Immunohistochemical distribution pattern of intermediate filament proteins and muscle actin in feline and human mammary carcinomas

J. Comp. Path. 1994 Vol. 111,365 381

Immunohistochemical Distribution Pattern of I n t e r m e d i a t e F i l a m e n t P r o t e i n s a n d M u s c l e A c t i n in Feline and Human Mammary Carcinomas J. Martin de las Mulas, A. E s p i n o s a de los Monteros*, i . J . Bautista, J. C. G 6 m e z - V i l l a m a n d o s and C. Morales t Department of Comparative Pathology, Veterinary Faculty, University of Cdrdoba, Avda. Medina Azahara s/n, 14005 Cdrdoba, *Department of Pathology, Veterinary Faculty, University of Las Palmas, c~ Francisco Inglott Artiles 12A, Las Palmas de Gran Canaria, and tDepartment of Pathology, Medical School, University of C6rdoba, Avda. Men#ndez Pidal s/n, 14071 Cdrdoba,

Spain Summary Thirty-seven feline and 38 human spontaneous mammary gland carcinomas were studied immunohistochemically. Commercially available antibodies directed against high and low molecular weight keratins (RCK-102 and NCL5D3), vimentin, desmin, glial fibrillary acidic protein (GFAP), neurofilament (NF) proteins and muscle actin (HHF35) were used in the avidin biotin peroxidase complex (ABC) technique on formalin-fixed paraffin wax-embedded tumour tissue samples. Healthy feline and human mammary gland tissue adjacent to the neoplasms was also examined. The distribution pattern of intermediate filament proteins and muscle actin was comparable in healthy mammary gland tissue of the two species: both RCK-102 and NCL-5D3 antibodies reacted with luminal epithelial cells of ducts and acini, but basal/ myoepithelial cells were stained by RCK-102 exclusively. In addition, basal/ myoepithelial cells expressed vimentin and muscle actin in both species, and GFAP was found in some feline basal/myoepithelial cells. No immunoreactivity to desmin and NF proteins Was observed. Feline m a m m a r y gland carcinoma cells reacted with RCK-102 (89%), NCL-5D3 (62%), vimentin (76%) and GFAP (30%) antibodies, while human mammary gland carcinoma cells reacted with RCK-102 (95 %), NCL-5D3 (100%) and vimentin (13%) antibodies. HHF35 immunoreactivity was observed in stromal cells only. These results indicate that mammary gland carcinomas of both species share a heterogeneous immunophenotype with respect to intermediate filament proteins, which adds to the list of known similarities between mammary gland carcinomas of both species.

Introduction M a m m a r y tumours are the third most common feline neoplasm, exceeded in incidence only by neoplasms of the skin and haematopoietic tissues. The most common m a m m a r y tumours in cats are carcinomas, which occur spontaneously in aged animals (Hampe and Misdorp, 1974; Madewell and Theilen, 1987; Moulton, 1990). Breast carcinoma is the most frequent type of cancer in 0021-9975/94/080365 + 17 $08.00/0

9 1994 Academic Press Limited

366

J. Martin de las Mulas e t al.

women in western societies and its incidence has been increasing (Haagensen et al., 1981). Feline mammary carcinomas have numerous features in common with human bi~east carcinomas, for example, the aetiological role of steroid hormones, age'incidence, morphology and prognostic value of some histological features. In addition, feline mammary carcinomas frequently show infiltrative growth, lymph vessel invasion, and early metastasis to the regional lymph nodes and lungs. For these reasons they are considered to represent a good model for the study of human breast carcinoma (Weijer et al., 1972; Weijer et al., 1973; Misdorp and Weijer, 1980; Weijer and Hart, 1983; Frese, 1986; Page, 1991). Specific. turnout cell constituents or products (tumour markers) can be demonstrated.immunohistochemically. The intermediate filament (IF) proteins, which are part of the cell cytoskeleton, have been extensively studied (Osborn and Weber, 1983; Erlandson, 1989; Goldman et al., 1990). The human mammary gland contains cells expressing the two major epithelial phenotypes, luminal and basal/myoepithelial, that line the ductal tree (TaylorPapadimitriou and Lane; 1987). The overwhelming majority of breast carcinomas are composed of cells with luminal epithelial cell immunophenotype, i.e. cells expressing keratins 7, 8, 15, 16, 18 and 19 (Tsubura et al., 1991), but some of them show immunohistochemical heterogeneity. Thus, basal/ myoepithelial cell markers, such as keratins 5, 14 and 17, vimentin, glial fibrillary acidic protein' (GFAP) and alpha-smooth muscle actin, are also expressed by.breast carcinoma cells (Guelstein et al., 1988; Wetzels et al., 1989; Gould et al., 1990; Viale et al., 1991; Senzaki et al., 1992). Actin, which is present in microfilaments, is known to occur in at least six isoforms in mammals (Vandekerckhove and Weber, 1978). Myoepithelial cells are characterized by the presence of actin in larger amounts than in secretory and ductal cells (Bussolati, 1980; Tsukada et al., 1987). The immunophenotype of non-neoplastic and neoplastic feline mammary glands is not well known. Isolated studies with a variety of tissue fixation methods, immunohistochemical procedures and intermediate filament p.rotein antibodies (Ivanyi et al., 1999; Martin de las Mulas et al., 1994) have revealed both similarities and differences between feline and human mammary tissue. In addition, the immunophenotype of two established feline mammary carcinoma cell lines was analysed by Ivanyi et al. (1992). The purpose of this study was to examine, by means of uniform procedures, intermediate filament proteins and muscle actin in feline and human healthy mammary glands and mammary carcinomas. Materials

and Methods

Mammary Tumours

Thirty-seven feline and 38 human mammary gland carcinomas were examined. The feline turnouts were classified histologically (Hampe and Misdorp, 1974) as follows: simple tubular (10), simple papillary (8), simple solid (6), mucinous (4) and simple but of combined histological types (9). The human carcinomas were classified (World Health Organization, 1982) as: infiltrative ductal but not otherwise specified (NOS)

