Immunohistological Staining of Reactive Mesothelium, Monoclonal

J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from J Clin Pathol 1987;40:19-25

Immunohistological staining of reactive mesothelium, mesothelioma, and lung carcinoma with a panel of monoclonal antibodies

ANNA K GHOSH, K C GATTER, M S DUNNILL, D Y MASON From the Nuffield Department of Pathology, John Radcliffe Hospital, Oxford

SUMMARY A panel of seven monoclonal antiepithelial antibodies of different specificities, including anticytokeratin, human milk fat globule membrane, Ca, and carcinoembryonic antigen (CEA) were used with the alkaline phosphatase-antialkaline phosphatase (APAAP) immunostaining technique to determine their value in the differentiation between benign and malignant mesothelial cells and lung carcinoma in histological preparations. The anticytokeratin antibody reacted strongly with all cases of reactive mesothelium, mesothelioma, and lung carcinoma. Antibodies to human milk fat globule membrane and the Ca antigen stained mesothelioma and carcinoma and 43% of cases of reactive mesothelium. Staining for carcinoembryonic antigen was not detected in reactive mesothelium or mesothelioma, but was present in most of the lung carcinomas. CEA seemed to be the single most useful marker in distinguishing carcinoma from mesothelioma in that a positive reaction for CEA would indicate carcinoma rather than mesothelioma. copyright.

The histological diagnosis ofmalignant mesothelioma membrane, Ca and carcinoembryonic antigen were in pleural biopsy specimens is a well recognised prob- evaluated to determine whether they could distinguish lem.1 Malignant mesothelioma may be indistinguish- between reactive and malignant mesothelial cells and http://jcp.bmj.com/ able from a reactive mesothelial cell proliferation, or lung carcinoma in histological preparations, using an from primary, or metastatic carcinoma. The same immunoalkaline phosphatase staining method. problem exists in the cytological diagnosis of serous fluids in which malignant mesothelioma cells may be Material and methods difficult to differentiate from benign mesothelial cells or from metastatic carcinoma by morphological SAM PLES criteria alone. All tissues were obtained from the surgical pathology Recent immunocytochemical studies of serous files of the histopathology department, John Radcliffe on September 28, 2021 by guest. Protected fluids have shown that monoclonal antibodies are use- Hospital, Oxford and had been fixed in 10% ful in distinguishing benign from malignant cells. unbuffered formalin and embedded in paraffin. Cases Human milk fat globule membrane antigen, carcino- had been classified according to conventional histo- embryonic antigen (CEA), and the Ca antigen have logical criteria. Samples studied comprised: seven been shown in a wide variety of malignant epithelial pleural mesotheliomas (most of which were referred cells but have rarely been seen in reactive mesothelial to and accepted by the Pneumoconiosis panel); 15 cells.2`4 A previous study on histological samples of primary lung tumours, which included four squamous mesothelium reported similar findings with two anti- cell carcinomas, four adenocarcinomas, five oat cell bodies against human milk fat globule membrane and two carcinoid tumours; and reactive mesothelial antigen and an anti-CEA antiserum.5 cells from seven cases of recurrent pneumothorax. In the present study a panel of seven monoclonal The lung tumours were selected as representative antiepithelial antibodies of different specificities, paraffin embedded specimens from a series of 54 lung including anticytokeratin, human milk fat globule tumours previously studied in cryostat sections.6 The samples of benign mesothelial cells were selected from Accepted for publication 9 July 1986 19 J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from 20 Ghosh, Gatter, Dunnill, Mason Table 1 Details ofmonoclonal antibodies usedfor cases of mesothelioma examined. Most of the tumour immunological analysis of mesothelioma, benign mesothelial cells were positive in each case, showing strong cyto- proliferation, and lung carcinoma plasmic staining (fig4a). The antiepithelial membrane antigen (anti-EMA) Antibody Specification Reference antibodies E29 and HMFG2 showed essentially iden- In four cases the staining was KL I Cytokeratin Viac et al! tical reaction patterns. E29 Human milk fat globule Cordell et al8 patchy and weak on solid lumps of tumour cells (fig HMFG2 membrane antigen Taylor-Papadimitriou eta!9 4b), but where clefts lined by tumour cells were seen J (EMA) Gatter et al'0 strong luminal staining. In one Cal Cancer associated Ashall et al" these generally gave Ca2 glycoprotein Bramwell et al'2 case there was only very occasional staining of cells at Ca3 J the periphery of the tumour, and in the other two 11.285.14 Carcinoembryonic Gatter et al'° antigen (CEA) cases very weak staining oftumour cells was observed. The three Ca antibodies showed similar staining reactions to the anti-EMA antibodies, although Cal was generally weaker. Anti-CEA was negative on a large number of cases as clear cut examples of reac- these seven cases of mesothelioma. tive mesothelium. Cases in which any possible con- pulmonary tis- fusion might be made with underlying CARCINOMA sue were excluded. All four cases ofadenocarcinoma gave strong positive reactions with the anticytokeratin antibody (KLI) IMMUNOCYTOCHEMISTRY and the anti-EMA antibodies. Staining was usually Table 1 gives details of the monoclonal antibodies distributed evenly throughout the tumour. The three used in this study. Sections (5 gm) were stained by the Ca antibodies stained all four cases of adeno- APAAP immunoalkaline phosphatase staining carcinoma. The staining pattern varied from small method, as described previously.'3 Prior trypsinisa- foci of positive staining seen in one case to a uniform tion of sections was not performed. distribution of positive reaction. Anti-CEA stained

