JOURNAL OF PATHOLOGY, VOL.
165: 137-143 (1991)
IMMUNOCYTOCHEMISTRY OF MALIGNANT MESOTHELIOMA: OV632 AS A MARKER OF MALIGNANT MESOTHELIOMA M. DELAHAYE, H.
c. HOOGSTEDEN* AND TH. H. VAN DER KWAST
Departments of Pathology and Pulmonology*, University Hospital Dijkzigt and Erasmus University, Rotterdam, The Netherlands Received 2 January 1991 Accepted 18 April 1991
SUMMARY In pleural or ascitic effusions the cytomorphological distinction of adenocarcinoma cells, reactive mesothelial cells, and malignant mesothelioma cells often causes a diagnostic dilemma. The value of immunocytochemistry was investigated on cytological smears of 24 well-established cases of malignant mesothelioma, a selected series of 31 metastatic adenocarcinomas, and 20 smears of patients without known malignancy. In these smears we scored the immunoreactivity with apanel of four monoclonal antibodies. In addition to antibodies for epithelial membrane antigen (EMA) and carcinoembryonic antigen (CEA), the monoclonal antibody MOC3 1 and the ovarian carcinoma specific antibody OV632 were incorporated in the panel. With none of these four antibodies was immunostaining of reactive mesothelial cells found. CEA- and MOC3 1 -positive tumour cells were frequent in metastatic adenocarcinomas, but occurred rarely in malignant mesotheliomas. EMA-positive tumour cells were found in all metastatic adenocarcinomas (100 per cent) and in most malignant mesotheliomas (83 per cent). In addition to the expected reactivity of OV632 with ovarian carcinomas, 22 of 24 malignant mesotheliomas contained immunopositive tumour cells, while only a small proportion of non-ovarian adenocarcinomas reacted with this antibody. This selective staining of malignant mesothelioma cells, but not reactive mesothelial cells, with OV632 now permits the positive identification of malignant mesothelioma cells in male patients. KEY
worn-Clinical cytology, serous effusions, malignant mesothelioma, immunocytochemistry
mesothelioma, but a positive confirmation of this diagnosis can generally not be achieved by these Morphological criteria for the cytological diag- techniques.'-' I Patients with a suspected diagnosis nosis of malignant mesothelioma in pleural, pericar- of malignant mesothelioma often have to undergo dial, or ascitic fluids are now well established.14 thoracoscopy for a guided biopsy to obtain a definite Reappraisal of some light microscopical details has diagnosis based on histopathological criteria. led to an even higher level of diagnostic competence By means of time-consuming, expensive, and not in the examination of these fluids.>' Nevertheless, always accessible techniques such as transmission the cytomorphological criteria still lack sufficient electron microscopy (TEM) features of mesothelial specificity and sensitivity to allow a reliable distinc- differentiation can be identified but the distinction tion in all cases of malignant mesothelioma cells of malignant from reactive mesothelial cells appears from adenocarcinoma cells, on the one hand, and to be d iffi~ u lt.'.~ It ,'~ has also been claimed that morfrom reactive mesothelial cells, on the other hand.* phometry may help to distinguish malignant from The use of (immuno) cytochemistry may be helpful reactive mesothelial cells,13 but this technique in some cases in excluding a diagnosis of malignant appeared of no value in differentiating malignant Addressee for correspondence: M. Delahaye, Department of mesothelioma cells from adenocarcinoma ~ e 1 ls .l~ A few previous reports have described the use of Pathology, Ee977, Erasmus University Rotterdam, P.O. Box antisera specific for (tumour) cells with mesothelial 1738,3000DR Rotterdam, The Netherlands. INTRODUCTION
0022-341 719 1/10013747 $05.00 0 1991 by John Wiley & Sons, Ltd.
