Histopathology 1986, 10, 1315-1324

An analysis of the sensitivity and specificity of the cytokeratin marker CAM 5.2 for epithelial tumours. Results of a study of 203 sarcomas, 50 carcinomas and 28 malignant melanomas

M. LEADER*, J . PATEL*, C. MAKINt & K. HENRY * Departments of * Histopathology and t Surgery, Westminster Site, Charing Cross and Westminster Medical School, London, U K

Accepted for publication 12 May 1986

LEADER M . , PATEL J. , MAKIN C. & HENRY K. (1986) Histopathology 10, 1315-1324

An analysis of the sensitivity and specificity of the cytokeratin marker CAM 5.2 for epithelial tumours. Results of a study of 203 sarcomas, 50 carcinomas and 28 malignant melanomas

Two hundred and three sarcomas, 40 carcinomas, 10 carcinomas with spindle cell features, 27 malignant melanomas and one spindle cell melanoma were examined using CAM 5.2, a monoclonal antibody to cytokeratin. This antibody which was prepared against colorectal carcinoma cells and which identifies low molecular weight intermedi- ate filament cytokeratin proteins is suitable for use in formalin fixed, paraffin embedded material. Seventeen of the 203 sarcomas showed positive staining. These included 15/21 synovial sarcomas, 1/5 epithelioid sarcomas and 1/18 malignant neural tumours. Five carcinosarcomas showed positive staining of their epithelial components but negative staining of their spindle cell components; three out of four pure spindle cell carcinomas stained positively; a metastasis from a spindle cell renal carcinoma was negative. A spindle cell thymoma also stained positively. Thirty-seven of the 40 carcinomas stained positively; the three negative carcinomas were a squamous cell carcinoma, a renal cell carcinoma and an oat cell carcinoma. All malignant melanomas were negative. These results are compared with those of other workers and the sensitivity and specificity of CAM 5.2 as an epithelial marker is assessed.

Keywords: cytokeratin, CAM 5.2, sarcoma, carcinoma, synovial sarcoma, spindle cell sarcoma, carcinosarcoma

Introduction

The distinction between epithelial tumours and those of mesenchymal origin can be a difficult and recurring problem. Cytokeratin, an intermediate filament (Rungger- Brandle & Gabbiani 1983), may be of value in tumour diagnosis as intermediate

Address for correspondence: Dr M. Leader, Department of Histopathology, Westminster Medical School, Horseferry Road, London SWlP 2AR, UK.

1315

1316 M. Leader ef al.

filaments are retained during neoplastic transformation (Altmannsberger et al. 1981, Ramaekers etaf . 1983~). Makin, Bobrow & Bodmer (1984) raised an antibody, CAM 5.2, against colorectal carcinoma cells, which they showed reacted with intermediate filament cytokeratin proteins. However, before it can be accepted as a reliable epithelial tumour marker, its sensitivity and specificity for epithelial tumours must first be examined. We have done this by reacting CAM 5.2 with a large series of sarcomas and carcinomas.

Materials and methods

Two hundred and three sarcomas, 40 carcinomas, 10 carcinomas with spindle cell features, 27 malignant melanomas and one spindle cell melanoma were included in the study and stained with CAM 5.2. In addition, the 10 carcinomas with spindle cell features and the single spindle cell melanoma were also reacted with a commercial monoclonal vimentin antibody. All tumours were fixed in neutral buffered formalin and all were paraffin embedded. The sarcomas were identified from the files of the Westminster Hospital. The age of the blocks ranged from 1 to 28 years. All sarcomas had been examined by Professor D.H.MacKenzie (MacKenzie 1970) and had been diagnosed on the basis of haematoxylin and eosin and other appropriate non- immunocytochemical stains. The carcinomas were chosen retrospectively from the files of the Westminster Hospital. The blocks ranged in age from 1 to 20 years. The immunoperoxidase technique was used. The cytokeratin antibodies (obtained from

