 0 3 0 4AEI 13 0 1 0 0 5AE3 11 0 1 2 1 3AE2 19 0 0 0 0 0

LP34 9 2 2 0 0 6AE5 4 2 1 2 2 8AE8 15 0 0 1 1 0RSKE60 18 1 0 0 0 06B10 19 0 0 0 0 0


epithelium. Similarly, other authors havedescribed heterogeneous reactions with anti-bodies to cytokeratins 8,18 18,18 19 and 19.20 Anumber of carcinomas in the present studywere entirely negative for one or more of thesesimple cytokeratins. Again, similar findingshave been reported in the literature (cytoker-atin 7 negative tumours7 17182122; cytokeratin 8negative tumours'8; cytokeratin 18 negativetumours1618 23; and cytokeratin 19 negativetumours'8 2024)

Table 10 Number of cases offibrocystic disease where themyoepithelial cells reacted, in different proportions, tovarious anticytokeratin antibodies

Grade of reactivityO 1 2 3 4 5

Antibody Number of cases per gradeRCK105 17 0 1 0 0 0BA17 15 1 1 0 0 1M20 15 1 0 0 0 0RGE53 19 0 0 0 0 0Cam52 12 2 0 1 0 3Anticytokeratin 14 1 1 2 1 2 8RCK102 14 0 0 0 1 3AEI 12 0 1 2 1 2AE3 10 3 1 0 0 4AE2 19 0 0 0 0 0LP34 8 0 1 2 1 6AE5 5 1 1 0 5 6AE8 17 0 0 0 1 0RSKE60 17 1 0 0 0 06B10 19 0 0 0 0 0


Table 1 Number of cases ofductal carcinoma in situwhere the myoepithelial cells reacted, in differentproportions, to various anticytokeratin antibodies

Grade of reactivity

O 1 2 3 4 5

Antibody Number of cases per gradeRCK105 15 3 0 1 0 1BA17 17 2 0 1 0 2M20 22 0 0 0 0 0RGE53 22 0 0 0 0 0Cam52 18 0 0 0 1 2Anticytokeratin 14 5 1 2 2 2 6RCK102 14 0 0 0 3 1AEI 11 0 2 0 2 7AE3 15 0 0 0 1 4AE2 22 0 0 0 0 0LP34 6 0 1 1 2 12AE5 10 0 1 0 3 4AE8 15 0 0 0 3 0RSKE60 19 0 0 0 0 06B10 22 0 0 0 0 0


The complete absence of simple cytoker-atins in breast ductal and lobular carcinomasmay be because these tumours may havearisen in those luminal epithelial cells of theterminal duct lobular units which did notexpress these proteins.20 That this represents agenuine absence of the cytokeratin, ratherthan masking of an epitopic site or the effectof fixation, is supported by studies in whichseveral antibodies were used, reacting withmore than one epitopic site on a given cyto-keratin,722 and by the use of frozen sections,in which immunoreactivity is optimally pre-served, as in the current series.As cases in which all of the epithelial cells

were negative for cytokeratins 7, 8, and 18were limited to carcinomas, the use of theseantibodies may provide a convenient meansfor distinguishing benign from malignantbreast disease in some borderline cases. Thecancer cells in two adenocarcinomas werecompletely negative for cytokeratin 19 and astatistically significant difference was foundwhen cytokeratin 19 expression was com-pared in ductal carcinomas and benign breast

30


http://jcp.bmj.com

/J C


ownloaded from

http://jcp.bmj.com/

Expression ofcytokeratins in breast disease

I

Figure 6 The myoepithelial cells also reacted to an antibody to actin. This antibody alsoreacts with the blood vessels in the adjacent stroma (frozen section x 240).

disease. However, as the luminal epitheliumin one case of fibroadenoma was entirely non-reactive for this antibody, its value as a diag-nostic discriminator is limited. Findingssimilar to those described in cases of benignbreast disease were identified in the luminalepithelium in the two phyllodes tumours.