138
M. DELAHAYE ET AL.
differentiation, but no monoclonal antibody with a high level of specificity for malignant mesothelioma is currently In previous studies, the selective reactivity of monoclonal antibody OV632 with ovarian carcinomas has been r e p ~ r t e d . ’ The ~.’~ suggested common histogenesis of ovarian carcinomas and malignant mesotheliomas led us to evaluate the reactivity of OV632 with malignant mesothelioma cells, reactive mesothelial cells, and a number of adenocarcinomas in cytological smears of pleural and ascitic fluids. The reactivity pattern of antibodies to carcinoembryonic antigen (CEA), epithelial membrane antigen (EMA), and the epithelial marker MOC3 1 was simultaneously examined on these cytological smears. MATERIALS AND METHODS Patient material
The study comprised cytological smears of pleural or ascitic fluids from 24 patients with histologically and/or TEM confirmed malignant mesothelioma, 3 1 patients with metastatic adenocarcinoma with known primary origin, and 20 patients without malignancy and negative follow-up (15 male, 5 female). The metastatic adenocarcinomas included 8 ovarian carcinomas, 10 pulmonary adenocarcinomas, 6 adenocarcinomas of the digestive tract, and 7 breast carcinomas. The nature of the primary tumour was confirmed histologically. After centrifugation of the pleural and ascitic fluids for 10 min at 2000 rpm, smears were prepared from the sediment. Thoroughly air-dried smears were dehydrated in acetone for 1 min and stored at - 20°C until use. Preliminary experiments had indicated that storage at -20°C for periods of up to 3 years did not affect immunohistochemical staining results. Immunocytochemistry
Prior to immunocytochemistry, slides were removed from the freezer and brought to room temperature. The smears of each case were incubated with monoclonal antibodies OV632 (Sanbio, Uden, The Netherlands), PARLAM 4 (anti-CEA), MOC3 1 (Organon, Oss, The Netherlands), and antiEMA (Dakopatts, Glostrup, Denmark). OV632 was diluted 1:5 in PBS (pH 7.4) and the other antibodies were diluted 1:10 in PBS. OV632 could not be stored for prolonged times at 4”C, as titres fell dramatically after several months. Before incubating the slides
with anti-CEA, MOC31 and anti-EMA, fixation and blocking of endogenous peroxidase activity were achieved simultaneously by incubation in methanol containing 3 per cent hydrogen peroxide for 25 min at room temperature. For demonstration of OV632 no prior blocking of endogenous peroxidase activity was allowed as hydrogen peroxide destroyed antigenicity completely. Incubation with the primary antibody was performed at 37°C for 30 min followed by three rinses of 3 min in PBS and incubation with horseradish peroxidase conjugated rabbit anti-mouse immunoglobulin antibody (Dakopatts, Glostrup, Denmark) diluted 1:50 in PBS for 30 rnin at room temperature. After visualization with diaminobenzidine tetrachloride containing 0.03 per cent H,O, in PBS, nuclear counterstaining with Mayer’s haematoxylin was performed and slides were mounted with malino1 for microscopical examination. Incubations were done on series of eight to ten slides including known positive and negative control slides. RESULTS Immunocytochemistry of malignant mesothelioma
The immunocytochemical staining data on a series of 24 cases of malignant mesothelioma with OV632, anti-EMA, anti-CEA, and MOC31 are listed in Table I. In 20 cases, a characteristic membranous staining pattern of tumour cells was obtained by immunostaining with anti-EMA. Staining with OV632 was of moderate to strong intensity in 22/24 cases, while the frequency of immunopositive tumour cells varied greatly. Macrophages and recognizable benign mesothelial cells were always OV632-negative. Staining was membranous and around the periphery of cell clusters and morulae (Fig. 1A). In a minority of cases, faint to moderate diffuse cytoplasmic reactivity was noted. In addition to their reactivity with OV632 and ant-EMA, a positive reaction was also obtained with MOC3 1 in one case (M 12), and in another case (M3), both MOC3 1 and, surprisingly, anti-CEA produced intense cytoplasmic immunostaining of most of the tumour cells (Fig. 1B). Metastatic adenocarcinoma
The immunocytochemical staining data on a series of 31 metastatic adenocarcinomas are listed in Table 11. All cases showed moderate to intense
IMMUNOHISTOCHEMISTRY OF MESOTHELIOMA
Table I-Immunocytochemical mesotheliomas Patient
Sex
M1 M2 M3 M4 M5 M6 M7 M8 M9 MI0 M11 MI2 MI3 M14 M15 M16 M17 M18 M19 M20 M21 M22 M23 M24
m m m
m m m m m m m m m f m m m m m m
m m m m m
results in 24 malignant
pl/as* OV632 MOC31 CEA EMA
+ + + + + + + + + + + + + + + + + +
+ ++ + -
+ + + + -
-
+ + + + + + + + + + + + + + + + -
*PI =pleural effusion; as = ascitic fluid
staining with anti-EMA. Varying proportions of tumour cells of ovarian adenocarcinomas and 2 out of 10 pulmonary adenocarcinomas were reactive with OV632 (Fig. IC). None of the adenocarcinomas of the digestive tract or the breast contained OV632-positive cells. MOC3 1 exhibited a diffuse cytoplasmic staining reaction in 18 metastatic adenocarcinomas (58 per cent), while CEA-positive cells were found in 14 cases (45 per cent). In five pulmonary adenocarcinomas, tumour cells positive for CEA and MOC3 1 were found, while two cases showed immunoreactivity for CEA alone and two cases for MOC31 alone in addition to their reactivity with anti-EMA (Fig. ID).