Table 1. Cytokeratin staining of 203 sarcomas, 40 carcinomas and 27 malignant melanomas using CAM 5.2

Results

No. Positive Negative

Synovial sarcoma Epithelioid sarcoma Neurofibrosarcomdmalignant schwannoma Leiom yosarcoma Rhabdomyosarcoma Clear cell sarcoma Kaposi’s sarcoma Angiosarcoma Haemangiopericytoma Malignant fibrous histiocytoma Liposarcoma Fibrosarcoma Sarcoma-unclassified Squamous cell carcinoma Adenocarcinoma Renal cell carcinoma Transitional cell carcinoma (bladder) Oat cell carcinoma Malignant melanoma

21 5

18 22 22 8 9 8

18 27 21 20

3 10 10 6 4

10 27

15 1 1 0 0 0 0 0 0 0 0 0 0 9

10 5 4 9 0

6 4

17 22 22 8 9 8

18 27 21 20

3 1 0 1 0 1

27

Cytokerutin staining of turnours 1317

the Imperial Cancer Research Fund) were used either neat or diluted to 1: 10 depend- ing on the batch, and vimentin (Labsystems) at a dilution of 1 :75. CAM 5.2 antibody is now available from Becton Dickinson.

Trypsinization for 30 min was required with the vimentin antibody. The need for trypsinization when using cytokeratin depended on the individual batch of CAM 5.2. It was used only in the latter part of the study. Whilst staining without trypsin was weaker, the use of trypsin did not appear to affect the number of cells staining. Appropriate controls were used throughout.

Sections were coded, examined and a positive or negative result was recorded on the basis of immunocytochemical staining in the absence of the original diagnosis. A positive result was defined as brown granular cytoplasmic staining. In the majority of cases this was diffusely seen within the cytoplasm. However, a pericellular distribution was noted in some cases. Other workers have had difficulty in distinguishing melanin staining from a positive DAB reaction. We did not have this difficulty. In our hands, melanin staining was darker and the granules coarser than in the immunocytochemical staining.

Results

CAM 5.2 stained all normal epithelial cells in the sections examined with the excep- tion of stratified squamous epithelium. The results of CAM 5.2 staining on the 281 tumours studied are summarized in Tables 1 & 2. For a more detailed analysis, we have divided the tumours into six categories, namely: carcinomas, carcinomas with spindle cell features, malignant melanomas, synovial sarcomas, epithelioid sarcomas and other sarcomas.

Table 2. CAM 5.2 and anti-vimentin staining of carcinomas with spindle cell features and a spindle cell melanoma

Results

CAM 5.2 Vimentin

Spindle cell carcinosarcoma of oesophagus Ep+ Sp- Ep- Sp+ Spindle cell carcinosarcoma of lung E p + Sp- Ep- Sp+ Spindle cell carcinosarcoma of endometrium Ep+ Sp- Ep- Sp+ Spindle cell carcinosarcoma of bladder E p + Sp- Ep- Sp- Spindle cell carcinosarcoma of pyriform fossa Ep+ Sp- Ep- Sp+

- Spindle cell carcinoma of lung Metastatic renal spindle cell carcinoma Metastatic renal cell carcinoma - Spindle cell carcinoma of the thyroid Spindle cell melanoma - Spindle cell thymoma +

+ + +

-

- -+ + -

Ep=Epithelial component; Sp=spindle cell component.

1318 M . Leader et al.

CARCINOMAS , CARCINOMAS WITH S P I N D L E CELL FEATURES A N D M A L I G N A N T

M E L A N O M A S

All 10 adenocarcinomas studied, regardless of primary site, stained positively (Figure 1). Nine of the 10 squamous cell carcinomas stained positively (Figure 2); four were well differentiated and six were moderately differentiated tumours. The negative result was obtained with a poorly differentiated squamous cell carcinoma. In general the better differentiated squamous cell carcinomas showed a greater number of positively staining cells than the moderately differentiated tumours. Five of six renal cell carcinomas showed cytoplasmic staining, including three clear cell carcinomas. All four transitional cell carcinomas stained positively. Nine of the 10 oat cell carcin- omas of lung stained positively. In all nine cases only a small number of tumour cells stained. All 27 malignant melanomas showed negative staining.

