Journal of PathologyJ Pathol 2002; 198: 458–467.Published online 14 October 2002 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/path.1241

Original Paper

Usual ductal hyperplasia of the breast is a committedstem (progenitor) cell lesion distinct from atypical ductalhyperplasia and ductal carcinoma in situ

Werner Boecker,1* Roland Moll,2 Peter Dervan,3 Horst Buerger,1 Christopher Poremba,1 RaihanatouIna Diallo,1 Hermann Herbst,1 Ansgar Schmidt,2 Markus M Lerch,4 and Igor B Buchwalow4

1 Gerhard-Domagk-Institute of Pathology, University of Munster, Domagkstr. 17, D-48149 Munster, Germany2 Institute of Pathology, Philipps University of Marburg, Baldingerstrasse, D-35033 Marburg, Germany3 Pathology Department, University College Dublin, Mater Hospital, Eccles Sr. Dublin 7, Ireland4 Department of Medicine B and Central Ultrastructure Research Unit, interdisciplinary Centre of Clinical Research, University of Munster,Domagkstrasse. 17, D-48149 Munster, Germany

*Correspondence to:Werner Boecker, GerhardDomagk-Institute of Pathology,Domagkstrasse 17, D-48149,Munster, Germany.E-mail:boeckew@uni-muenster.de

Received: 30 January 2002Revised: 9 May 2002Accepted: 12 August 2002

AbstractCurrent classification systems in proliferative mammary gland pathology are based ona two-cell system, recognizing only glandular and myoepithelial lines of differentiation.A third cell type has recently been characterized in normal breast tissue by double-immunofluorescence analysis to express cytokeratin 5 (Ck5) only. These cells were shownto represent progenitor or adult stem cells that give rise to the glandular and myoepithelialcell lineage. The double-labelling technique has been applied to characterize a spectrumof intraductal epithelial proliferations, namely benign usual ductal hyperplasia, atypicalductal hyperplasia, and ductal carcinoma in situ, all of which are thought to representthe gradual steps of a sequence in the development of breast cancer. Immunofluorescencestudies with specific antibodies against Ck5, Ck8/18/19, and smooth muscle actin werecomplemented by western blotting analysis of Ck5 and Ck8/18/19 expression in normalbreast tissue and in proliferative lesions. Usual ductal hyperplasia appears to be a Ck5-positive committed stem (progenitor) cell lesion with the same differentiation potential asseen in the normal breast. This is in sharp contrast to atypical ductal hyperplasia/ductalcarcinoma in situ, which display the differentiated glandular immunophenotype (Ck8/18/19-positive, but Ck5-negative). These data require the abandonment of the idea of an obligatebiological continuum of intraductal proliferations from benign to malignant. This studyprovides evidence that cells undergoing malignant transformation tend to be fairly advancedin the glandular lineage of differentiation. The committed stem (progenitor) cell model maycontribute to a better understanding of both benign proliferative breast disease and breastcancer development. Copyright 2002 John Wiley & Sons, Ltd.

Keywords: breast; progenitor (stem) cells; ductal carcinoma in situ; stem cell model

Introduction

Current descriptions recognize two cell types withinthe human mammary gland epithelium: the inner glan-dular and the outer myoepithelial cell, which arecharacterized by the expression of glandular cytoke-ratins (Cks) 8/18/19 and smooth muscle actin (SMA),respectively. The finding of an additional cell type,characterized by Ck5 expression in the absence ofCk8/18/19 and SMA, as well as of cells with anintermediate phenotype differentiating along the glan-dular lineage, led to a progenitor cell function beingattributed to this population within the mammaryepithelial compartment [1,2]. In a previous study wehave shown that these Ck5-positive cells have pro-genitor cell function both for the myoepithelial andglandular cell lineages [3].

In the currently favoured working hypothesis ofhuman breast cancer evolution, usual ductal hyper-plasia (UDH), atypical ductal hyperplasia (ADH), andductal carcinoma in situ (DCIS) are considered to besuccessive steps in a linear progression model termi-nating in invasive breast cancer (IBC) [4–9]. On theother hand, an increasing set of data is not consis-tent with this view [10–13]. We therefore investigatedthe whole range of intraductal proliferations. We useddouble-labelling immunofluorescence techniques withdigital image processing for the semi-quantitative anal-ysis of co-localized antigens by applying antibodiesspecific for different sets of glandular and basal cytok-eratins [14], especially against Ck8/18/19 and Ck5/6,as well as for myoepithelial differentiation markers(SMA). Western blotting experiments complementedthese studies.

Copyright 2002 John Wiley & Sons, Ltd.

Usual ductal hyperplasia of the breast 459

We describe qualitative differences in the expres-sion patterns between UDH, on the one hand, andADH / DCIS, on the other, that cannot be easily recon-ciled with the view of a gradual evolution of invasivebreast cancer from UDH to ADH/DCIS and, further,to definitive invasive cancers. Thus, UDH on the onehand and ADH/DCIS on the other represent differ-ent types of intraductal proliferations that appear to becaused by alterations of proliferative and differentia-tion capacities at different cellular levels.