Reactive mesothelial cells No positive immunostaining reaction for EMA, OV632, MOC3 1 or CEA was obtained on cytological negative smears of pleural ( n = 7 ) or ascitic (n = 13) fluids from patients without a known malignancy.
139
DISCUSSION The value of cytology for the diagnosis of malignant mesothelioma is still considered a controversial issue, although recent studies have suggested that implementation of ancillary techniques has led to improved To date, immunocytochemistry has contributed to the positive identification of adenocarcinoma cells in serous fluids by using markers for tumour-associated glycoproteins such as CEA. CEA has been widely applied in cytology owing to its ability to distinguish a substantial proportion of adenocarcinomas from reac1,14915*20 tive and malignant mesothelial Malignant mesotheliomas rarely show focal, occasionally granular, cytoplasmic reactivity for CEA. Some authors attributed this phenomenon to cross-reactivity with epitopes on non-CEA glycoproteins such as blood group antigens.8321 In our study, only 1 out of 24 malignant mesotheliomas showed immunoreactivity with anti-CEA, while 14 of the 3 1 metastatic adenocarcinomas were positive. Most of the metastatic adenocarcinomas of the digestive tract and lung in this study were CEApositive, but most breast and ovarian carcinoma cells lacked this marker. This finding is in agreement with previous Several other markers have been described which positively identify adenocarcinoma cells and lack reactivity with benign or malignant mesothelial cells, e.g., gastrointestinal carcinoma antigen, stagespecific embryonic antigen, Leu MI, surfactant apoproteins, Lewis blood group antigens, and Tn antigen.20’2k26 As single markers, none of these antibodies is very useful and an impractically large number need to be incorporated in a panel in order to obtain specificity. Depending on the epitope specificity of the antibody directed against human milk fat globules (HMFG) or epithelial membrane antigen (EMA), reactivity with benign mesothelial cells can be observed.” Although the antibody used in our studies generally does not react with reactive mesothelial cells, it must be emphasized that in both paraffin-embedded tissue sections of inflammatory pleura and in cytological smear preparations of reactive serous fluids we did occasionally observe sporadic EMA-positive mesothelial cells. Similarly, in a study using a polyclonal antiserum EMApositive cells with normal morphology were found in a few cases without known malignancy.28In contrast to the rare expression of EMA by reactive mesothelial cells, data from the literature indicate
Fig. I-(A) Small groups of mainly membranously OV632 stained cells of malignant mesothelioma. Reactive mesothelial cells in background (case M12). (B) Strong membranous staining of malignant mesothelioma cells with MOC31. Note some MOC31negative reactive mesothelial cells in the background (case M 12). (C) This case represents mainly solitary cells and some small groups of metastatic ovarian carcinoma cells. Most tumour cells show intense membranous and cytoplasmic staining with OV632 (case C4). (D) MOC3 1-positive tumour cells in a case of pulmonary adenocarcinoma (case C13)
141
IMMUNOHISTOCHEMISTRY OF MESOTHELIOMA
Table 11-Immunocytochemical
results in 3 1 metastatic adenocarcinomas
Patient
Sex
pl/as*
Primary tumour
c1 c2 c3 C4 C5 C6 c7 C8
f f f f f f f f
as as as as as as as
PI
Ovary Ovary Ovary Ovary Ovary Ovary Ovary Ovary
c9 C10 c11 c12 C13 C14 C15 C16 C17 C18
m m m m m m m m m
PI PI PI PI Pl PI PI PI PI PI
Lung Lung Lung Lung Lung Lung Lung Lung Lung Lung
C19 c20 c 21 c22 C23 C24
m m m m m m
as Pl PI as PI as
Digestive tract Digestive tract Digestive tract Digestive tract Digestive tract Digestive tract
C25 C26 C27 C28 C29 C30 C3 1
f f f f f f f
Pl PI PI PI PI as PI
Breast Breast Breast Breast Breast Breast Breast
m
OV632 MOC31 CEA EMA
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
*pl =pleural effusion; as = ascitic fluid.