Amongst the carcinomas with spindle cell features, positive cytokeratin staining was seen in all but one tumour, a metastatic renal spindle cell carcinoma. Vimentin positivity was demonstrated in the sarcomatous component of 4/5 carcinosarcomas, in two spindle cell renal carcinomas and in the spindle cell melanoma.

SARCOMAS

Seventeen of the 203 sarcomas studied showed positively staining cells. These included 15 synovial sarcomas, one epithelioid sarcoma and one malignant schwannoma.

Figure 1 . Poorly differentiated carcinoma of the stomach: the turnour cells are stained with CAM S.2. x 400.

Cytokeratin staining of tumours 1319

Figure 2. This is a moderately differentiated squamous cell carcinoma of the cervix. The tumour cells are stained intensely with CAM 5.2. X400.

Figure 3. Biphasic synovial sarcoma stained with CAM 5.2. The epithelial-like cells have stained intensely. The spindle cell component stained weakly. ~ 4 0 0 .

1320 M . Leader et al.

Figure 4. Monophasic synovial sarcoma. The cytoplasm of a few spindle cells has stained with CAM 5.2 in a discrete and granular manner. ~ 4 0 0 .

Synovial sarcomas

The synovial sarcoma group consisted of 21 tumours, 12 monophasic synovial sar- comas and nine biphasic tumours. Amongst the biphasic tumours, the glandular areas in all stained positively, but to a variable extent (Figure 3 ) ; in seven nearly all epithelial-like cells stained positively, whilst in two cases only a minority did. The spindle cell component stained positively in six out of the nine biphasic tumours and in six of the 12 monophasic tumours (Figure 4). Thus, almost 60% of synovial sarcomas showed positively staining spindle cells. Seven tumours showed foci of calcification and could thus be classified as calcifying synovial sarcomas; six of these tumours showed a positively staining spindle cell component.

Epithelioid sarcomas

Five tumours were studied but only one showed positive staining.

Other sarcomas

Only one of the remaining 176 sarcomas showed positively staining cells. This was a malignant schwannoma; it did not contain glandular areas.

Cytokeratin staining of tumours 1321

Discussion

CAM 5.2 is a sensitive marker of most carcinomas with the possible exception of a small number of renal carcinomas. One of six renal carcinomas in this study failed to stain with CAM 5.2 and Herman et al. (1983) failed to demonstrate keratin in three of 13 renal carcinomas. The variability of staining of renal cell carcinomas is probably a reflection of their mesenchymal derivation. CAM 5.2 is also valuable in distinguishing between carcinomas and malignant melanomas.

The positive staining of pure spindle cell carcinomas of the lung and thyroid suggests that CAM 5.2 is a useful marker in these tumours. The staining patterns of carcinosarcomas are of interest. Since the spindle cell component of the carcinosar- comas in this study stained negatively for cytokeratin and positively in 4/5 cases for vimentin this suggests that these tumours are true carcinosarcomas and not carcin- omas with pseudosarcomatous differentiation. Our results are different from those obtained by Addis & Corrin (1985) who demonstrated cytokeratin in the epithelial component of all of eight carcinomas and also in the sarcomatous component of three of these tumours. They were unable to obtain consistent staining results with a vimentin antibody. These markers of course do not distinguish the sarcomatous component from a pseudosarcomatous stromal reaction but this distinction should be possible on the basis of cytological detail.