Feline and Human

Mammary

367

Carcinomas

Table 1 Antibodies used in this study

Antibody

7~pe

Specificity

Dilution

Source

RCK-102 NCL-5D3 NFs HHF35

Mousemonoclonal Mousemonoclonal Mouse monoclonal Mouse monoclonal

1 in 20 1 in 20 1 in 20 1 in 300

Euro-Diagnostics B.V.* Euro~ B.V.* Euro-Diagnostics B.V.* Enzo-Diagnostics Inc.~"

Vimentin Desmin GFAP

Rabbit polyclonal Rabbit polyclonal Rabbit polyclonal

Human K5+8 Human K8 + 18 + 19 Human 70 and 200 kDa Human alpha and gamma muscle actin Calf lens vimentin Chickengizzard desmin Human GFAP

l in 160 1 in 160 1 in 160

Euro-Diagnostics B.V.* Euro-Diagnostics B.V.* Euro-Diagnostics B.V.*

NFs=neurofilament triplet proteins; GFAP=glial fibrillary acidic protein; K=human keratin type (s), according to the human catalogue of Moll et al. (1982). *Apeldoorn, The Netherlands; ~New York.

(24), papillary (2), colloid (4), medullary (4) and infiltrative lobular (4). The tumour tissue samples were submitted as surgical excision specimens in 10 % buffered formalin. The fixation time, though uncertain, was probably 24-48 h in most cases.

Healthy Mammary 7~ssue Samples Four feline and five human tissue samples of histologically normal m a m m a r y gland adjacent to the neoplasms were evaluated and served as internal controls.

Examination of 77ssue Samples Diseased and normal specimens were embedded in paraffin wax and sections were cut at a thickness of 4 gm. The antibodies employed in this study, their dilutions and their sources are listed in Table 1. The immunohistochemical technique used was the avidin biotin peroxidase complex (ABC) method (prediluted kit; Vector Corp., Burlingame, CA, USA). Dewaxed sections were incubated with 1% hydrogen peroxide in methanol for 30 min to block endogenous peroxidase activity. Predigestion of tissue sections with 0'1% pronase (Protease E; Sigma, St Louis, MO, USA) was used with the RCK-102 and NCL-5D3 monoclonal antibodies (MABs) according to the recommendations of the manufacturer for formalin-fixed tissue samples. Optimal concentration (0"1%) and incubation time (10 rain) of pronase were determined at room temperature on feline and h u m a n m a m m a r y gland tissue. After the proteolytic enzyme treatment, the slides were covered with normal swine (for polyclonal antibodies) or goat (for MABs) sera 1% in phosphate buffered saline (PBS) for 30 min before incubation with the primary antibodies for 18 h at 4~ The biotinylated swine antirabbit immunoglobulin (Vector Corp.; pre-diluted) and goat anti-mouse immunoglobulin (Nordic Imm, Tilburg, The Netherlands) antibodies, diluted 1 in 20 in PBS containing normal swine and goat serum 1%, respectively, were incubated for 30 min and the ABC (Vector Corp., Burlingame, CA, USA) for 60 min, both at room temperature. Between each step, slides were washed three times for 10min in PBS. The chromogen, 0"5% 3,3 diaminobenzidine tetrahydrochloride (Sigma, St Louis, MO, USA) diluted 1 in 10 in 0"05M Tris-containing hydrogen peroxide 0"3%, was applied to the slides for 2-3 min at 20-22~ Slides were counterstained with Mayer's haematoxylin. Routinely processed tissue samples of skin, m a m m a r y gland, cerebrum, skeletal muscle and intestinal wall were initially tested to establish the appropriate dilution of each antibody in each species. As negative controls, the substitution of primary

368

J.

Martin

d e l a s M u l a s e t al.

Table 2 Positive reactions with various m o n o c l o n a l and p o l y c l o n a l a n t i b o d i e s in s e c t i o n s of n o r m a l and c a r c i n o m a t o u s feline and h u m a n m a m m a r y t i s s u e Positive reactions in Antibody

Cells of normal* Feline mammary tissue

RCK-102 NCL-5D3 NFs HHF35 Vimentin Desmin GFAP

Luminal cells, 50% Basal cells, 100% Luminal cells, 90% . Basal cells, 100% Basal cells, 80% -Basal cells, 10%

Feline carcinomas Human carcinomas (n = 37) (n = 38)

Human mammary tissue Luminal cells, 60% Basal cells, 100% Luminal cells, 90% . . Basal cells, 100% Basal cells, 90% -None

33 (89%)

36 (95%)

23 (62%)

38 (100%)

2l (57%)'~ 28 (76%)

16 (42%)-~ 7 (13%) -0

.