three of the four adenocarcinomas, one with a focal copyright. Results staining pattern. Squamous cell carcinomas showed strong cyto- REACTIVE MESOTHELIUM plasmic staining with the anticytokeratin antibody, Monoclonal antibody KLI (anticytokeratin) reacted with a patchy distribution in one case. One anti-EMA with mesothelium in all seven benign cases. It charac- antibody (E29) showed positive staining of tumour teristically stained the single layer of cuboidal meso- cells in two cases while the other anti-EMA antibody thelial cells lining the pleural surface with a strong (HMFG2) stained all four cases, although in two http://jcp.bmj.com/ cytoplasmic and surface reaction (figs la, 2a). Areas cases only occasional tumour cells were positive. The that contained clumps or multilayers of plump reac- Ca antibodies showed focal staining of two squamous tive mesothelial cells, either on the pleural surface or cell carcinomas. The other two were negative with Cal exfoliated, also reacted strongly with KLI and Ca2, while one showed focal staining with Ca3. Antibodies E29 and HMFG2 (antiepithelial mem- Anti-CEA stained three of the squamous cell car- brane antigen) and the three Ca antibodies gave sim- cinoma cases in a patchy staining pattern. ilar staining reactions to each other. Three of the Oat cell carcinomas showed positive staining with seven benign cases showed positive staining with these the anticytokeratin and anti-EMA antibodies in four on September 28, 2021 by guest. Protected antibodies. In these cases the single cuboidal layer of of five cases. The Ca antibodies were different in their mesothelial cells showed both cytoplasmic and surface staining reactions. Cal stained four cases of oat cell staining (figs lb, c). In areas in which there were multi- carcinoma, while Ca2 and Ca3 stained only two cases. layers of reactive mesothelial cells these were usually All showed a focal or patchy staining pattern. Anti- negatiye (fig 2b), although the occasional exfoliated CEA stained three cases of oat cell carcinoma; in two mesothelial cell was positive with these five antibodies. of these cases the staining was weak with a patchy Two cases had clefts lined by reactive mesothelial cells distribution. that were strongly positive (figs 3a, b). Staining for Carcinoid carcinomas were positive with the anti- HMFG2 was similar to that seen for E29 and Cal in cytokeratin antibody and negative with anti-EMA, all cases of benign mesothelium. Ca, and anti-CEA antibodies. Anti-CEA did not stain any reactive mesothelial cells. Discussion MESOTHELIOMA Antibody KLI (anticytokeratin) stained all seven The aim of this study was to evaluate the reactivity of J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from Immunohistological staining of reactive mesothelium, mesothelioma, and lung carcinoma 21

* a

.t ACO* .~~~~~~~~~~~~~ .I.. - - -- %,114k---0-' .1 .. 1 i 4 _' b '1 * 't 1 ob

..i...... 4'_ p

S.p

X..' S. - .. *,, K

- 4.,

.trj '1 -a . I'4 r ,4. .i. 0 . .