that 50-95 per cent of malignant mesotheliomas express detectable amounts of EMA.8,1@’2,20 In keeping with these reports, we found that 20 of 24 cases of malignant mesotheliomas and all adenocarcinomas contained variable proportions of EMA-positive tumour cells. Some authors suggest a membranous staining pattern of malignant mesothelial cells with anti-EMA, whereas adenocarcinoma cells more often display cytoplasmic Although we could substantiate this claim, the practical use has to be established. Anti-EMA antibody thus appears to be a sensitive marker for epithelial and mesothelial malignancy, but additional markers are required for the definite identification of malignancy.
The newly described epithelial marker MOC3 1 is reported to react with squamous cell carcinomas, adenocarcinomas, and small cell carcinomas of the lung.30i31 As 2 of 24 malignant mesotheliomas and only 18 of 3 1 adenocarcinomas contained MOC3 1 positive tumour cells, this marker seems to be of limited use for the distinction of adenocarcinomas from malignant mesotheliomas. We did not demonstrate MOC31-positive cells in our small series of reactive effusions, which suggests that MOC3 1 expression by pleural or ascitic fluid cells may be used as a marker for malignancy. Obviously, a larger number of serous effusions need to be screened for MOC3 1-positive cells before a final conclusion can be drawn.
142
M. DELAHAYE ET AL.
Table 111-Sensitivity and specificity of immunocytochemistry for the diagnosis of malignant mesothelioma
Sensitivity Specificity Predictive value of a positive result Predictive value of a negative result
OV632 alone
All four markers*
92% (91%)t 80% (94%)
71% (70%) 94% (loo%o)
69% (9 1%)
85% (100%)
95% (94%)
87% (82%)
*Expected staining pattern is positive for OV632 and EMA combined, with negative staining for CEA and MOC31. ?In parentheses: only male patients.
Inclusion of OV632 in our panel proved to be most useful, since (1) this antibody is highly specific in its ability to distinguish malignant from reactive mesothelial cells and (2) the majority of malignant mesotheliomas and ovarian adenocarcinomas show immunoreactivity with OV632 in contrast to adenocarcinomas of other origin. The original publication on the immunohistochemical reactivity of OV632 on frozen tissue sections reported that benign, borderline, and malignant ovarian tumours were frequently stained with this antibody in contrast to adenocarcinomas of other origin." Remarkably, in this study none of 17 pulmonary adenocarcinomas was stained with OV632. Koelma et a l . 1 9 demonstrated that in 97 per cent of malignant effusions of patients with ovarian carcinomas OV632-positive tumour cells were found. Like us, these authors did not find OV632-positive cells in 20 reactive exudates. Thus, EMA + OV632 can now be applied to distinguish malignant from reactive mesothelial cells in cytological preparations. The use of OV632 alone in this group of selected patients gave high specificity for the diagnosis of malignant mesothelioma, particularly in male patients (Table 111). With a panel of four markers this specificity is increased. ACKNOWLEDGEMENTS
We wish to thank Mrs P. Delfos for photography, Mrs M. Hanegraaff for secretarial assistance, and Mr R. Teygeman for excellent immunostaining of the slides. REFERENCES 1. Suzuki Y. Pathology of human malignant mesothelioma. Semin Oncol
1980; 8: 268-282. 2. Whitaker D, Shilkin KB. Diagnosisofpleural malignantmesothelioma in life-a practical approach. J Pathol1984; 143: 147-175.