11 (30%)

* Results in normal tissues based on four feline and five h u m a n specimens. "~Stromal cells. NFs, neurofilament proteins; GFAP, glial fibrillary acidic protein; - , negative.

antibodies by PBS, non-immune rabbit serum (for polyclonal antibodies) and mouse ascitic fluid (for MABs) at the same dilution as the specific reagents was employed. Results

Normal Feline and Human Mammary Glands The results are shown in Table 2. Luminal epithelial cells from the ducts and acini of both feline and human tissues reacted with antibodies directed against keratin proteins. The RCK-102 (MAB) staining reaction was weak and heterogeneous (i.e. not all cells of a single type were stained) in feline luminal epithelial cells and s t r o n g a n d homogeneous (the majority of cells of a single type were stained) in h u m a n luminal epithelial cells. The NCL-5D3 (MAB) staining reaction was strong and homogeneous in both feline and human luminal epithelial cells. Basal/myoepithelial cells from ducts and acini were reactive with RCK-102, vimentin and HHF35 (actin) antibodies. In one cat, the basal/myoepithelial cells were additionally stained by the GFAP antiserum. Mesenchymal stromal cells were reactive with the vimentin antibody, vascular wall smooth muscle cells with the desmin antibody, and peripheral nerves with both GFAP and neurofilament (NF) protein antibodies.

Feline Mammary Tumours The carcinoma cells reacted with RCK-102 and NCL-5D3 (keratin) MABs, and with vimentin and GFAP polyclonal antibodies (Table 2). The reactivities of the different histological types of tumour are shown in Table 3. R C K - 102 (MAB) reacted with 33 out of 37 tumours (89%). Immunoreactivity occurred in all tubular and mucinous carcinomas (Fig. 1), being strong and heterogeneous in >60% of the tumour cells; it also occurred in most of the

Feline and Human Mammary

369

Carcinomas

Table 3 Reactlvitles of feline and h u m a n m a m m a r y gland carcinomas with various antibodies

Species Feline

Human

Number of tumour specimens reacting with the following antibodies

Histological type of tumour

Total number of tumour specimens RCK-102

Simple tubular Simple papillary Simple solid Mucinous Simple combined Infiltrative ductat NOS Papillary Medullary Colloid Infiltrative lobular

10 8 6 4 9

I0 6 5 4 8

8 4 2 3 6

6 6 5 3 8

5 2 0 2 2

6 4 5 3 3

24 2 4 4 4

23 2 3 4 4

24 2 4 4 4

3 0 2 0 0

0 0 0 0 0

10 1 3 0 2

NCL-5D3

r/~mentin GFAP

HHF35*

*Stromal cells. NOS, not otherwise specified.

papillary, solid and combined carcinomas (Fig. 2). All immunoreactive cells were of the luminal epithelial type. Squamous metaplastic cells (in two cases of the combined type) were more deeply stained than their non-metaplastic counterparts (Fig. 2). NCL-5D3 (MAB) gave positive results in fewer cases (62%) and the staining reaction was weaker and heterogeneous. All immunoreactive cells were of the luminal epithelial type. Squamous metaplastic cells were unreactive. Vimentin immunoreactivity, which occurred in 76% of feline tumours, was seen in luminal epithelial-like cells indistinguishable from their non-reactive neighbours in all five histological types of tumour (Fig. 3). The positive parenchymal neoplastic cells varied from l0 to 70%. In addition, sloughed tumour cells and basally located cells at the epithelial stromal junction of some tumour nests were also reactive. All vimentin-positive cells were also labelled by RCK-102 (keratin) antibody, and some of the luminal epitheliallike and sloughed tumour cells were additionally labelled by NCL-5D3 (keratin) antibody. In the stroma, vimentin immunoreactivity was mostly fibrillar, with wide bands sometimes surrounding neoplastic ducts and nests. GFAP immunoreactivity, which occurred in 30% of tumours, was seen in luminal epithelial-like cells in all types of tumours except solid carcinomas. Positive cells were scattered among non-reactive cells in variable numbers, with no particular distribution pattern (Fig. 4). In addition, GFAP antibody reacted with a single, continuous or discontinuous layer of cells at the epithelial stromal junction of some tumour nests (Fig. 5A). Both types of GFAP-positive cells were also labelled by vimentin and RCK-102 antibodies, but GFAPpositive and vimentin-negative tumour cells were occasionally seen. In one example of the combined histological type, wide bands of immunoreactive fibrillar material were seen surrounding nests and cords of tumour cells. HHF35 (actin) MAB immunoreactivity was observed both in elongated cells

370

J. Martin de las Mulas et al.

Fig. 1.

Feline mucinous carcinoma. RCK'- 102 keratin antibody. Immunoreactivity is seen in the cytoplasm of some tumour cells. ABC method, x 20.

Fig. 2.

Feline combined histological type adenocarcinoma. RCK-102 antibody. Squamous metaplastic cells (lower half) are more deeply stained than non-metaplastic cells (upper half). ABC method. x20.

Feline and H u m a n M a m m a r y Carcinomas

371

Fig. 3.

Feline solid carcinoma. Vimentin antibody. Immunoreactivity in the cytoplasm of some luminal epithelial-like ceils (right upper corner). Positive internal controls (endothelial and stromal cells, left lower corner) are seen. ABC method, x 40.

Fig. 4.

Feline combined histological type adenocarcinoma. Glial fibrillary acidic protein antibody. Immunoreactivity is seen in the cytoplasm of some luminal epithelial-type cells scattered a m o n g nonreactive counterparts. Positive internal control (peripheral nerve, right lower corner) is seen. ABC method, x 40.