1 Fig la Fig lb Fig Ic

SO pR copyright. aW 4*. S... P , -z WM

Aw." Syt~' 40 ^ W ., p~ C-,

$P S. http://jcp.bmj.com/ Ry% _ .4 *+ ..tr V- y. F*a 4p4 441 r t w; Ai _s ,' _ A.; "X 1% + A. ' 4 -,r :k? 4 0S N ..... ti . *1 t 4k

41- on September 28, 2021 by guest. Protected a* 3Y.. z., .....1 ?.. .. Fig 2a Fig 2b

Fig 1 Section ofpleura stainedfor cytokeratin showing strong staining ofplump cuboidal reactive mesothelial cells (open arrows) and scattered exfoliated cells (bottom ofpicture). Flatter overlying mesothelium (closed arrows) is more weakly stained.

Fig 1 b Comparable section from same pleura specimen stainedfor epithelial membrane antigen (antibody E29). Line of plump mesothelial cells and cells exfoliatingfrom this layer are negative (compare with those in fig la), but overlying flattened mesothelium (arrowed) is strongly stained.

Fig ic High power view of area outlined in fig lb.

Fig 2 Same biopsy specimen as seen in fig 1. High power view of reactions for cytokeratin (a) and epifhelial membrane antigen (b) ofplump reactive mesothelial cells seen in figs Ja andb. J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from 22 Ghosh, Gatter, Dunnill, Mason

Fig 3a Fig 3b

4. I 'ft a, r

.. .

t'-i .t .*Mil S 1 C A 4 "¼ t . W'

0 copyright.

}

tf. http://jcp.bmj.com/

,~rs- es{ v r Fig 4a Fig 4b Fig 3 Clefts in pleural tissue lined with mesothelial cells, which are strongly reactive for (a) epithelial membrane antigen (antibody E29) and (b) Ca antigen (antibody Cal).

Fig 4 Mesothelioma showing strong labelling for (a) cytokeratin and (b) much more restricted and patchy staining for epithelial membrane antigen (antibody E29). on September 28, 2021 by guest. Protected several different monoclonal antiepithelial antibodies effusions. Previous studies on the diagnosis of these on histological preparations of mesothelioma, benign conditions in solid tumours using antikeratin anti- mesothelial proliferation, and carcinoma and to assess bodies have differed in their results. Corson and their value in the differential diagnosis of these condi- Pinkus,'5 using a polyclonal antikeratin antibody, tions. found that mesotheliomas were strongly positive but Positive staining with the anticytokeratin. antibody that adenocarcinomas were only weakly positive or KLl was observed in all cases of benign and malig- negative. They suggested that this would be a useful nant mesothelium and carcinoma. These findings are marker to distinguish these two neoplasms. Holden consistent with those of a previous study on exfoliated and Churg`6 and Loosli and Hurlimann,'7 however, cells in serous fluids using another monoclonal anti- found keratins expressed in both adenocarcinomas cytokeratin antibody of similar specificity.2 Kahn et and mesotheliomas. al' also showed that keratin was present in benign Other workers have shown varying patterns of and malignant mesothelial cells and carcinoma cells in keratin expression in bronchial carcinomas.8 19 All of J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from Immunohistological staining of reactive mesothelium, mesothelioma, and lung carcinoma 23 Table 2 Immunohistological results ofpanel ofmonoclonal antibodies with benign mesothelium and mesothelioma