3. Tao LC. The cytopathology of mesothelioma. Acta Cytol 1979; 2 3 209-2 13. 4. Sherman ME, Mark EJ. Effusion cytology in the diagnosis of malignant epitheloid and biphasic pleural mesothelioma. Arch Pathol Lab Med 1990; 1 1 4 845-851. 5. Trio1JH, Conston AS, Chandler SV. Malignant mesothelioma. Cytopathology of 75 cases seen in a New Jersey Community Hospital. Acta Cyiol1984; 28: 3 7 4 5 . 6. Delahaye M, De Jong AAW, Versnel MA, Hoogsteden HC, Teeling P, Van der Kwast ThH. Cytopathology of malignant mesothelioma. Reappraisal of the diagnostic value of collagen cores. Cytopathology 1990; 1: 137-145. 7. Boon ME, Veldhuizen RW, Ruinaard C, Snieders MW, Kwee WS. Qualitative distinctive differences between the vacuoles of mesothelioma cells and of cells from metastatic carcinoma exfoliated in pleural fluid. Acta Cytol1984; 28: 4 4 3 4 4 9 . 8. Cibas ES, Corson JM, Pinkus GS. The distinction of adenocarcinoma from malignant mesothelioma in cell blocks of effusions: the role of routine mucin histochemistry and immunohistochemical assessment of carcinoembryonic antigen, keratin proteins, epithelial membrane antigen, and milk fat globule-derived antigen. Hum Pathol1987; 18: 67-74. 9. Dewar A, Valente M, Ring NP, Corrin B. Pleural mesothelioma of epithelial type and pulmonary adenocarcinoma: an ultrastructural and cytochemical comparison. J Pathol1987; 154: 309-316. 10. Kawai T, Greenberg SD, Truong LD, Mattioli CA, Klima M, Titus JL. Differences in lectin binding of malignant pleural mesothelioma and adenocarcinoma of the lung. Am J Puthol1988; 1 3 0 4 0 1 4 1 0 . 11. Tron V, Wright JL, Churg A. Carcinoembryonic antigen and milkfat globule protein staining of malignant mesothelioma and adenocarcinoma of the lung. Arch Pathol Lab Med 1987; 111: 291-293. 12. Van der Kwast ThH, Versnel MA, Delahaye M, De Jong AAW, Zondervan PE, Hoogsteden HC. Expression of epithelial membrane antigen on malignant mesothelioma cells. Actu Cyfol 1988; 3 2 169-1 74. 13. Kwee WS, Veldhuizen RW, Alons CA, Morawetz F, Boon ME. Quantitative and qualitativedifferencesbetween benign and malignant mesothelial cells in pleural fluid. Acta Cytol1982; 2 6 4 0 1 4 0 6 . 14. Kwee WS, Veldhuizen RW, Golding UP, et al. Histologic distinction between malignant mesothelioma, benign pleural lesion and carcinoma metastasis. Evaluation of the application of morphometry combined with histochemistry and immunostaining. Virchows Arch [Pathol Anat] 1982; 397: 287-299. 15. O H a r a CJ, Corson JM, Pinkus GS, Stahel RA. ME 1. A monoclonal antibody that distinguishes epithelial-type malignant mesothelioma and extrapulmonary malignancies. Am J Pathol1990; 136: 4 2 1 4 2 8 . 16. Singh G, Whiteside TL, Dekker A. Immunodiagnosis of mesothelioma. Use of a mesothelial cell serum in an indirect immunofluorescence assay. Cancer 1979; 43: 2288-2296. 17. Donna A, Betta PG, Bellingeri D, Marchesini A. New marker for mesothelioma: an immunoperoxidase study. J CIin Pathol 1986; 39: 961-968.