Materials and methods

Tissues

Breast tissue biopsy samples were obtained from 56patients. The lesions were diagnosed and classifiedon routine haematoxylin and eosin (H&E)-stainedsections by three pathologists, according to currentclassification schemes [15–19] before complementaryCk5/6 immunohistochemistry was undertaken. Tencases of normal mammary gland tissue, 17 cases ofusual ductal hyperplasia, 16 cases of ductal carcinomain situ, and six cases of atypical ductal hyperplasiawere investigated. In accordance with recent nomen-clature, ADH refers to a small lesion with evenlyspaced cells and uniform round to oval nuclei [19].

Antibodies and immunohistochemical staining

For immunocytochemical analysis, routine paraffinsections of formaldehyde-fixed tissues were mountedon SuperFrost microslides (Menzel Glaser, Braun-schweig, Germany) coated with poly-L-lysine (Sigma).For antigen retrieval, dewaxed and rehydrated sectionswere autoclaved in citrate buffer, pH 6.0, at 120 ◦C for10 min. Sections were incubated for 45 min at roomtemperature with primary monoclonal mouse antibod-ies (MAbs), diluted in 10% RPMI (Sigma) supple-mented with 10% bovine serum albumin (Sigma) and0.1% NaN3, pH 7.5. Myoepithelial differentiation wasdemonstrated using a MAb specific for SMA (clone1A4; Sigma, Taufkirchen, Germany; diluted 1 : 200).Glandular differentiation was assessed with a MAbagainst Ck8/18 (used as a cocktail of Ck8 and Ck18;clones C 51 and DC 10, dilution 1 : 10 and 1 : 20,respectively; Quartett, Berlin, Germany) or Ck8/18/19(clone KL1; Immunotech, Marseille, France; dilution1 : 50), and progenitor cells were detected with a MAbagainst Ck5/6 (clone D5/16B4; Zymed, San Francisco,USA; 1 : 50, and Dako, Hamburg, Germany; 1 : 80).For simultaneous visualization of pairs of antibod-ies on the same tissue section, we performed subse-quent immunolabelling. The sections were incubatedwith a first primary MAb, washed, and incubatedwith FITC-conjugated rabbit anti-mouse antibodies(Dianova, Hamburg, Germany; diluted 1 : 50). Sectionswere then incubated with the second primary MAb thathad been biotinylated previously (ARK biotinylatingkit; Dako, Hamburg, Germany). The biotin label was

subsequently visualized with Cy3-conjugated strep-tavidin (Dianova, 1 : 200). SMA was detected witha directly FITC-conjugated antibody. Finally, nucleiwere counterstained with DAPI (Sigma, 5 µg/ml Trisbuffer) for 15 s and specimens were mounted withVectashield (Vector Laboratories, Burlingame, USA).Controls were incubated without a primary MAb orwith mouse IgG applied at the same concentrationas the primary MAb. In the controls, no specificimmunolabelling was observed. All sections were alsostained using conventional enzyme immunolabellingas described elsewhere [20–23].

Microscopy and image processing

Immunostained slides were examined on a ZeissAxioplan 2 bright field/fluorescence microscopeequipped with appropriate filters. Separate imagesfor DAPI, Cy3, and FITC staining were captureddigitally from triple-stained specimens into colour-separated components using an AxioCam digitalcamera (Carl Zeiss, Werke Gottingen, Germany) andAxioVision2.05 multi-channel image processing (CarlZeiss Vision GmbH, Hallbergmoos, Germany). Thered (for Cy3), blue (for DAPI), and green (for FITC)components were merged and composite imageswere imported as JPEG files into PhotoImpact 3.0(Ulead Systems, Inc, Torrance, CA, USA) for furtheranalysis. The images shown are representative of twoindependent experiments, which gave similar results.

Western blotting

For the western blotting experiments, biopsy speci-mens containing ductal hyperplasia (usual type), DCIS(non-high grade), and normal mammary gland tis-sue were snap-frozen in liquid nitrogen. Cytokeratinswere extracted routinely and separated by sodiumdodecyl sulphate polyacrylamide gel electrophoresis(SDS-PAGE) as described elsewhere [24] and trans-ferred onto a polyvinylidene difluoride (PVDF) mem-brane (Immobilon P; Millipore, Eschlhofen, Germany)using a semi-dry blotting system (Trans-Blot SD; Bio-Rad, Munich, Germany). Membranes were blockedwith 10% skimmed milk in phosphate-buffered saline(PBS) containing 0.1% Tween (PBS-Tween) and thenincubated with the Ck5/6 MAb (purchased fromBoehringer Mannheim, Mannheim, Germany; nowavailable from Chemicon) diluted 1 : 3000 in PBS-Tween for 1 h, followed by exposure to horseradishperoxidase (HRP)-conjugated antibodies to mouse IgG(DAKO) for 20 min (diluted 1 : 500 in PBS-Tween).Specific binding of the Ck5/6 MAb was determinedusing enhanced chemiluminescence (ECL; Amersham-Pharmacia Biotech, Freiburg, Germany).