No of cases KLI E29 HMFG2 Cal Ca2 Ca3 CEA Benign mesotheium: 4 + ------Benign mesothelium: 3 + + + + + + Mesothelioma: 7 + +/- +/- +/- +/- +1- - + = strong staining; + /- = weak or focal staining;- = no staining. these studies used polyclonal antikeratin antibodies. of seven cases of reactive mesothelium, however, were As keratins are a group of proteins ofdifferent molec- strongly positive with these antibodies. Most carcino- ular weights the antisera used in these studies almost mas were positive with both antibodies, except two certainly had different specificities. Gatter et al6 have carcinoid tumours and one oat cell carcinoma. recently shown heterogeneous expression of keratins Our previous experience with HMFG2 and E29 on in lung tumours using different monoclonal anti- cytological specimens showed strong staining of keratin antibodies. The detection of cytokeratins in exfoliated mesothelioma cells. Reactions with mesothelioma, benign mesothelium, and carcinomas exfoliated reactive mesothelial cells were seen in 14 of limits the use of these reagents in their differential 22 benign cases stained with HMFG2, and occa- diagnosis, particularly in routinely fixed material sionally, with E29, although not all mesothelial cells in where the number of suitable anticytokeratin anti- any one sample were positive.2 8 Epenetos et al,4 on bodies is so limited. the other hand, using wet fixed cytological material, Immunohistological studies with polyclonal anti- did not detect HMFG2 on benign mesothelial cells. sera against keratins of different molecular weights Although these differences could be due to different have shown that mesotheliomas and benign meso- techniques and fixatives used, a study comparing thelium express a 63 kd keratin not detected in most different staining and fixation methods on the same lung carcinomas.20 Such a difference in staining for specimens showed that the HMFG2 determinant was copyright. this 63 kd keratin between mesothelioma and ade- occasionally expressed on reactive mesothelial cells.2' nocarcinoma may have diagnostical application, and These previous reports indicated that mesothelial further work, particularly with specific monoclonal cells could express the HMFG2 determinant. The antikeratin antibodies, should be undertaken to clar- present study has shown the presence of HMFG2 in ify these results. reactive mesothelium in three of seven cases. This is in Both monoclonal antibodies (HMFG2 and E29) to contrast to the findings of Marshall et al,S who the human milk fat globule membrane antigen (anti- observed focal staining with HMFG2 in only one of http://jcp.bmj.com/ EMA) reacted with all cases of mesothelioma. Three 13 cases of reactive mesothelium. These workers used

Table 3 Immunohistological results ofpanel ofmonoclonal antibodies with lung carcinoma