The staining results of other types of carcinomas with anti-keratin antibodies are conflicting and confusing (Schlegel er af. 1980, Osborn & Weber 1983). This is due to a number of factors, probably the most important being the source of the antibody used and which of the 19 different cytokeratins (Moll et al. 1982) it recognizes. Ramaekers et al. (1983a) found that a polyclonal cytokeratin antiserum stained a higher number of squamous cell carcinomas than a monoclonal cytokeratin antibody. Cytokeratin antiserum has been used to distinguish poorly differentiated nasopharyngeal carcin- omas (lymphoepitheliomas) from lymphomas (Miettinen, Lehto & Virtanen 1982a, Madri & Barwick 1982). Renal cell carcinomas show variable cytokeratin staining and some also contain vimentin (Herman et al. 1983). The positive staining of oat cell carcinomas (Sheppard & Bobrow 1985), mesothelial cells and mesotheliomas (Holden & Churg 1984, Heydermann et al , 1985) is well documented as is the negative staining of malignant melanomas (Ramaekers et al. 1982, 1983b, 1983~) .

The staining results of sarcomas in this study may seem at first glance to suggest that CAM 5.2 is not a specific marker of epithelial tumours. However, the positive staining of certain sarcomas may be interpreted as support for either an epithelial derivation or differentiation in these tumours. This is certainly likely in the case of synovial sarcomas where an epithelial histogenesis or differentiation is also supported by tissue culture (Alvarez-Fernandez & Escalona-Zapata 1981) and ultrastructural studies (Gabbiani et al. 1971). The earlier suggestion of a synovial derivation is unlikely as normal synovium does not contain cytokeratin (Corson et al. 1984). Some workers (Corson et al. 1984, Salisbury & Isaacson 1985), but not all (Miettinen et al. 1982b), also report cytokeratin staining in the spindle cell component of synovial sarcomas. In addition these tumours also stain with the oligosaccharide HMFG-2 (Salisbury & Isaacson 1985). It is interesting to note that in our study calcifying

1322 M . Leader et al.

synovial sarcomas showed a higher incidence of staining than non-calcifying tumours. It has previously been suggested that calcification occurs especially in tumours show- ing obvious epithelial differentiation (Mirra, Want & Bhuta 1984).

The histogenesis of epithelioid sarcomas is still unresolved but other workers (Chase el al. 1984) also report cytokeratin positivity in these tumours and a similarity to synovial sarcomas has been suggested (Mukai et al. 1985). The positive staining of a malignant schwannoma (this patient had a 20 year history of neurofibromatosis) may also provide evidence of a tendency to epithelial differentiation, a feature which has been recorded in 11 of these tumours (Lanford & Cohn 1927, Foraker 1948, Woodruff 1976, Hajdu 1979, Garre 1982, Warner & Louie 1983).

We therefore suggest that CAM 5.2 is a specific epithelial marker and that its positivity in certain ‘sarcomas’ may merely reflect an epithelial differentiation or support an epithelial histogenesis in these cases. A number of studies have examined the cytokeratin staining of sarcomas. Thirty-five sarcomas examined by Miettinen et al. (1982b), six by Schlegel et al. (1980), six by Ramaekers et al. (1983c), four by Gabbiani et al. (1981), six by Makin et al. (1984) and 52 by Corson et al. (1984) all stained negatively. The study by Corson and his colleagues included two schwan- nomas. However, Salisbury & Isaacson (1985), using a similar source of CAM 5.2 to that used in our study, reported a weak and diffuse cross-reaction type of staining in 8/20 leiomyosarcomas and 3/20 fibrosarcomas. We did not see this type of staining. Makin et al. (1984) report some pale cytoplasmic staining of normal smooth muscle, the only difference between our staining methods and theirs being the absence of trypsinization in this study (except for the staining of the spindle cell carcinomas in the latter part of the study). This may explain the difference between the results.