The same PVDF membrane was then subjectedto a second immunoreaction using monoclonal MAbM20 against Ck8 (Euro-Diagnostica, Arnhem, TheNetherlands). The resulting second ECL revealedboth polypeptides to be immunoreactive for Ck5

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

460 W Boecker et al

and Ck8. Finally, the PVDF membrane was stainedwith Coomassie Brilliant Blue, revealing completepolypeptide transfer throughout all gel lanes.

Results

Immunohistochemistry anddouble-immunofluorescence staining experiments

All tissue samples were cut in serial sections andimmunostained using the indirect peroxidase techniqueas previously described [20–23] with antibodies spe-cific for Ck5/6, Ck8/18, Ck8/18/19, and SMA. Similarstaining patterns were obtained with different MAbclones specific for the respective glandular cytoker-atins.

Normal breast

The Ck5/6-specific antibody consistently immunore-acted with many basal (myoepithelial) cells and asubset of luminal cells of the ductal system as shownin Figures 1 and 2. The luminal and the basal epithelialcompartments of the lobules, however, showed volatilestaining that varied in intensity from moderate to min-imal or even absent, with a great variety of stagesbetween these extremes.

Double-staining with Ck5/6 and Ck8/18/19 MAbsrevealed heterogeneous expression patterns of thesetwo cytokeratin subgroups in the inner cell layer.The luminal cells of ducts contained only a fewcells that expressed Ck5 in the absence of Ck8/18/19(Figures 1c–1g). In addition, all samples containedvarying numbers of cells staining with Ck8/18/19 inthe absence of Ck5. Most luminal cells reacted withboth the Ck5/6 and the Ck8/18/19 MAbs in variableintensity, indicating intermediate cells in transitionfrom Ck5-positive cells to Ck8/18/19-positive glandu-lar end cells. The lobules again showed considerablevolatility in staining patterns. In most cases, Ck5,Ck8/18/19, as well as Ck5 and Ck8/18/19-positiveluminal cells could be found. In addition, there werelobules that contained exclusively Ck8/18/19-positiveluminal end cells.

Double-staining with MAbs specific for Ck5/6 andSMA revealed that the normal mammary gland har-bours single cells within the outer layer that expressCk5 in the absence of SMA (Figures 2b–2d), whereasmany cells in that location co-expressed Ck5 andSMA, and terminally differentiated myoepithelial cellsexpressed SMA alone. The presence of two distinctcell layers was corroborated by staining with Absagainst SMA and Ck8/18/19 (Figure 2e).

Usual ductal hyperplasia

Seventeen cases of UDH displayed immunohisto-chemical staining with the Ck5/6-specific MAb inthe majority of the proliferating intraductal cells(Figure 3b). Double-labelling displayed Ck5-positive

cells, both Ck5- and Ck8/18/19-positive cells, as wellas cells exclusively expressing Ck8/18/19 within theintraductal cell population (Figure 3c), whereas themyoepithelial marker SMA was never detected inthose cells (Figures 3d and 3e). The outer epitheliallayer contained a mixture of cells co-expressing Ck5and SMA and myoepithelial end cells expressing SMAalone (Figure 3d).

Atypical ductal hyperplasia and ductal in situcarcinomas

Sixteen cases of DCIS of varying grades (five lownuclear grade, five intermediate nuclear grade, andsix of seven high nuclear grade) displayed an intensereaction with the Ck8/18/19-specific MAb among allneoplastic cells (Figures 4c and 4e). The same stain-ing pattern was observed in all the ADH cases, withCk8/18/19 expression in the absence of reactivity withthe Ck5-specific MAb in the neoplastic cell popu-lation. In contrast to DCIS, however, a few cells,morphologically consistent with residual non-atypicalcells, expressed Ck5. Double-labelling for Ck5 andSMA confirmed the absence of Ck5 among the atyp-ical cells (Figure 4d). In contrast to UDH, transitionsfrom Ck5-positive cells to Ck8/18/19-positive cellswere not observed (Figures 3, 4, and 5c). Double-staining for Ck5 and SMA showed a mixture ofintermediate myoepithelial cells (Ck5-positive, SMA-positive) and myoepithelial end cells (SMA-positive),with occasional Ck5-positive cells in the outer celllayer, thus indicating a normal basal layer (Figure 4d).In one of the 17 in situ malignancies, a ductal carci-noma in situ of high nuclear grade, the tumour cellsexpressed Ck5 and Ck8/18/19 (not shown).