IMMUNOHISTOCHEMISTRY OF MESOTHELIOMA
18. Fleuren GJ, Coerkamp EG, Nap M, Van den Broek LJCM, Warnaar SO. Immunohistological characterization of a monoclonal antibody (OV632) against epithelial ovarian carcinomas. Virchows Arch [Pathol Anur] 1987; 410 481486. 19. Koelma IA, Nap M, Van Steenis GJ, Fleuren FJ. Tumor markers for ovarian cancer. A comparative immunohistochemical and immunocytochemical study of two commercial monoclonal antibodies (OV632 and OC125). Am JClin Pathol1988; 9 0 391-396. 20. Ordonez NG. The immunohistochemical diagnosis of mesothelioma. Differentiation of mesothelioma and lung carcinoma. Am J Surg Pathol1989; 13: 276291. 21. Stirling JW, Henderson DW, Spagnolo DV, Whitaker D. Unusual granular reactivity for carcinoembryonic antigen in malignant mesothelioma (Letter to the Editor). Hum Pathol 1990; 21: 678479. 22. Permanetter W, Wiesinger H. Immunohistochemical study of lysozyme, alpha,-anti-chymotrypsin, tissue polypeptide antigen, keratin and carcinoembryonic antigen in effusion sediments. Acta Cylol1987; 31: 1041 12. 23. Pinto MM, Bernstein L.H, Brogan DA, Criscuolo EM. Carcinoembryonic antigen in effusions. A diagnostic adjunct to cytology. Acta Cyrol1987;31: 113-118. 24. Ernst CS, Atkinson B, Chianese D, et al. Differential diagnosis between mesotheliomas and metastatic adenocarcinomas using monoclonal antibodies against gastrointestinal carcinoma antigen and stage-specificembryonic antigen. Appl Pathol1986; 4 115-124.
143
25. Sheibani K, Battifora H, Burke JS. Antigenic phenotypeofmalignant mesothelioma and pulmonary adenocarcinomas. An immunohistologic analysis demonstrating the value of LEU MI antigen. Am J
Pathol1986; 123: 212-219. 26. Noguchi M, Nakajima T, Hirohashi S, Akiba T, Shimosato Y. Immunohistochemical distinction of malignant mesothelioma from pulmonary adenocarcinoma with anti-surfactant apoprotein, anti-Lewis (a), and anti-Tn antibodies. Hum Pathol1990; 2 0 54-57. 27. Hilkens J, Buijs F, Hilgers J, ef al. Monoclonal antibodies against human milk-fat globule membranes detecting differentiation antigens of the mammary gland and its tumors. Int J Cancer 1984; 31: 1977206. 28. To A, Dearnaley DP, Ormerod MG, Canti G, Coleman DV. Epithelial membrane antigen. Its use in the cytodiagnosis of malignancy in serous effusions. Am JClin Pathol 1982; 7 8 214-219. 29. Leong ASY, Parkinson R, Milios J . ‘Thick’cellmembranes revealed by immunocytochemicalstaining: a c h e to thediagnosisofmesothelioma. Diagn Cytopathol1990; 6: 9-1 3. 30. Berendsen HH, De Leij L, Postmus E, el a/. Small cell lung cancer. Tumour cell phenotype detected by monoclonal antibodies and response to chemotherapy. Chest 1987; 91 Suppl. 1 lS-12S. 31. De Leij L, Postmus PE, Poppema S, Elema JD, The TH. The use of monoclonal antibodies for the pathological diagnosis of lung cancer. In: Hansen HH, ed. Lung Cancer: Basic and Clinical Aspects. Boston: Martinus Nijhoff, 1986; 3 1 4 8 .
165: 137-143 (1991)
IMMUNOCYTOCHEMISTRY OF MALIGNANT MESOTHELIOMA: OV632 AS A MARKER OF MALIGNANT MESOTHELIOMA M. DELAHAYE, H.