Three cytokeratin 14 positive tumours wereobserved in this series and in one case all ofthe tumour cells were positive. Gustersen eta125 and Wetzels et al'6 identified basementmembrane components in the tumour cells in aproportion of such tumours, suggesting thatthey arose from the myoepithelial layer of thebreast. However, cytokeratin 14 expression inthe absence of other markers of myoepithelialdifferentiation has also been described inbreast tumours.26 Occasional luminal cells inthe terminal duct lobular unit, which reactwith antibodies to cytokeratin 14, have beendescribed by Nagle et al 27 and these cells maybe the origin of some cytokeratin 14 positivetumours. Cytokeratin 14 expression mayalso represent a change in the cytokeratinprofile with the acquisition of malignant char-

Table 12 Summary of the reaction of antibodies to inter-mediatefilament proteins tomyoepithelial cells from benign and malignant conditions of the breast

The proportion of cells reacting to antibody

Ductal carcinomaFibroadenoma Fibrocystic disease in situ

Antibody Positive Negative Positive Negative Positive Negative

RCK105 4 15 1 17 5 15BA17 4 15 3 15 5 17M20 3 13 1 15 0 22RGE53 0 17 0 19 0 22Cam 5-2 7 12 6 12 3 18Anticytokeratin 14 14 4 14* 1 13 5RCK102 9 10 4 14 4 14AEI 6 13 6 12 11 11AE3 7 11 8 10 5 15AE2 0 19 0 19 0 22LP34 10 9 10 18 16 6AE5 15 4 13 5 8 10AE8 2 15 1 17 3 15RSKE60 1 18 1 17 0 196B10 0 19 0 19 0 22

* p < 0-05 Mann-Whitney U test compared with fibroadenoma.

acteristics. The acquisition of cytokeratin 14expression has been described in tumours ofthe urinary bladder.8 The finding that bothsimple cytokeratins and cytokeratin 14 wereco-expressed in three tumours in the presentstudy and that none of them reacted to a spe-cific antibody to actin would lend support toeither of these latter hypotheses.

Reactions with the antibodies AE2 (cyto-keratins 1, 2, 10, and 11), AE5 (cytokeratin3), AE8 (cytokeratin 13), RSKE60 (cytoker-atin 10), and 6B10 (cytokeratin 4) were iden-tified in the luminal epithelium in occasionalcases of benign breast disease and of invasivecancers. Of these, only cytokeratin 13 hasbeen previously reported within breast epithe-hium.'6 Statistically significant differences incytokeratin expression with increasing tumourgrade did not occur with any of the antibodiesstudied. Therefore, the value of the cyto-keratin profile in assessing tumour grade andprognosis in breast carcinoma appears to belimited.The detection in myoepithelium of cyto-

keratins 7, 18, 19, and 14 with specific anti-bodies to these proteins reflects the results ofprevious studies. 15-182326-30 The reaction withmyoepithelium of a number of antibodieswhich detect more than one cytokeratinreflects their ability to detect specific cytoker-atins which are known to be expressed bythese cells (Cam 5-2, cytokeratins 18 and 19;RCK102, cytokeratin 5; AE1, cytokeratin 14;AE3, cytokeratin 5; LP34, cytokeratin 18). Ofthese, cytokeratin 14 has been most exten-sively studied. Cytokeratin 14 expression hasbeen localised to the myoepithelial cells in thenormal breast, in cases of benign breast dis-ease16182326-30 and in surrounding foci of insitu carcinoma.'62729 In this study and that ofDairkee et a126 a reaction to cytokeratin 14 inthe myoepithelial cells was absent in somecases of benign disease. As in this series,Gottlieb et al30 found that a monoclonal anti-body to cytokeratin 14 was a less sensitivemeans of detecting myoepithelial cells thanthe use of antiactin antibodies.

This is the first study in which cytokeratins1, 2, 10, or 11 (AE2), cytokeratin 3 (AE5),cytokeratin 10 (RSKE60), and cytokeratin4 (6B 10) have been immunolocalised inmyoepithelial cells. In contrast to our find-ings Nagle et al27 were unable to detect cyto-keratin 13 in these cells using an alternativeantibody (2D7). Neither we nor Taylor-Papadimitriou24 or Guelstein et al3 were ableto detect cytokeratin 8 in myoepithelial cells.