Western blotting

Western blotting analysis of normal breast tissue,UDH, and DCIS with the Ck5/6-specific MAb dis-played a 58 kD band in normal breast tissue and UDH,but not in DCIS (Figure 5). At variance, substitutingthe Ck5/6 antibody for a Ck8-specific reagent resultedin the detection of a 52.5 kD band of similar intensityin the same three lesions. Thus, basal cytokeratins Ck5and Ck8 are present in both normal mammary tissueand UDH, whereas DCIS is characterized by Ck8 inthe absence of Ck5. Western blotting of control cellswith a previously established cytokeratin composition,such as A-431 cells with Ck5/Ck8 co-expression andA-549 cells with Ck8 positivity, revealed bands of theappropriate size. The exclusive reactivity of the Ck5/6-specific MAb with a major band of 52.5 kD indi-cates that, in breast tissue, the MAb (clone D5/16B4)detected Ck5, but not Ck6.

Discussion

Adult human breast tissue contains distinct epithe-lial cell populations as evidenced from double-immunolabelling and western blotting. The terminal

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

Usual ductal hyperplasia of the breast 461

cell types of the glandular and the myoepithelial linesof differentiation originate from a Ck5-positive cellpopulation that thus has to be regarded as progenitoror committed stem cells of the epithelial cells of thehuman breast. During the lineage-specific differentia-tion process, Ck5 is down-regulated, ultimately lost,and replaced by Ck8/18/19 or SMA [3].

According to current concepts, intraductal prolifer-ations such as UDH, ADH, and DCIS are consideredto be precursor lesions in a sequence ultimately lead-ing to invasive breast cancer [7]. Yet a series of recentdata failed to support this idea [4,3,25], and reachinga better understanding of how the breast gland givesrise to those different types of proliferation is now oneof the central challenges in the breast cancer devel-opment debate. Our double-immunolabelling studiesrevealed that UDH, on the one hand, and ADH/DCIS,on the other, seem to represent phenotypically distinctlesions. Considered as progenitor or committed stemcells, Ck5-positive cells in UDH proliferate and dif-ferentiate into glandular cells via intermediate stages.The presence of numerous Ck5-positive cells was con-firmed biochemically by large amounts of Ck5 inwestern blotting of the UDH lesion. Thus, the basicbiological mechanisms leading to proliferation in UDHseem to occur at the glandularly committed stem (pro-genitor) cell level.

In contrast, all of the ADH and more than 95%of the DCIS cases studied were characterized by theexpression of Ck8/18/19, which is indicative of apurely glandular phenotype. For DCIS, Ck5 expressionwas absent or only minimal in western blotting. Thesedata suggest that ADH and most DCIS are derivedfrom functional glandular cells, or from cells at alate stage of the glandular cell lineage (Figure 5 andFigure 6). Support for this hypothesis comes from thelack of Ck5 expression in neoplastic cells.

Since the classic studies of Page’s group, UDH isregarded as a benign, non-atypical proliferation whichis associated with a slight general increase in cancerrisk [4–6,8,9,26]. The question of whether UDH is adirect precursor of DCIS remains contentious. Whileit is true that in situ malignancies, usually of ductaltype, may occasionally occur within the confines ofUDH, such a coincidence is, in our opinion, tooinfrequent to account for more than a random link.Moreover, in contrast to DCIS, a direct progressionfrom UDH to invasive breast cancer has so far not beenconvincingly demonstrated at the morphological ormolecular level. Furthermore, recent research equallysupports the view that UDH is distinct from DCIS.Firstly, results obtained by means of comparativegenomic hybridization reveal that all non-invasiveand invasive breast cancers are associated with grosscytogenetic alterations, while UDH does not displayany such changes [10,27]. Secondly, microsatelliteanalyses have revealed that about a third of all UDHcases exhibit allelic imbalances [11,12]. However, incontrast to DCIS, these are randomly distributed overthe genome and are inconsistent (non-recurrent) atany one locus. Although the LOH data indicate aclonal proliferation of at least a part of UDH andclassify these as benign neoplasms (adenomas), thisdoes not necessarily mean that these lesions followa cancerous pathway [9]. Thirdly, Shoker et al. [28]showed that, in contrast to UDH, in ADH/DCIS thenegative association between oestrogen receptor andKi-67 is lost.

These data, in conjunction with our immunofluores-cence studies of UDH and ADH/DCIS cases, suggestthat despite the intraductal proliferation which bothlesions have in common, UDH is unlikely to be anobligate and direct precursor of DCIS. Thus, UDHshould not be conceived as an early obligate precursor