From the results of this study we conclude that CAM 5.2 is a sensitive marker of epithelial tumours and of those showing epithelial differentiation. It is also useful in confirming the epithelial nature of pure spindle cell carcinomas with the exception of a small number of renal carcinomas. When used in addition to vimentin it is valuable in the diagnosis of carcinosarcomas. The presence of CAM 5.2 staining in a small group of sarcomas, e.g. synovial sarcomas, probably reflects either an epithelial derivation or an epithelial differentiation in these tumours. Where the histological diagnosis is thought to be one of a sarcoma rather than a carcinoma, CAM 5.2 is of value in separating both biphasic and monophasic synovial sarcomas from other spindle cell sarcomas.

Acknowledgements

We are most grateful to Professor D.H.MacKenzie who very kindly allowed us full access to his collection of soft tissue sarcomas; without his diagnostic expertise this study’could not have been possible. We would also like to thank Dr S.Baithun and Dr J.Van der Walt who kindly lent us blocks of five spindle cell carcinomas. ML is in receipt of a grant from the North West Thames Area Health Authority and the Westminster Hospital Research Trust.

Cytokeratin staining of tumours 1323

References

ADDIS B.J. & CORRIN B. (1985) Pulmonary blastoma, carcinosarcoma and spindle cell carcinoma. An immunohistochemical study of keratin intermediate filaments. Journal of Pathology 147, 291-301

ALTMANNSBERGER M., OSBORN M., HOLSCHER A,, SCHAUER A. & WEBER K. (1981) The distribution of keratin type intermediate filaments in human breast cancer. An immunohistological study. Virchows Archive 37, 277-284

ALVAREZ-FERNANDEZ E. & ESCALONA-ZAPATA J . (1981) Monophasic mesenchymal synovial sarcoma: Its identification by tissue culture. Cancer 47, 628-635

CHASED.R., WEISSS.W.. ENZ~NGERF.M. & L A N G ~ 0 s s D . v . M . (1984) Keratininepithelioidsarcomas. American Journal of Surgical Pathology 8, 435-441

CORSON J.M., WEISS L.M., BANKS-SCHLEGEL S.P. & PINKUS G.S. (1984) Keratin proteins and car- cinoembryonic antigen in synovial sarcomas. An immunohistochemical study of 24 cases. Human Pathology 15,615-621

FORAKER A.G. (1948) Gland-like elements in peripheral neuro-sarcoma. Cancer 1, 286-293 GABBIANI G., KAYE G.I., LATTES R. & MAJNO G. (1971) Synovial sarcoma. Electron microscopic study of a

typical case. Cancer 28, 1031-1039 GABBIANI G., KAPANCI Y . , BARAZZONE P. & FRANKE W.W. (1981) Immunochemical identification of

intermediate-sized filaments in human neoplastic cells. American Journal of Pathology 104,206-216 GARRE C . (1982) Uber Sekundar maligne neurome. Beitrage Klin Chir 9, 465-495 HAJDU S.I. (1979) Tumours of peripheral nerves. In Pathology of Sofr Tissue Tumours, pp. 448-450. Lea &

Febiger, Philadelphia HERMAN C.J., MOESKER O., KANT A . , HUYSMAN A,, VOOIJS G.P. & RAMAEKERS F.G. (1983) Is renal cell

(Grawitz) tumour a carcinosarcoma? Virchows Archive (Cell Pathology) 44,77-83 HEYDERMANN E., LARKIN S.E., MAKIN C.A. & CORRIN B. (1985) Epithelial markers in pleural

mesotheliomas. A comparison with primary lung carcinoma. Abstract Pathological Society of Great Britain and Ireland. 150th meeting, January

HOLDEN J. & CHURG A. (1984) Immunohistochemical staining for keratin andcarcinoembryonicantigen in the diagnosis of malignant mesothelioma. American Journal of Surgical Pathology 8, 277-279

LANFORD J .A . & COHN I . ( I 927) Ependymal neoplasm of the median nerve with case report. Sourh African Medical Journal 20, 273-278

MACKENZIE D.H. (1970) The Dtfferential Diagnosis of Fibroblasric Disorders. Blackwell Scientific Publications, Oxford

MADN J.A. & BARWICK K.W. (1982) An immunohistochemical study of nasopharyngeal neoplasms using keratin antibodies. American Journal of Surgical Pafhology 6, 143-149