372

j: Martin

d e l a s M u l a s e t al.

Fig. 5.

Feline combined histological type adenocarcinoma. Glial fibrillary acidic protein (A) and HHF35 muscle actin (B,C) antibodies. A single, continuous or discontinuous layer of cells reacts with both glial fibrillary acidic protein (A, x 40) and muscle actin (B, x 20) antibodies. Abundant muscle actin immunoreactivity is seen in elongated cells in the tumour stroma (C, • 10). Positive internal controls (vascular smooth muscle ceils, arrows) are seen (B,C). ABC method.

Fig. 6.

H u m a n NOS infiltrative ductal carcinoma. RCK-102 keratin antibody. Immunoreactivity is strong in the cytoplasm of some tumour cells. ABC method. • 20.

Feline and Human Mammary Carcinomas

373

(57%) in the t u m o u r stroma (Fig. 5C) and in a single, continuous layer of cells at the epithelial stromal junction of some tumour nests (Fig. 5B). T h e former reacted with the vimentin antibody also, and the latter with vimentin, RCK-102 and GFAP antibodies. In addition, very small, isolated charcoallike areas (spherical masses of elongated or stetlate cells separated by a colourless to slightly basophilic staining ground substance), present in two tumours, were picked out by HHF35 immunoreactivity; the positive cells were star-shaped or elongated and reacted also with the vimentin antibody but not with GFAP or RCK-102 antibodies. Human Mammary Turnouts

The carcinoma cells reacted with RCK-102 and NCL-5D3 (keratin) MABs, and with vimentin polyclonal antibody (Table 2). T h e reactivities of the different histological types of tumour are shown in Table 3. RCK-102 MAB reacted with neoplastic cells in 36 out of 38 (95%) tumours. Immunoreactivity was strong and heterogeneous in both infiltrative ductal (Fig. 6) and infiltrative lobular carcinomas. NCL-5D3 reacted with all tumours, strongly and homogeneously (Fig. 7). Vimentin immunoreactivity was found in luminal, epithelial-type cells in two N O S and three medullary infiltrative ductal carcinomas (13%) (Fig. 8). In addition, basally located cells at the epithelial stromal junction of some t u m o u r nests and stromal cells were also stained by this antibody. Some of the former had clearly defined "in situ" patterns of growth, such as comedocarcinoma or cribiform carcinoma (Fig. 9), and vimentin-positive cells were also labelled by the RCK-102 antibody. HHF35 (actin) MAB immunoreactivity was observed both in elongated cells in the tumour stroma (42%) and in a single, continuous layer of cells at the epithelial stromal junction of some t u m o u r nests with clearly defined "in situ" patterns of growth (Fig. 10). The former reacted with the vimentin antibody also, and the latter with vimentin and RCK-102 antibodies. Discussion

The commercially available IF protein antibodies, which were raised against antigens from different m a m m a l i a n species (Table 1) reacted with both feline and h u m a n tissues. Acinar and ductal luminal epithelial cells in feline and h u m a n healthy m a m m a r y glands showed a reaction with both NCL-5D3 and RCK-102 MABs. This accords with the previous demonstration of low molecular weight (simple epithelial) keratins in these cells. Thus, in man, keratins 7, 8, 15, 16, 18 and 19 are expressed by luminal cells, while keratins 5, 14 and 17 are found in basal celts (Guelstein et al., 1988; Wetzels et al., 1989; Tsubura et al., 1991; B6cker et al., 1992a; Ivanyi et al., 1992); and in the cat keratins 5 + 8 and 6 + 18 are expressed by acinar luminal cells, and keratins 5 + 8, 6 + 18 and 7 by ductat luminal celIs (Ivanyi et al., 1992). This keratin protein distribution pattern in h u m a n and feline m a m m a r y glands is similar to that described in dogs, in which keratins 6 + 8 (Hellm~n and

374

J.

M a r t i n de l a s M u l a s e t M.

Fig. 7.

H u m a n N O S infiltrative ductal carcinoma. NCL-5D3 antibody. All tumour cells in both tubular and cribiform structures are reactive. Positive internal control (trapped lobule) is seen. ABC method, x I0.

Fig. 8.

H u m a n medullary carcinoma. Vimentin antibody. Immunoreactivity is strong in the cytoplasm of some tumour cells which are intermingled with morphologically indistinguishable non-reactive counterparts. Internal control (stromal cells and inflammatolN cells) are positive. ABC method. x 20.

Feline and H u m a n M a m m a r y C a r c i n o m a s

Fig. 9.

Fig. 10.

375

H u m a n N O S infiltrative ductal carcinoma. Vimentin antibody. Immunoreactivity is seen both in basally-located cells at the epithelial-stromal junction and in some t u m o u r cells in areas of "in situ" cribiform carcinoma. Positive internal controls (endothelial, stromal and vascular smooth muscle cells) are seen. ABC method, x 20. H u m a n N O S infiltrative ductal carcinoma. H H F 3 5 muscle actin antibody, lmmunoreactivity is seen in a continuous or discontinuous layer of cells at the epithelial-stromal junction in solid t u m o u r nests and areas of "in situ" comedocarcinoma. Positive internal control (vascular smooth muscle cells) is seen. ABC method, x 10.

376

J.

Martin de las Mulas e t

al.