Figure 1 (overleaf). Normal mammary gland. H&E-stained section (a), horseradish-peroxidase (HRP) immunohistochemicallabelling of Ck5 (b), and double-immunolabelling of paraffin-embedded tissues, stained with MAb to cytokeratin subgroups Ck5and Ck8/18/19 (c–g). (a) H&E-stained section with a small duct (D) and a lobule (L). The epithelium consists of an inner glandularand an outer basal (myoepithelial) cell layer. (b) HRP-immunohistochemical labelling of Ck5 reveals a heterogeneous stainingpattern. Cells of the inner and the outer layer of the intralobular terminal duct (TD) immunoreact for Ck5. The immunolabellingof adjacent acini varies widely. In one acinus, luminal and basal cells are stained (arrow); other acini display only a moderate toweak Ck5 reaction in basal cells, whereas luminal cells are Ck5-negative. (c,d) Double staining of ductal epithelium with Ck5/6MAb (green) and Ck8/18/19 MAb (red). (c) Composite image with only one luminal cell displaying an intense green signal (arrow)indicating a Ck5+ progenitor cell; several luminal cells have only red signal, indicative of Ck8/18/19+ glandular cells. In addition,several cells show hybrid (weak yellow) signal, indicating a Ck5+, Ck8/18/19+ intermediate status (asterisks). (d) Same area as inc with a 65-pixel shift of the red signal (Ck8/18/19+) upwards, revealing clearly the two-colour status of green (Ck5+) and redsignal (Ck8/18/19+) in each cell. Note that glandular cells strongly express only Ck8/18/19 (red), whereas intermediate glandularcells (asterisks), in addition to Ck5 (red), also express Ck8/18/19 (red), but in trace amounts. (e) Terminal duct immunostainedfor Ck5 and Ck8/18/19. This composite image includes cells expressing only Ck5 (green signal). Many Ck5+ and Ck8/18/19+intermediate cells with hybrid yellow colours (asterisks) and a few glandular cells with expression of Ck8/18/19 alone (red signal)reveal a gradual transition from Ck5+ progenitor cells to Ck8/18/19+ glandular cells via intermediate cells (Ck5+, Ck8/18/19+).(f) Acinar structures of a lobule double-immunostained for Ck5 (green signal) and Ck8/18/19 (red signal) with a 56-pixel shift ofthe red signal (Ck8/18/19) to the right. Note that there are only a few cells expressing Ck5 alone, indicating progenitor cells(arrow) and many intermediate cells co-expressing both keratin subgroups (asterisks), whereas some glandular cells exclusivelyexpress Ck8/18/19 (red signal). # indicates the cell exhibiting exclusively the red signal shifted 56 pixels to the right. (g) Part of alarger duct double-immunostained for Ck5 (green signal) and Ck8/18/19 (red signal). Note the intense expression of Ck5 in basal(myoepithelial) cells. The luminal epithelium with a mixed staining pattern contains a few Ck5+ progenitor cells (green signal),Ck5+, Ck8/18/19+ intermediate cells (asterisks), and Ck8/18/19+ glandular cells (intense red signal)

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

462 W Boecker et al

in a progression cascade to breast cancer [29]. It may,however, be tempting to hypothesize that some of therare Ck5-positive, Ck8/18/19-positive DCIS cases maybe derived from UDH lesions.

On the basis of the pioneering work of Pageet al. [30], it was possible to define another intraduc-tal proliferation, ADH, as an indicator lesion of amoderately increased general cancer risk. Yet contro-versy concerning these lesions persists in both practi-cal and theoretical terms. In this study, we followed

the explicit positive criteria and included six cases withonly tiny focal proliferations of evenly spaced cellswith monotonous round to oval nuclei that were con-fined to lobules [19]. Associated DCIS was excluded.In all cases, we found the same Ck8/18/19-positiveand Ck5-negative phenotype. Based on these find-ings, we conclude that ADH represents the lowest endof the low-nuclear-grade DCIS spectrum. This viewis further supported by a host of data from otherstudies (e.g. ref 28). Furthermore, recurring allelic

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

Usual ductal hyperplasia of the breast 463

Figure 2. Normal mammary gland. HRP-immunohistochemical labelling of SMA (a) and double-immunolabelling ofparaffin-embedded tissues, stained with MAb to Ck5, SMA, and Ck8/18/19 (b–e). (a) HRP-immunohistochemical labelling ofSMA of a lobule reveals characteristic positive staining of the basal (myoepithelial) layer. (b–d) Double-staining of TDLU withCk5/6 MAb (red) and SMA MAb (green). (b) Composite image with a mixture of staining patterns in the basal (myoepithelial) layerof a terminal duct (TD) and adjacent acini of a lobule. Most basal cells display a hybrid (yellow) colour (Ck5+, SMA+), indicativeof intermediate myoepithelial cells, and a green colour (SMA+), indicative of differentiated myoepithelial end cells. There aresome basal cells with red fluorescence (Ck5+), indicating progenitor cell function. (c, d) Lobular acinus double-immunolabelledfor Ck5 and SMA with a 6-pixel shift of the red signal (Ck8/18/19) to the right in d, revealing clearly a green and a red signalin intermediate myoepithelial cells (asterisk). There is only one oval-shaped progenitor cell, expressing Ck5 (red signal) alone(arrow). (e) Double-labelling for SMA (green signal) and Ck8/18/19 (red signal). Note the clear distinction in staining patternbetween the outer and the inner cell layer