MAKIN C.A. , BOBROW L.G. & BODMER W.F. (1984) Monoclonal antibody tocytokeratin for use in routine histopathology. Journal of Clinical Pathology 37, 975-983

M I E ~ N E N M., LEHTO V.P. & VIRTANEN 1. (1982a) Nasopharyngeal lymphoepithelioma. Histological diagnosis as aided by immunohistochemical demonstration of keratin. Virchows Archive (Cell PUthOl-

MIEITINEN M., LEHTO V.P., BADLEY R.A. & VIRTANEN 1. (1982b) Expression of intermediate filaments in soft tissue sarcomas. International Journal of Cancer 30, 541-546

MIRRA J.M., WANTS. & BHUTA S. (1984) Synovial sarcoma with squamous differentiation of its mesenchy- ma1 glandular elements. Amercian Journal of Surgical Parhology 8, 791-796

MOLL R., FRANKE W.W., SCHILLER D.L., GEIGER B. & KREPLER R. (1982) The catalogue of human cytokeratins: patterns of expression in normal epithelia, tumours and cultured cells. Cell 31, 11-24

MUKAI M. , TORIKATA C., IRI H., HANOKA H., KAWAI T., YAKUMARU K., SHIMODA T.. MIKATA A. & KAGEYAMA K . (1985) Cellular differentiation of epithelioid sarcoma. American Journal of Patho/ogy 119, 44-56

OSBORN M. & WEBER K . (1983) Biology of disease. Tumour diagnosis by intermediate filament typing: A novel tool for surgical pathology. Laboratory Investigation 48, 372-394

RAMAEKERS F., PUTS J . , KANT A , , MOESKER O., JAP P. & VOOIK P. (1982) Differential diagnosis of human

ogy) 40, 163-169

1324 M . Leader et al.

carcinomas, sarcomas and their metastases using antibodies to intermediate-sized filaments. Euro- pean Journal of Cancer and Clinical Oncology 18, 1251-1257

RAMAEKERSF., HUYSMAN A., MOESKERO., KANT A., JAP P., HERMAN C. & VOOIJS P. (1983a) Monoclonal antibody to keratin filaments, specific for glandular epithelia and their tumours. Laboratory Investigation 49, 353-361

RAMAEKERS F.C.S., PUTS J.J.G., MOESKER O., KANT A., VOO~JS G.P. & JAP P.H. (1983b) Intermediate filaments in malignant melanomas. Journal of Clinical fnvesrigafion 71, 635-643

RAMAEKERS F.C., PUTS J.J.G., KANT A . , MOESKER O., JAP P.H. & VOOIJS G.P. (1983~) Antibodies to intermediate filament proteins in the imrnunohistochemical identification of human tumours: An overview. Histochemical Journal 15, 691-713

RUNGGER-BRANDLE E. & GABBIANO G. (1983) The role of cytoskeletal and cytocontractile elements in pathologic processes. American Journal of Pathology 110, 361-392

SALISBURY J.R. & ISAACSON P.G. (1985) Synovial sarcoma: An immunohistochemical study. Journal of Pathology 147, 49-57

SCHLEGEL R., BANKS-SCHLECEL S., MCLEOD J.A. & PINKUS G.S. (1980) Imrnunoperoxidase localization of keratin in human neoplasms. American Journal of Pathology 101, 41-48

SHEPPARD M.N. & BOBROW L.G. (1985) Irnmunostaining of human lung tumours using a panel of monoclonal antibodies. Abstract Pathological Society of Great Britain and Ireland. 150th meeting, January

WARNER T. & LOUIE R. (1983) Malignant nerve sheath turnour containing endocrine cells. American Journal of Surgical Pathology 7,583-590

WOODRUFF J.M. (1976) Peripheral nerve turnours showing glandular differentiation (glandular schwan- nomas). Cancer 37, 2399-2413