Epithelial membrane antigen Ca antigen Case No Cytokeratin E29 HMFG2 Cal Ca2 Ca3 CEA on September 28, 2021 by guest. Protected Squamous cell I + +/- + +1- ±1- +/- +I_ carcinoma 2 + - +/- 3 + - +1- _ _ +1- +1- 4 + + + +1- +1- +1- +1- Adenocarcinoma 5 + + + + + + + 6 + + + + + + 7 + + + +/- +/- +/- +/- 8 + + + + +1- + + Oat cell carcinoma 9 + + + +/- - - +1- 10 - - - _ _ _ 11 + + + +/- +/- NT +/- 12 + + + +1- +1- + + 13 + + + + - +/- Carcinoid 14 + 15 + - - _ _ _ + = strong staining; +/- = weak or focal staining;- = no staining; NT = not tested. J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from 24 Ghosh, Gatter, Dunnill, Mason an immunoperoxidase technique and trypsinised their using polyclonal CEA antisera on reactive meso- tissue. Marshall et al5 detected the HMFG2 deter- thelium in which no staining was observed.5 15 24 Our minant in mesothelioma and lung carcinoma and sug- results are in agreement with these findings. gested that positive staining with this antibody indi- The reactions of the monoclonal CEA in this study cates malignancy, while negative staining is more with lung carcinomas were similar to those observed likely to be benign rather than malignant meso- by Gatter et al6 using the same reagent: most of these thelium. Our findings, however, differ from this and tumours showed positive staining. Other workers indicate that a positive reaction cannot be used to have detected CEA in lung tumours, particularly in distinguish a benign from malignant condition. bronchial adenocarcinomas.5 16 17 24 The three monoclonal antibodies directed against The difference in CEA staining in mesotheliomas different determinants on the Ca antigen gave similar and carcinomas makes it a useful marker for staining reactions and were positive with three of differentiating these two neoplasms. A positive reac- seven cases of reactive mesothelium, all the meso- tion with CEA suggests carcinoma and makes a diag- theliomas, and most of the carcinomas. Positive reac- nosis of mesothelioma unlikely. A distinction cannot tions ofthese three antibodies with mesothelioma cells be made (at least with the antibody used in this study), in cytological preparations have been noted. (AK however, between mesothelioma and reactive meso- Ghosh, unpublished observations).2 3 12 The reac- thelium, or between the different histological types of tions of the antibodies on benign mesothelial cells in lung carcinoma. cytological preparations have differed. Cal was At present there are no reagents that can conclu- weakly expressed on mesothelial cells in two of 47 sively distinguish benign from malignant meso- benign effusions (combined data from two series).23 thelium. Although monoclonal antibodies to milk fat Ca2 was not detected on reactive mesothelial cells, globule and Ca determinants can help to make this while Ca3 was detected in benign mesothelial cells in distinction in cytological preparations, our findings in eight of 23 cases.12 These latter findings suggest that histological preparations are less helpful. The location the antigenic determinant recognised by at least two of positive staining or intensity does not help to make ofthe Ca antibodies is variably present on mesothelial this distinction, as cases of benign mesothelium cells. exhibited strong staining, especially in cleft like areas A previous immunohistological study of pleural of cells. This was a feature also observed in meso- copyright. biopsy specimens detected Cal in four of eight cases thelioma. More specific markers for mesothelioma of benign mesothelium, seven of 12 cases of meso- will therefore be required to facilitate their histologi- thelioma, and three of five cases of metastatic car- cal diagnosis. cinoma. 22 Our findings are consistent with these observations and iidicate that these antibodies do not This work was supported by grants from the Cancer distinguish between malignant and benign meso- Research Campaign and the Wellcome Trust. We are thelium in histological sections. grateful to the research workers listed in table 1 for http://jcp.bmj.com/ CEA was absent in all cases of mesothelioma and generously providing antibodies for use in this study. reactive mesothelium, but was present in most cases of KCG is a Wellcome senior research fellow and holds carcinoma, both here and in a previous study of 54 the Gillson scholarship ofthe Society of Apothecaries lung tumours, which included cryostat sections from of London. the present cases.6 These findings agree with our observations on cytological samples where we found References on September 28, 2021 by guest. Protected that CEA was absent from benign mesothelial cells, 1 McCaughey WTE. Criteria for diagnosis of diffuse mesothelial weakly expressed in one case of mesothelioma (sub- tumours. Ann NY Acad Sci 1985;132:603-13. sequent cases of mesothelioma have been negative, 2 Ghosh AK, Spriggs AI, Taylor-Papadimitriou J, Mason DY. AK Ghosh, unpublished observations), and present in Immunocytochemical staining ofcells in pleural and peritoneal effusions with a panel of monoclonal antibodies. J Clin Pathol 80% of carcinomas.2 1983;36: 1154-64. Previous immunohistological studies of CEA in 3 Woods JC, Spriggs AI, Harris H, McGee JO'D. A new marker for mesothelioma have given conflicting results, with human cancer cells. 