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

464 W Boecker et al

Figure 3. Usual ductal hyperplasia. H&E-stained section (a), HRP-immunohistochemical labelling of Ck5 (b), and doubleimmunolabelling of paraffin-embedded tissues, stained with Abs to keratin subgroups Ck5, Ck8/18/19, and to SMA (c–e).(a) Characteristic H&E view of UDH with irregular spacing of nuclei and uneven secondary lumina. (b) HRP-immunohistochemicallabelling of Ck5 showing many cells of this intraductal proliferation expressing Ck5, whereas some cells are clearly negative.(c) Double-labelling for Ck5 (green signal) and Ck8/18/19 (red signal). Note that some cells display a clear green (Ck5+) orred signal (Ck8/18/19+) with, however, many cells showing a hybrid (yellow) signal (Ck5+, Ck8/18/19+) (asterisks), indicatingintermediate cells in transition from Ck5+ progenitor cells to Ck8/18/19 differentiated glandular cells. (d) Double-labelling for Ck5(red signal) and SMA (green signal). The basal (myoepithelial) cell layer displays a mixture of cells with hybrid (yellow) and greencolour. The cells of the intraductal proliferation display a mixture of intensely red (Ck5+) and unstained cells. The unstained cellsprobably represent functional glandular cells (Ck8/18/19+). (e) Double-labelling for Ck8/18/19 (red signal) and SMA (green signal).Two completely distinct cell types are seen — (basal) myoepithelial cells (green signal, indicating SMA+) and intraductal glandularcells (red signal, indicating Ck8/18/19). Note that the myoepithelial cells are not a component of the intraductal cell proliferation

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

Usual ductal hyperplasia of the breast 465

Figure 4. Ductal carcinoma in situ. H&E-stained section (a), HRP-immunohistochemical labelling of Ck5 (b), anddouble-immunolabelling of paraffin-embedded tissues, stained with Abs to keratin subgroups Ck5, Ck8/18/19, and to SMA(c–e). (a) Typical H&E view of a high-grade DCIS with large nuclei, several mitoses, and a small area of comedo type necrosis.(b) HRP-immunohistochemical labelling of Ck5 showing positively stained basal (myoepithelial) cells. The neoplastic cells of theintraductal proliferation do not immunoreact with Ck5/6 Ab. (c) Double-labelling for Ck8/18/19 (red signal) and Ck5 (green signal).There is a prominent basal layer, stained for Ck5 (green signal). The intraductal neoplastic proliferation displays only a red signal,indicating a purely glandular phenotype. (d) Double-labelling for Ck5 (red signal) and SMA (green signal). Note that only basal(myoepithelial) cells are stained, whereas the luminal cells do not immunoreact at all. There is only one Ck5+ cell (red signal), anoccasional cell with hybrid (yellow) colour (Ck5+, SMA+), and many cells with a green signal (SMA+). (e) Double-labelling forCk8/18/19 (red signal) and SMA (green signal) with the characteristic staining pattern of basal (myoepithelial) cells (green signal,SMA+) and neoplastic cells (red signal, Ck8/18/19+)

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

466 W Boecker et al

1 – A 431

2 – A 549

3 – Normal

4 – UDH

5 – DCIS

1 2 3 4 5

K5

K8

58.0 kDa

52.5 kDa

Figure 5. Western blotting of breast tissue preparations andcultured cells. Lanes from left to right: A-431 cells (total cellularproteins); A-549 cells (total cellular proteins); normal breasttissue (micro-dissected terminal ducts and lobular units); usualductal hyperplasia (micro-dissected lesions); ductal carcinomain situ (micro-dissected lesions). The panel shows the westernblotting reaction produced by monoclonal MAb D5/16B4against Ck5. Clear-cut Ck5 bands at 58.0 kDl are presentin A-431 epidermoid carcinoma cells, in normal breast tissue,and — particularly prominent — in usual ductal hyperplasia. InDCIS, only a very faint Ck5 reaction is seen, probably due tothe presence of normal basal cells. The subsequent westernblotting immunoreaction with anti-Ck8 after incubation withMAb M20 against Ck8 shows the additional appearance of Ck8bands (52.5 kD) in all preparations, being particularly strongin DCIS

imbalance has been found with similar frequency inADH and DCIS [11,12], indicating a causal relation-ship to breast cancer.

In view of the cellular and molecular data, there issubstantial evidence to suggest that ADH representsa pre-cancerous cell population which is not only

an indicator of a general cancer risk — similar toALH/CLIS — but also a direct precursor of IBC. Itmay therefore be more appropriate to use a term suchas early ductal in situ (intraepithelial) neoplasia ratherthan ADH with its clearly benign connotations.

On the basis of double-labelling and biochemi-cal studies, we conclude that fully developed UDHdoes not represent an immediate obligate precursorof ADH/DCIS. This view is supported by cytomor-phological differences between the two lesions and bycytogenetic data that revealed recurrent genetic alter-ations in all epithelial malignancies, including ADHand DCIS [27,31–33], whereas cytogenetic abnormal-ities in benign proliferative breast disease are ran-domly distributed [5,10–12,34,35]. These data, how-ever, require us to abandon the idea of an obligatebiological continuum of intraductal proliferations frombenign to malignant [25,36].