c. HOOGSTEDEN* AND TH. H. VAN DER KWAST
Departments of Pathology and Pulmonology*, University Hospital Dijkzigt and Erasmus University, Rotterdam, The Netherlands Received 2 January 1991 Accepted 18 April 1991
SUMMARY In pleural or ascitic effusions the cytomorphological distinction of adenocarcinoma cells, reactive mesothelial cells, and malignant mesothelioma cells often causes a diagnostic dilemma. The value of immunocytochemistry was investigated on cytological smears of 24 well-established cases of malignant mesothelioma, a selected series of 31 metastatic adenocarcinomas, and 20 smears of patients without known malignancy. In these smears we scored the immunoreactivity with apanel of four monoclonal antibodies. In addition to antibodies for epithelial membrane antigen (EMA) and carcinoembryonic antigen (CEA), the monoclonal antibody MOC3 1 and the ovarian carcinoma specific antibody OV632 were incorporated in the panel. With none of these four antibodies was immunostaining of reactive mesothelial cells found. CEA- and MOC3 1 -positive tumour cells were frequent in metastatic adenocarcinomas, but occurred rarely in malignant mesotheliomas. EMA-positive tumour cells were found in all metastatic adenocarcinomas (100 per cent) and in most malignant mesotheliomas (83 per cent). In addition to the expected reactivity of OV632 with ovarian carcinomas, 22 of 24 malignant mesotheliomas contained immunopositive tumour cells, while only a small proportion of non-ovarian adenocarcinomas reacted with this antibody. This selective staining of malignant mesothelioma cells, but not reactive mesothelial cells, with OV632 now permits the positive identification of malignant mesothelioma cells in male patients. KEY
worn-Clinical cytology, serous effusions, malignant mesothelioma, immunocytochemistry
mesothelioma, but a positive confirmation of this diagnosis can generally not be achieved by these Morphological criteria for the cytological diag- techniques.'-' I Patients with a suspected diagnosis nosis of malignant mesothelioma in pleural, pericar- of malignant mesothelioma often have to undergo dial, or ascitic fluids are now well established.14 thoracoscopy for a guided biopsy to obtain a definite Reappraisal of some light microscopical details has diagnosis based on histopathological criteria. led to an even higher level of diagnostic competence By means of time-consuming, expensive, and not in the examination of these fluids.>' Nevertheless, always accessible techniques such as transmission the cytomorphological criteria still lack sufficient electron microscopy (TEM) features of mesothelial specificity and sensitivity to allow a reliable distinc- differentiation can be identified but the distinction tion in all cases of malignant mesothelioma cells of malignant from reactive mesothelial cells appears from adenocarcinoma cells, on the one hand, and to be d iffi~ u lt.'.~ It ,'~ has also been claimed that morfrom reactive mesothelial cells, on the other hand.* phometry may help to distinguish malignant from The use of (immuno) cytochemistry may be helpful reactive mesothelial cells,13 but this technique in some cases in excluding a diagnosis of malignant appeared of no value in differentiating malignant Addressee for correspondence: M. Delahaye, Department of mesothelioma cells from adenocarcinoma ~ e 1 ls .l~ A few previous reports have described the use of Pathology, Ee977, Erasmus University Rotterdam, P.O. Box antisera specific for (tumour) cells with mesothelial 1738,3000DR Rotterdam, The Netherlands. INTRODUCTION
0022-341 719 1/10013747 $05.00 0 1991 by John Wiley & Sons, Ltd.
138
M. DELAHAYE ET AL.
differentiation, but no monoclonal antibody with a high level of specificity for malignant mesothelioma is currently In previous studies, the selective reactivity of monoclonal antibody OV632 with ovarian carcinomas has been r e p ~ r t e d . ’ The ~.’~ suggested common histogenesis of ovarian carcinomas and malignant mesotheliomas led us to evaluate the reactivity of OV632 with malignant mesothelioma cells, reactive mesothelial cells, and a number of adenocarcinomas in cytological smears of pleural and ascitic fluids. The reactivity pattern of antibodies to carcinoembryonic antigen (CEA), epithelial membrane antigen (EMA), and the epithelial marker MOC3 1 was simultaneously examined on these cytological smears. MATERIALS AND METHODS Patient material