Lindgren, 1989), 8 + 1 8 and 19 (Destexhe et al., 1993), 1-8, 10, 14-16, 18 and 19 (Griffey et a[., 1993) and 5 + 8 , 8, 18, 7 and 19 (Vos et al., 1993a) are expressed by luminal cells. However, acinar and ductal luminal cells that express basal keratins 5 and 14 have been described in man (Wetzels eta[., 1989; B6cker et al., 1992@ Because of this, the immunoreactivity observed in our study, as well as in that of Ivanyi et al. (1992), of acinar and ductal luminal cells with RCK-102 MAB may be attributed to the recognition of keratin 8, keratin 5, or both. Acinar and ductal basal/myoepithelial cells in human healthy mammary glands reacted with RCK-102, vimentin and HHF35 antibodies, and feline basal/myoepithelial cells reacted additionally with GFAP antibody. The presence of high molecular weight keratin proteins in human basal/myoepithelial cells is well known (Guelstein et a[., 1988; Wetzels et al., 1989; Tsubura et al., 1991; B6cker et aL, 1992a; Ivanyi et al., 1992) and has also been described in cats (Ivanyi et al., 1992), dogs (Griffey et al., 1993; Vos et al., 1993a), rats (Lichtner et al., 1991), and mice (Smith et al., 1990). In addition, vimentin, alpha-smooth muscle actin, common acute lymphoblastic leukaemia antigen (CALLA), myosin, S-100 protein and GFAP have been described in human basal/myoepithelial cells (Gusterson et al., 1986; Guelstein et al., 1988; Gillet et al., 1990; Gould et al., 1990; Viale et al., 1991; B6cker et al., 1992a; Senzaki et aL, 1992); low molecular weight keratin, vimentin and alpha-smooth muscle actin in canine mammary cells (Hellm~n and Lindgren, 1989; Destexhe et al., 1993; Griffey et al., 1993; Vos et al., 1993a); vimentin and desmin in routine mammary cells (Pefia et aL, 1990; Lichtner et al., 1991); and vimentin in feline mammary cells (Martin de las Mulas et al., 1994). All these cells were of the basal/myoepithelial type. In contrast, however, lack of vimentin immunoreactivity has also been reported in the species mentioned above (TaylorPapadimitriou et al., 1983; Trojani et al., 1991; Vos et al., 1993a), with the exception of cats. In man, alpha-smooth muscle actin seems to be the most consistent and diagnostically reliable marker, whereas vimentin, GFAP and even keratins 5 and 14 are only sometimes present in a subset ofmyoepithelial cells (Tsukada eta[., 1987; Gould et al., 1990; Viale eta[., 1991; B6cker et al., 1992a). The feline and human tumours included in this study covered the range of most frequently diagnosed mammary gland carcinomas in the two species. Most mammary carcinomas in cats are tubular, papillary, or solid, or a combination of all three (Hampe and Misdorp, 1974). In women, infiltrating carcinoma (World Health Organization, 1982) and ductal carcinoma with productive fibrosis (McDivitt et al., 1968; Adair et al., 1974) are the most common types. Complex carcinomas and sarcomas are rare in the cat, as in women (McDivitt et al., 1968; Hampe and Misdorp, 1974; Moulton, 1990). In this study both feline and human luminal epithelial-type carcinoma cells reacted with RCK-102, NCL-5D3 and vimentin antibodies. Feline luminal epithelial-type tumour cells were additionally labelled by GFAP antibody. The immunoreactivity of human and feline mammary gland carcinomas with RCK- 102 monoclonal antibody was comparable. Thus, a high percentage

Feline and H u m a n M a m m a r y Carcinomas

377

of tumours were labellect by this antibody, and distribution among the histological types was homogeneous. As a result, RCK-102 MAB would appear to be a good marker for formalin-fixed, paraffin wax-embedded feline and human mammary carcinoma tissue. In contrast, NCL-5D3 MAB is a highly sensitive marker for routinely processed human breast carcinoma tissue, as has already been reported (Angus et al., 1987), but is less sensitive for feline mammary carcinoma tissue. Vimentin antibody reacted with luminal epithelial-type tumour cells in both species, but the percentage of positive cases was much higher in cats. Coexpression of vimentin and keratin IF protein has been described in some epithelial tumours (Azumi and Battifora, 1987), including mammary gland tumours, with a percentage frequency ranging from 9"2% (Raymond and Leong, 1989) to 23% (Domagala et al., 1990). In addition to expression induced by the loss of cell-to-cell contact in sloughed tumour cells (Raymond and Leong, 1989), which was also observed in our study, vimentin immunoreactivity in epithelial tumours may be attributed to aberrant expression or to a reversion to a more primitive or embryonic cell type (Gould, 1986). Thus, the expression of vimentin (Vos et al., 1993a, 1993b) and NF protein (Hellm~n and Lindgren, 1989) has been described in canine mammary tumours without expression of those antigens in non-neoplastic cells. In addition, vimentin expression may be attributed to the presence of a subset of rapidly proliferating cells. In fact, vimentin expression in renal (Dierick et al., 1991), pulmonary (Upton et al., 1986) and breast (Raymond and Leong, 1989; Domagala et al., 1990) carcinomas has been associated with poor prognosis. Some of these vimentin-positive cells were additionally labelled in feline carcinomas by the GFAP antibody, but all of them were unreactive with HHF35 antibody. In other species, it has been suggested that the basal/myoepithelial layer of ducts and acini contains different types of cells (Hamperl, 1970), including stem cells (TaylorPapadimitriou et al., 1983; Sonnenberg et al., 1986; Gugliotta et al., 1988; Vos et al., 1993b), and post-stem cells at different stages of proliferation (Gould et al., 1990; B6cker et al., 1992b). The possibility exists, however, that GFAP antibody cross-reacts with an epitope recognized by vimentin antibody in feline but not in human tissues. GFAP antibody immunoreactivity has been described in both healthy, and pathological human mammary glands (Gould et al., 1990; Diaz et al., 1991, 1992; Viale et al., 1991). The lack of reactivity of human tissues with GFAP antibody in our study may be attributable to differences in antibodies, fixation and immunohistochemical methods employed. However, variation in differentiation of the cells should be considered also. Basally located cells at the epithelial-stromal junction of some feline and human tumour nests would represent keratin (5 +8), vimentin and muscle actin positive (human) and keratin (5 + 8), vimentin, muscle actin and GFAP positive (feline) non-tumour myoepithelial cells, indicating non-invasive foci of carcinoma (Guelstein et al., 1988; Smith et al., 1990; Vos et al., 1993b). Finally, elongated stroma cells positive with both vimentin and HHF35 antibodies probably represent activated fibroblasts or myofibroblasts; they have been observed in mammary gland carcinomas of other species, including