In summary, our data allowed us the following con-clusions. Based on morphology, immunophenotype,molecular pathology, and clinical outcome, UDH andADH/DCIS seem to constitute distinct clinical andpathological entities. The committed stem (progenitor)cell concept might serve as a useful model of breastregeneration and provide us with a new operativeparadigm for a better understanding of proliferativediseases of the breast.

AcknowledgementsThis study is dedicated to Professor Ekkehard Grundmannand to Professor Georg Dhom on the occasion of their 80thbirthdays. The expert technical assistance of Mrs BettinaBroemmelkamp is gratefully acknowledged. We also thank MsMonika Dirkwinkel for preparing the schematic drawings andMs Susanne Koelsch for editing the manuscript.

Breast cancerCk5-; Ck8/18/19+

Myoepithelial lineage Progenitor cell Glandular lineage

5-10% >90%

Usual ductal hyperplasia

Breast cancerCk5+; Ck8/18/19+

Figure 6. Hypothetical differentiation pathways for human mammary gland epithelial cells based on patterns of keratin subgroupand SMA staining and on data from the literature. Progenitor (adult stem) cells (Ck5/14+ only) give rise to either intermediateglandular (Ck5/14+; Ck8/18/19+) or intermediate myoepithelial cells (Ck5/14+; SMA+). The intermediate glandular cellsdifferentiate further into transit glandular cells (Ck8/18/19+ only) before final differentiation to lactating end cells (Ck8/18/19+only) occurs. The intermediate myoepithelial cells differentiate to myoepithelial end cells (SMA+ only). Most breast cancers appearto arise from differentiated glandular cells. Usual ductal hyperplasia seems to involve the glandularly committed progenitor (stem)cell with differentiation to glandular cells

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.

Usual ductal hyperplasia of the breast 467

References

1. Purkis PE, Steel JB, Mackenzie IC, Nathrath WB, Leigh IM,Lane EB. Antibody markers of basal cells in complex epithelia.J Cell Sci 1990; 97(Pt 1): 39–50.

2. Taylor-Papadimitriou J, Wetzels RHW, Ramaekers FC. Intermedi-ate filament protein expression in normal and malignant humanmammary epithelial cells. In Genes, Oncogenes and Hormones:Advances in Cellular and Molecular Biology of Breast Cancer,Dickson RB, Lippmann ME (eds). Kluwer Academic Publishers:Boston, 1991; 355–378.

3. Boecker W, Moll R, Poremba C, et al. Common adult stem cellsin the human breast give rise to glandular and myoepithelial celllineages: a new cell biological concept. Lab Invest 2002; 82(6):737–746.

4. Bodian CA, Perzin KH, Lattes R, Hoffmann P. Reproducibilityand validity of pathologic classifications of benign breast diseaseand implications for clinical applications. Cancer 1993; 71:3908–3913.

5. Carter CL, Corle DK, Micozzi MS, Schatzkin A, Taylor PR. Aprospective study of the development of breast cancer in 16 692women with benign breast disease. Am J Epidemiol 1988; 128:467–477.

6. Dupont WD, Page DL. Risk factors for breast cancer in womenwith proliferative breast disease. N Engl J Med 1985; 312:146–151.

7. Lakhani SR, Chaggar R, Davies S, et al. Genetic alterations in‘normal’ luminal and myoepithelial cells of the breast. J Pathol1999; 189: 496–503.

8. London SJ, Connolly JL, Schnitt SJ, Colditz GA. A prospectivestudy of benign breast disease and the risk of breast cancer[published erratum appears in J Am Med Assoc 1992; 267: 1780].J Am Med Assoc 1992; 267: 941–944.

9. Mommers EC, van Diest PJ, Leonhart AM, Meijer CJ, Baak JP.Expression of proliferation and apoptosis-related proteins in usualductal hyperplasia of the breast. Hum Pathol 1998; 29: 1539–1545.

10. Boecker W, Buerger H, Schmitz K, et al. Ductal epithelialproliferations of the breast: a biological continuum? Comparativegenomic hybridisation and high-molecular-weight cytokeratinexpression patterns. J Pathol 2001; 195: 415–421.

11. Lakhani SR, Slack DN, Hamoudi RA, Collins N, Stratton MR.Detection of allelic imbalance indicates that a proportion ofmammary hyperplasia of usual type are clonal, neoplasticproliferations. Lab Invest 1996; 74: 129–135.

12. O’Connell P, Pekkel V, Fuqua SA, Osborne CK, Clark GM,Allred DC. Analysis of loss of heterozygosity in 399 premalignantbreast lesions at 15 genetic loci. J Natl Cancer Inst 1998; 90:697–703.

13. Shoker BS, Jarvis C, Clarke RB, et al. Abnormal regulation of theoestrogen receptor in benign breast lesions. J Clin Pathol 2000; 53:778–783.