3. Immunocytochemical detection of some showing no staining5 23 24 and others positive malignant cells in serous fluids with the Cal antibody. Lancet 16 1982;ii:51 2-5. staining.'5 Different staining techniques, the use of 4 Epenetos AA, Canti G, Taylor-Papadimitriou J, Curling M, trypsin, and differing polyclonal antisera could Bodmer WF. Use of two epithelium-specific monoclonal account for these differences. Whether or not the antibodies for diagnosis of malignancy in serous effusions. monoclonal or polyclonal antibodies react with non- Lancet 1982;ii: 1004-6. 5 Marshall RJ, Herbet A, Braye SG, Jones DB. Use of antibodies specific crossreacting antigen may also be important to carcinoembryonic antigen and human milk fat globule to (the monoclonal anti-CEA used here does not react). distinguish carcinoma, mesothelioma and reactive meso- There have only been three histological studies thelium. J Clin Pathol 1984;37:1215-21. J Clin Pathol: first published as 10.1136/jcp.40.1.19 on 1 January 1987. Downloaded from Immunohistological staining of reactive mesothelium, mesothelioma, and lung carcinoma 25 6 Gatter KC, Dunnill MS, Pulford KAF, Heryet A, Mason DY. 16 Holden J, Churg A. Immunohistochemical staining for keratin Human lung tumours: a correlation of antigenic profile with and carcinoembryonic antigen in the diagnosis of malignant histological type. Histopathology 1985;9:805-23. mesothelioma. Am J Surg Pathol 1984;8:277-9. 7 Viac J, Reano A, Brochier J, Staquet MJ, Thivolet J. Reactivity 17 Loosli H, Hurlimann J. Immunohistological study of malignant pattern of a monoclonal antikeratin antibody (KLI). J Invest diffuse mesotheliomas of the pleura. Histopathology Dermatol 1983;81:351-4. 1984;8:793-803. 8 Cordell J, Richardson TC, Pulford KAF, et al. Production of 18 Gusterson B, Mitchell D, Warburton M, Sloane J. Immuno- monoclonal antibodies against human epithelial membrane histochemical localisation of keratin in human lung tumours. antigen for use in diagnostic immunocytochemistry. Br J Virchows Arch (Pathol Anal) 1982;394:269-77. Cancer 1985;52:347-54. 19 Said JW, Nash G, Tepper G, Banks-Schlegel S. Keratin proteins 9 Taylor-Papadimitriou J, Peterson JA, Arklie J, Burchell J, Ceriani and carcinoembryonic antigen in lung carcinoma: an immuno- RL, Bodmer WF. Monoclonal antibodies to epithelium- peroxidase study of fifty-four cases, with ultrastructural specific components of the human milk fat globule membrane: correlations. Hum Pathol 1983;14:70-6. production and reaction with cells in culture. Int J Cancer 20 Said JW, Nash G, Banks-Schlegel S, Sasson AF, Murakami S, 1981;28:17-21. Shintaku IP. Keratin in human lung tumors. Patterns of local- 10 Gatter KC, Abdulaziz Z, Beverley P, et al. Use of monoclonal ization of different-molecular-weight keratin proteins. Am J antibodies for the histopathological diagnosis of human malig- Pathol 1983;113:27-32. nancy. J Clin Pathol 1982;35:1253-67. 21 Ghosh AK, Spriggs AI, Curling M. Tumour markers in serous 11 Ashall F, Bramwell ME, Harris H. A new marker for human effusions. Lancet 1984;i:338. cancer cells. 1. The Ca antigen and the Cal antibody. Lancet 22 Burnett RA, Deery ARS, Adamson MR, Liddle C, Thomas M, 1982;ii:1-60. Roberts GH. Evaluation of Cal antibody in pleural biopsy 12 Bramwell ME, Ghosh AK, Smith WD, Wiseman G, Spriggs AI, material. Lancet 1983;i: 158. Harris H. Ca2 and Ca3: new monoclonal antibodies evaluated 23 Wang N, Huang S, Gold P. Absence of carcinoembryonic as tumor markers in serous effusions. Cancer 1985;56:105-10. antigen-like material in mesothelioma. An immuno- 13 Cordell JL, Falini B, Erber WN, et al. Immunoenzymatic label- histochemical differentiation from other lung cancers. Cancer ling of monoclonal antibodies using immune complexes of 1979;44:937-43. alkaline phosphatase and monoclonal anti-alkaline phos- 24 Kwee WS, Veldhuizen RW, Golding RP, et al. Histological dis- phatase (APAAP complexes). J Histochem Cytochem tinction between malignant mesothelioma, benign pleural 1984;32:219-29. lesion and carcinoma metastasis. Evaluation of the application 14 Kahn HJ, Hanna W, Yeger H, Baumal R. Immunohistochemical of morphometry combined with histochemistry and immuno- localization of prekeratin filaments in benign and malignant staining. Virchows Arch (Pathol Anat) 1982;397:287-99. cells in effusions. Comparison with intermediate filament distribution by electron microscopy. Am J Pathol 1982;109:206-14. 15 Corson JM, Pinkus G. Mesothelioma: of keratin proteins Requests for reprints to: Dr AK Ghosh, Immunology profile copyright. and carcinoenibryonic antigen. An immunoperoxidase study of Laboratory, Paterson Laboratories, Christie Hospital and 20 cases and comparison with pulmonary adenocarcinomas. Holt Radium Institute, Wilmslow Road, Manchester Am J Pathol 1982;108:80-8. M20 9BX, England. http://jcp.bmj.com/ on September 28, 2021 by guest. Protected