378

J. Martin de las Mulas e t al.

man and the dog (Tsukada et al., 1987; Destexhe et al., 1993; Griffey et al., 1993). In conclusion, our results show that feline and human healthy m a m m a r y glands have comparable immunohistochemical distribution patterns with respect to high and low molecular weight keratin proteins, vimentin and muscle actin, and that m a m m a r y gland carcinomas of both species share a heterogeneous immunophenotype with respect to IF proteins. Possibly different immunophenotypes in feline and human m a m m a r y gland carcinomas reflect biologically distinct forms of proliferation. The analysis of non-malignant tumours of the m a m m a r y glands, as well as prognosis-related studies, are needed to clarify this point. Acknowledgments

The excellent technical assistance of Pilar Cano is greatly appreciated. References

Adair, F., Berg, J., Joubert, L. and Robbins, G. F. (1974). Long-term follow-up of breast cancer patients: the 30-year report. Cancer, 33, 1145-1150. Angus, B., Purvis, J., Stock, D., Westley, B. R., Samson, A. C. R., Routledge, E. G., Carpenter, E H. and Horne, C. H. W. (1987). NCL-5D3: a new monoclonal antibody recognizing low molecular weight cytokeratins effective for immunohistochemistry using fixed paraffin-embedded tissue. Journal of Pathology, 153, 377-384. Azumi, N. and Battifora, H. (1987). The distribution of keratin and vimentin in epithelial and non-epithelial neoplasms. American Journal of Clinical Pathology, 88, 286-296. B6cker, W., Bier, B., Freytag, G., Br6mmelkamp, B., Jarasch, E.-D., Edel, G., Dockhorn-Dworniczak, B. and Schmid, K. W. (1992a). An immunohistochemical study of the breast using antibodies to basal and luminal keratins, alpha-smooth muscle actin, vimentin, collagen IV and laminin. Part I: normal breast and benign proliferative lesions. Nrchows Archiv. A Pathological Anatomy, 421, 315-322. B6cker, W., Bier, B., Freytag, G., Br6mmelkamp, B., Jarasch, E.-D., Edel, G., Dockhorn-Dworniczak, B. and Schmid, K. W. (I992b). An immunohistochemical study 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. Nrchows Archiv. A Pathological Anatomy, 421, 323-330. Bussolati, G. (1980). Actin-rich (myoepithelial) cells in lobular carcinoma in situ of the breast. FirchowsArchiv. B Cell Pathology, 32, 165 176. Destexhe, E., Lespagnard, L., Degeyter, M., Heyman, R. and Coignoul, F. (1993). Immunohistochemical identification of myoepithelial, epithelial and connective tissue cells in canine mammary tumors. Veterinary Pathology, 30, 146-154. Diaz, N. M., McDivitt, R. W. and Wick, M. R. (1991). Pleomorphic adenoma of the breast: a clinicopathologic and immunohistochemical study of 10 cases. Human Pathology, 22, 1206-1214. Diaz, N. M., Palmer, J. O. and Wick, M. R. (1992). Erosive adenomatosis of the nipple: histology, immunohistology, and differential diagnosis. Modern Pathology, 5, 179-184. Dierick, A., Praet, M., Roels, H., Verbeeck, E, Robyns, C. and Oosterlinck, W. (1991). Vimentin expression in renal cell carcinoma in relation to DNA content and histological grading: a combined light microscopic, immunocytochemical and cytophotometrical analysis. Histopathology, 18, 315-322. Domagala, W., Lasota, J., Bartkowiak, J., Weber, K. and Osborn, M. (1990). Vimentin