14. Moll R, Franke WW, Schiller DL, Geiger B, Krepler R. Thecatalog of human cytokeratins: patterns of expression in normalepithelia, tumors and cultured cells. Cell 1982; 31: 11–24.

15. Elston CW, Ellis IO. The Breast (1st edn). Harcourt Brace:Edinburgh, 1998.

16. Holland R, Peterse JL, Millis RR, et al. Ductal carcinoma in situ:a proposal for a new classification. Semin Diagn Pathol 1994; 11:167–180.

17. Page DL, Anderson TJ. Diagnostic Histopathology of the Breast(1st edn). Churchill Livingstone: Edinburgh, 1987.

18. Silverstein MJ, Poller DN, Waisman JR, et al. Prognostic clas-sification of breast ductal carcinoma-in-situ. Lancet 1995; 345:1154–1157.

19. Sloane JP. Biopsy Pathology of the Breast (2nd edn). Arnold:London, 2001.

20. Boecker W, Bier B, Freytag G, et al. An immunohistochemicalstudy of the breast using antibodies to basal and luminal keratins,alpha-smooth muscle actin, vimentin, collagen IV and laminin.Part II: Epitheliosis and ductal carcinoma in situ. Virchows Arch A[Pathol Anat Histopathol] 1992; 421: 323–330.

21. Moinfar F, Man YG, Lininger RA, Bodian C, Tavassoli FA. Useof keratin 35βE12 as an adjunct in the diagnosis of mammaryintraepithelial neoplasia, ductal type — benign and malignantintraductal proliferations. Am J Surg Pathol 1999; 23: 1048–1058.

22. Otterbach F, Bankfalvi A, Bergner S, Decker T, Krech R,Boecker W. Cytokeratin 5/6 immunohistochemistry assists thedifferential diagnosis of atypical proliferations of the breast.Histopathology 2000; 37: 232–240.

23. Raju U, Crissman JD, Zarbo RJ, Gottlieb C. Epitheliosis of thebreast. An immunohistochemical characterization and comparisonto malignant intraductal proliferations of the breast. Am J SurgPathol 1990; 14: 939–947.

24. Achtstaetter T, Hatzfeld M, Quinlan RA, Parmelee DC,Franke WW. Separation of cytokeratin polypeptides by gelelectrophoretic and chromatographic techniques and theiridentification by immunoblotting. Methods Enzymol 1986; 134:355–371.

25. Walker RA. Are all ductal proliferations of the breastpremalignant? J Pathol 2001; 195: 401–403.

26. Marshall LM, Hunter DJ, Connolly JL, et al. Risk of breast cancerassociated with atypical hyperplasia of lobular and ductal types.Cancer Epidemiol Biomarkers Prev 1997; 6: 297–230.

27. Buerger H, Mommers EC, Littmann R, et al. Correlation ofmorphologic and cytogenetic parameters of genetic instability withchromosomal alterations in in situ carcinomas of the breast. Am JClin Pathol 2000; 114: 854–859.

28. Shoker BS, Jarvis C, Sibson DR, Walker C, Sloane JP. Oestrogenreceptor expression in the normal and pre-cancerous breast. JPathol 1999; 188: 237–244.

29. Tavassoli FA. Pathology of the Breast (2nd edn). Appleton andLange: Norwalk, 2000.

30. Page DL, Salhany KE, Jensen RA, Dupont WD. Subsequentbreast carcinoma risk after biopsy with atypia in a breast papilloma.Cancer 1996; 78: 258–266.

31. Buerger H, Otterbach F, Simon R, et al. Comparative genomichybridization of ductal carcinoma in situ of the breast — evidenceof multiple genetic pathways. J Pathol 1999; 187: 396–402.

32. Buerger H, Otterbach F, Simon R, et al. Different geneticpathways in the evolution of invasive breast cancer are associatedwith distinct morphological subtypes. J Pathol 1999; 189:521–526.

33. Diallo R, Schaefer KL, Poremba C, et al. Monoclonality in normalepithelium and in hyperplastic and neoplastic lesions of the breast.J Pathol 2001; 193: 27–32.

34. Kasami M, Vnencak-Jones CL, Manning S, Dupont WD,Page DL. Loss of heterozygosity and microsatellite instabilityin breast hyperplasia. No obligate correlation of these geneticalterations with subsequent malignancy. Am J Pathol 1997; 150:1925–1932.

35. Radford DM, Fair KL, Phillips NJ, et al. Allelotyping of ductalcarcinoma in situ of the breast: deletion of loci on 8p, 13q, 16q,17p and 17q. Cancer Res 1995; 55: 3399–3405.

36. Ellis IO, Pinder SE, Lee AH, Elston CW. A critical appraisal ofexisting classification systems of epithelial hyperplasia and insitu neoplasia of the breast with proposals for future methods ofcategorization: where are we going? Semin Diagn Pathol 1999;16: 202–208.

Copyright 2002 John Wiley & Sons, Ltd. J Pathol 2002; 198: 458–467.