In conclusion, cytokeratins 7, 8, and 18were invariably expressed in benign breast dis-ease and when absent, were reliable indicatorsof malignancy. Cytokeratin expressionshowed no statistically significant associationwith breast tumours of differing histologicalgrade, but was of some value in outlining themyoepithelial layers surrounding foci of in situcarcinoma, enabling these lesions to be differ-entiated from invasive carcinomas.

Supported by the Ulster Cancer Foundation and The RoyalGroup of Hospitals Trust. We are grateful for the photo-graphic expertise provided by Mr R Creighton and to Mrs E

31

r:4.

.:. :...


http://jcp.bmj.com

/J C


ownloaded from

http://jcp.bmj.com/

Heatley, Maxwell, Whiteside, Toner

McNamee who typed the paper. We also thank Dr CPatterson for statistical advice.

1 Fuchs E. Keratins as biochemical markers of epithelial dif-ferentiation. Trends Genet 1988;4:227-81.

2 Domagala W, Lubinski J, Weber K, Osborn M.Intermediate filament typing of tumor cells in fine needleaspirates by means of monoclonal antibodies. ActaCytologica 1986;30:214-24.

3 Lehto VP, Miettinen M, Virtanen I. Antibodies to inter-mediate filaments in surgical pathology. Arch GeschwulstForsch 1986;56:283-98.

4 Moll R, Schiller DL, Franke WW. Identification of proteinIT of the intestinal cytoskeleton as a novel type 1 cyto-keratin with unusual properties and expression patterns._J Cell Biol 1990;111:567-80.

5 Osborn M, Weber K. Biology of disease. Tumor diagnosisby intermediate filament typing: a novel tool for surgicalpathology. Lab Invest 1983;48:372-93.

6 Nagle RB. Intermediate Filaments: A review of the basicbiology. Am _7 Surg Pathol 1988;12(Suppl 1):4-16.

7 Ramaekers F, van Niekerk C, Poels L, Schaafsma E,Huijsmans A, Robben H, et al. Use of monoclonal anti-bodies to keratin 7 in the differential diagnosis of adeno-carcinomas. Am Pathol 1990;136:641-55.

8 Schaafsma HE, Ramaekers FCS, van Muijen GNP, LaneEB, Leigh IM, Robben H, et al. Distribution of cytoker-atin polypeptides in human transitional cell carcinomas,with special emphasis on changing expression patternduring tumor progression. Am _7 Pathol 1990;136:329-43.

9 Smedts F, Ramaekers F, Robben H, Pruszczynski M, van

Muijen G, Lane B, et al. Changing patterns of keratinexpression during progression of cervical intraepithelialneoplasia. Amj7Pathol 1990;136:657-68.

10 Broers JLV, Ramaekers FCS, Rot MK, Oostendorp T,Huysmans A, van Muijen GNP, et al. Cytokeratins indifferent types of human lung cancer as monitored bychain-specific monoclonal antibodies. Cancer Res 1988;48:3221-9.

11 Battifora H. Clinical applications of the immunohisto-chemistry of filamentous proteins. Am _7 Surg Pathol1988;12(Suppl 1):24-42.

12 Corson JM. Keratin protein immunohistochemistry insurgical pathology practice. PatholAnnu 1986;21:47-81.

13 Bloom HJG, Richardson WW. Histological grading andprognosis in breast cancer. A study of 1409 cases ofwhich 359 have been followed for fifteen years. BrJCancer 1957;11:359-77.

14 Elston CW, Gresham GA, Rao GS, Zebro T, Haybittle JL,Houghton J, et al. (On behalf of cancer research cam-

paign working party). The cancer research campaign(King's/Cambridge) trial for early breast cancer: Clinico-pathological aspects. BrJ Cancer 1982;45:655-69.

15 Mork C, van Deurs B, Petersen OW. Regulation ofvimentin expression in cultured human mammary

epithelial cells. Differentiation 1990;43:146-56.16 Wetzels RHW, Kuijpers HJH, Lane EB, Leigh IM,

Troyanovsky SM, Holland R, et al. Basal cell-specificand hyperproliferation-related keratins in human breastcancer. AmJ7 Pathol 1991;138:751-63.