Feline and H u m a n M a m m a r y Carcinomas

379

is preferentially expressed in human breast carcinomas with low estrogen receptor and high Ki-67 growth fraction. American Journal of Pathology, 136, 219 227. Erlandson, R. A. (1989). Cytoskeletal proteins including myofilaments in human tumors. Ultrastructural Pathology, 13, 155-186. Frese, K. (1986). Comparative pathology of mammary tumors of domestic animals. In: Pathology of Neoplastic and Endocrine induced Diseases of the Breast. R. B~issler and K. Habner, Eds. Gustav Fischer Verlag, Stuttgart, New York, pp. 44-61. Gillett, C. E., Bobrow, L. G. and Millis, R. B. (1990). S-100 protein in human mammary tissue. Immunoreactivity in breast carcinoma, including Paget's disease of the nipple, and value as a marker of myoepithelial cells. Journal of Pathology, 160, 19-24. Goldman, .R.D., Zackroff, R. V. and Steinert, E M. (1990). Intermediate filaments. An overview. In: Cellular and Molecular Biology of Intermediate Filaments. R. D. Goldman and E M. Steinert, Eds. Plenum Publishing Corporation, New York, pp. 3-17. Gould, V. E. (1986). Histogenesis and differentiation: a reevaluation of these concepts as criteria for the classification of tumours. Human Pathology, 17, 212-214. Gould, V. E., Koukoulis, G. K., Jansson, D. S., Nagle, R. B., Franke, W. W. and Moll, R. (1990). Coexpression patterns of vimentin and glial filament protein with cytokeratins in the normal, hyperplastic and neoplastic breast. American Journal of Pathology, 137, 1143-1155. Griffey, S. M., Madewell, B. R., Dairkee, S. H., Hunt, J. E., Naydan, D. K. and Higgins, R. J. (1993). Immunohistochemical reactivity of basal and luminal epithelium-specific cytokeratin antibodies within normal and neoplastic canine mammary glands. Veterinary Pathology, 30, 155-161. Guelstein, V. I., Tchypysheva, T. A., Ermilova, V. D., Litvinova, L. V., Troyanovsky, S. M. and Bannikov, G. A. (1988). Monoclonal antibody mapping of keratins 8 and 17 and of vimentin in normal human mammary gland, benign tumors, dysplasias and breast cancer. International Journal of Cancer, 42, 147-153. Gugliotta, E, Sapino, A., Macri, L., Skalli, O., Gabbiani, G. and Bussolati, G. (1988). Specific demonstration of myoepithelial cells by anti-alpha smooth muscle actin antibody. Journal of Histochemistry and Cytochemistry, 36, 659-663. Gusterson, B. A., Monaghan, P., Mahendran, R., Ellis, J. and O'Hare, M.J. (1986). Identification of myoepithelial cells in human and rat breast by anti-common acute lymphoblastic leukemia antigen antibody A12. Journal of the National Cancer Institute, 77, 343-349. Haagensen, C. D., Bodian, C. and Haagensen, D. E. Jr (1981). The frequency of breast carcinoma. In: Breast Carcinoma: Risk and Detection. W.B. Saunders Company, Philadelphia, pp. 1-5. Hampe, J. E and Misdorp, W. (1974). Tumours and dysplasias of the mammary gland. Bulletin of the World Health Organization, 50, 111-133. Hamperl, H. (1970). The myothelia (myoepithelial cells) normal state; regressive changes; hyperplasia; tumors. Current Topics in Pathology, 53, 161-220. Hellmdn, E. and Lindgren, A. (1989). The expression of intermediate filaments in canine mammary glands and their tumors. Veterinary Pathology, 26, 420-428. Ivanyi, D., Minke,J. M. H. M., Hageman, C., Groeneveld, E. and van Doornewaard, G. (1992). Patterns of expression of feline cytokeratins in healthy epithelia and mammary carcinoma cells. American Journal of Veterinary ~esearch, 53, 304-314. Lichtner, R. B.,Julian, J. A., North, S. M., Glasser, S. R. and Nicolson, G. L. (1991). Coexpression of cytokeratins characteristic for myoepithelial and luminal cell lineages in rat 13762 NF mammary adenocarcinoma tumors and their spontaneous metastases. Cancer Research, 51, 5943-5950. Madewell, B. R. and Theilen G. H. (1987). Tumors of the mammary gland. In: Veterinary Cancer Medicine, G. H. Theilen and B. R. Madewell, Eds. Lea and Febiger, Philadelphia, pp. 327-344. Martin de las Mulas, J., Espinosa de los Monteros, A., Carrasco, L., Sierra, M. A.

380

J. Martin de las Mulas et al.

and Vos, J. H. (1994). Immunohistochemical distribution of vimentin, desmin, glial fibrillary acidic protein and neurofilament proteins in feline tissues. Journal of VeterinaryMedicine, A, 40, 1-15. McDivitt, R. W., Stewart, F. W. and Berg, J. W. (1968). Tumors of the Breast. Atlas of Tumor Pathology, series 2, fascicle 2. Armed Forces Institute of Pathology, Washington, D.C., pp. 22-132. Misdorp, W. and Weijer, K. (1980). Feline mammary carcinoma. AmericanJournal of Pathology, 98, 573-576. Moll, R., Franke, W. W., Schiller, D. L., Geiger, B. and Krepler, R. (1982). The catalogue of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Cell, 31, 11-24. Moulton, J. E. (1990). Tumors of the mammary gland. In: Tumors of DomesticAnimals, 3rd Edit.,J. E. Moulton, Ed. University of California Press, Berkeley, pp. 518-552. Osborn, M. and Weber, K. (1983). Tumor diagnosis by intermediate filament typing: a novel tool for surgical pathology. Laboratory Investigation, 48, 372-394. Page, L. (1991). Prognosis and breast cancer. Recognition of lethal and favorable prognostic types. AmericanJournal of Surgical Pathology, 15, 334-349. Pefia, L., Flores, J. M., Castafio, M. and GonzAlez, M. (1990). Intermediate filaments in DMBA-induced mammary adenocarcinomas in rats. Journal of VeterinaryMedicine, A, 37, 614-621. Raymond, W. A. and Leong, A. S.-Y. (1989). Vimentin--a new prognostic parameter in breast carcinoma? Journal of Pathology, 158, 107-114. Senzaki, H., Tsubura, A., Shoji, T., Okada, H., Takahashi, H. and Morii, S. (1992). Immunohistochemical detection of keratin, actin and type-IV collagen in serial sections of methacarn-fixed breast cancer tissues. Acta Histochemicaand Cytochemica, 25, 31 38. Smith, G. H., Mehrel, T. and Roop, D. R. (1990). Differential keratin gene expression in developing, differentiating, preneoplastic and neoplastic mouse mammary epithelium. Cell Growth and Differentiation, 1, 161-170. Sonnenberg, A., Daams, H., Valk, M. A. van der, Hilkens, J. and Hilgers, J. (1986). Development of mouse mammary gland: identification of stages in differentiation of luminal and myoepithelial cells using monoclonal antibodies and polyvalent antiserum against keratin. Journal of Histochemistry and Cytochemistry,8, 1037-1046. Taylor-Papadimitriou, J. and Lane, E. B. (1987). Keratin expression in the mammary gland. In: The Mammary Gland. Development, Regression and Function, M. C. Neville and C. W. Daniel, Eds. Plenum Press, New York, pp. 181-215. Taylor-Papadimitriou, J., Lane, E. B. and Chang, S. E. (1983). Cell lineages and interactions in neoplastic expression in the human breast. In: UnderstandingBreast Cancer. Clinical and Laboratory Concepts,M. A. Rich, J. C. Hager and R Furmanski, Eds. Marcel Decker, Inc., Basel, New York, pp. 215 246. Trojani, M., Demascarel, I. and Coindre, J. M. (1991). Adenoid cystic carcinoma of the breast value of immunohistochemical study in diagnosis. Tumori, 77, 130-135. Tsubura, A., Okada, H., Senzaki, H., Hatano, T. and Morii, S. (1991). Keratin expression in the normal breast and in breast carcinoma. Histopathology, 18, 517 522. Tsukada, T., Tippens, D., Gordon, D., Ross, R. and Gown, A. M, (1987). HHF 35, a muscle-actin-specific monoclonal antibody. American Journal of Pathology, 126, 51-60. Upton, M. R, Hirohashi, S., Tome, Y., Miyazawas, N., Suemasu, K. and Shimosato, Y. (1986). Expression of vimentin in surgically resected adenocarcinomas and large cell carcinomas of lung. AmericanJournal of Surgical Pathology, 10, 560-567. Vandekerckhove, J. and Weber, K. (1978). At least six different actins are expressed in a higher mammal: an analysis based on the amino acid sequence of aminoterminal tryptic peptide. Journal of Molecular Biology, 126, 783-802. Viale, G., Gambacorta, M., Coggi, G., Dell'Orto, R, Milani, M. and Doglioni, C.

Feline and H u m a n M a m m a r y Carcinomas

381

(1991). Glial fibrillary acidic protein immunoreactivity in normal and diseased human breast, ldrchows Archiv. (A), 418, 339 348. Vos, J. H., van den Ingh, T. S. G. A. M., Misdorp, W., Molenbeek, R. E, van Mil, E N., Rutteman, G. R., Ivanyi, D. and Ramaekers, E C. S. (1993a). Immunohistochemistry with keratin, vimentin, desmin, and alpha-smooth muscle actin monoclonaI antibodies in canine mammary gland: normal mammary gland tissue. Veterinary Quarte@, 15, 102-107. Vos, J. H., van den Ingh, T. S. G. A. M., Misdorp, W., Molenbeek, R. E, van Mil, F. N., Rutteman, G. R., Ivanyi, D. and Ramaekers, F. C. S. (1993b). Immunohistochemistry with keratin, vimentin, desmin, and alpha-smooth muscle actin monoclonal antibodies in canine mammary gland: malignant mammary tumours. Veterinary Quarterly, 15, 96-102. Weijer, K., Hampe, J. K. and Misdorp, W. (1973). Mammary carcinoma in the cat. A model in comparative cancer research? Archivum Chirurgicum Neerlandicum, 24-25, 413 425. Weijer, K. and Hart, A. M. (1983). Prognostic factors in feline mammary carcinoma. Journal of the National Cancer Institute, 70, 709 716. Weijer, K., Head, K. W., Misdorp, W. and Hampe, J. K. (1972). Feline malignant mammary tumors. I. Morphology and biology: some comparisons with human and canine mammary carcinomas. Journal of the National Cancer Institute, 49, 1697 1704. Wetzels, R. H., Holland, R., van Haelst, U . J . G . M . , Lane, E. B., Leigh, I. M. and Ramaekers, E C. S. (1989). Detection of basement membrane components and basal cell keratin 14 in noninvasive and invasive carcinomas of the breast. American Journal of Pathology, 134, 571 579. World Health Organization (1982). Histological Typing of Breast Tumors. International Histological Classification of Tumors, No. 2, 2nd Edit., Geneva, pp. 15-25.

Received, March 31st, 1994] Accepted, July 12th, 1994 J