17 Ferrero M, Spyratos F, Le Doussal V, Desplaces A,

Rouesse J. Flow cytometric analysis ofDNA content andkeratins by using CK7, CK8, CK18, CK19, and KL1monoclonal antibodies in benign and malignant humanbreast tumors. Cytometry 1990;11:716-24.

18 Tsubura A, Okada H, Senzaki H, Hatano T, Morii S.Keratin expression in the normal breast and in breastcarcinoma. Histopathology 1991;18:517-22.

19 Goddard MJ, Wilson B, Grant JW. Comparison of com-mercially available cytokeratin antibodies in normal andneoplastic adult epithelial and non-epithelial tissues._J Clin Pathol 199 1;44:660-3.

20 Bartek J, Taylor-Papadimitriou J, Miller N, Millis R.Patterns of expression of keratin 19 as detected withmonoclonal antibodies in human breast tissues andtumours. Int7Cancer 1985;36:299-306.

21 Bartek J, Vojtesek B, Staskova Z, Bartkova J, Kerekes Z,Rejthar A, et al. A series of 14 new monoclonal antibod-ies to keratins: Characterization and value in diagnostichistopathology. _7 Pathol 1991 ;164:215-24.

22 Vojtesek B, Staskova Z, Nenutil R, Bartkova J, Kovarik J,Rejthar A, et al. A panel of monoclonal antibodies to ker-atin No. 7: Characterization and value in tumor diagno-sis. Neoplasma 1990;37:333-42.

23 Dairkee SH, Puett L, Hackett AJ. Expression of basal andluminal epithelium-specific keratins in normal, benign,and malignant breast tissue. J Natl Cancer Inst1988;80:691-5.

24 Taylor-Papadimitriou J. Immunohistochemical analysis ofbenign and malignant breast tumours. Pathol Biol 1990;38:778.

25 Gusterson BA, Warburton MJ, Mitchell D, Ellison M,Neville AM, Rudland PS. Distribution of myoepithelialcells and basement membrane proteins in the normalbreast and in benign and malignant breast diseases.Cancer Res 1982;42:4763-70.

26 Dairkee SH, Ljung BM, Smith H, Hackett A.Immunolocalization of a human basal epitheliumspecific keratin in benign and malignant breast disease.Breast Cancer Res Treat 1987;10: 11-20.

27 Nagle RB, Bocker W, Davis JR, Heid HW, Kaufmann M,Lucas DO, et al. Characterization of breast carcinomasby two monoclonal antibodies distinguishing myoepithe-lial from luminal epithelial cells. _7 Histochem Cytochen1986;34:869-81.

28 Gould VE, Koukoulis GK, Jansson DS, Nagle RB, FrankeWW, Moll R. Coexpression patterns of vimentin andglial filament protein with cytokeratins in the normal,hyperplastic and neoplastic breast. Am _7 Pathol 1990;137:1143-55.

29 Wetzels RHW, Holland R, van Haelst UJGM, Lane EB,Leigh IM, Ramaekers FCS. Detection of basementmembrane components and basal cell keratin 14 in non-invasive and invasive carcinomas of the breast. Am _7Pathol 1989;134:571 -9.

30 Gottlieb C, Raju U, Greenawald KA. Myoepithelial cellsin the differential diagnosis of complex benign andmalignant breast lesions: an immunohistochemicalstudy. Mod Pathol 1990;3:135-40.

31 Guelstein VI, Tchrypysheva TA, Ermilova VD, LitvinovaLV, Troyanovsky SM, Bannikov GA. Monoclonal anti-body mapping of keratins 8 and 17 and of vimentin innormal human mammary gland, benign tumours, dys-plasias and breast cancer. Int_ Cancer 1988;42:147-53.

32


http://jcp.bmj.com

/J C


ownloaded from

http://jcp.bmj.com/

clin cytokeratin intermediate filament expression benign ...heatley, maxwell, whiteside, toner...

Documents