The study comprised cytological smears of pleural or ascitic fluids from 24 patients with histologically and/or TEM confirmed malignant mesothelioma, 3 1 patients with metastatic adenocarcinoma with known primary origin, and 20 patients without malignancy and negative follow-up (15 male, 5 female). The metastatic adenocarcinomas included 8 ovarian carcinomas, 10 pulmonary adenocarcinomas, 6 adenocarcinomas of the digestive tract, and 7 breast carcinomas. The nature of the primary tumour was confirmed histologically. After centrifugation of the pleural and ascitic fluids for 10 min at 2000 rpm, smears were prepared from the sediment. Thoroughly air-dried smears were dehydrated in acetone for 1 min and stored at - 20°C until use. Preliminary experiments had indicated that storage at -20°C for periods of up to 3 years did not affect immunohistochemical staining results. Immunocytochemistry
Prior to immunocytochemistry, slides were removed from the freezer and brought to room temperature. The smears of each case were incubated with monoclonal antibodies OV632 (Sanbio, Uden, The Netherlands), PARLAM 4 (anti-CEA), MOC3 1 (Organon, Oss, The Netherlands), and antiEMA (Dakopatts, Glostrup, Denmark). OV632 was diluted 1:5 in PBS (pH 7.4) and the other antibodies were diluted 1:10 in PBS. OV632 could not be stored for prolonged times at 4”C, as titres fell dramatically after several months. Before incubating the slides
with anti-CEA, MOC31 and anti-EMA, fixation and blocking of endogenous peroxidase activity were achieved simultaneously by incubation in methanol containing 3 per cent hydrogen peroxide for 25 min at room temperature. For demonstration of OV632 no prior blocking of endogenous peroxidase activity was allowed as hydrogen peroxide destroyed antigenicity completely. Incubation with the primary antibody was performed at 37°C for 30 min followed by three rinses of 3 min in PBS and incubation with horseradish peroxidase conjugated rabbit anti-mouse immunoglobulin antibody (Dakopatts, Glostrup, Denmark) diluted 1:50 in PBS for 30 rnin at room temperature. After visualization with diaminobenzidine tetrachloride containing 0.03 per cent H,O, in PBS, nuclear counterstaining with Mayer’s haematoxylin was performed and slides were mounted with malino1 for microscopical examination. Incubations were done on series of eight to ten slides including known positive and negative control slides. RESULTS Immunocytochemistry of malignant mesothelioma
The immunocytochemical staining data on a series of 24 cases of malignant mesothelioma with OV632, anti-EMA, anti-CEA, and MOC31 are listed in Table I. In 20 cases, a characteristic membranous staining pattern of tumour cells was obtained by immunostaining with anti-EMA. Staining with OV632 was of moderate to strong intensity in 22/24 cases, while the frequency of immunopositive tumour cells varied greatly. Macrophages and recognizable benign mesothelial cells were always OV632-negative. Staining was membranous and around the periphery of cell clusters and morulae (Fig. 1A). In a minority of cases, faint to moderate diffuse cytoplasmic reactivity was noted. In addition to their reactivity with OV632 and ant-EMA, a positive reaction was also obtained with MOC3 1 in one case (M 12), and in another case (M3), both MOC3 1 and, surprisingly, anti-CEA produced intense cytoplasmic immunostaining of most of the tumour cells (Fig. 1B). Metastatic adenocarcinoma
The immunocytochemical staining data on a series of 31 metastatic adenocarcinomas are listed in Table 11. All cases showed moderate to intense
IMMUNOHISTOCHEMISTRY OF MESOTHELIOMA
Table I-Immunocytochemical mesotheliomas Patient
Sex
M1 M2 M3 M4 M5 M6 M7 M8 M9 MI0 M11 MI2 MI3 M14 M15 M16 M17 M18 M19 M20 M21 M22 M23 M24
m m m
m m m m m m m m m f m m m m m m
m m m m m
results in 24 malignant
pl/as* OV632 MOC31 CEA EMA
+ + + + + + + + + + + + + + + + + +
+ ++ + -
+ + + + -
-
+ + + + + + + + + + + + + + + + -
*PI =pleural effusion; as = ascitic fluid
staining with anti-EMA. Varying proportions of tumour cells of ovarian adenocarcinomas and 2 out of 10 pulmonary adenocarcinomas were reactive with OV632 (Fig. IC). None of the adenocarcinomas of the digestive tract or the breast contained OV632-positive cells. MOC3 1 exhibited a diffuse cytoplasmic staining reaction in 18 metastatic adenocarcinomas (58 per cent), while CEA-positive cells were found in 14 cases (45 per cent). In five pulmonary adenocarcinomas, tumour cells positive for CEA and MOC3 1 were found, while two cases showed immunoreactivity for CEA alone and two cases for MOC31 alone in addition to their reactivity with anti-EMA (Fig. ID).
Reactive mesothelial cells No positive immunostaining reaction for EMA, OV632, MOC3 1 or CEA was obtained on cytological negative smears of pleural ( n = 7 ) or ascitic (n = 13) fluids from patients without a known malignancy.
139
DISCUSSION The value of cytology for the diagnosis of malignant mesothelioma is still considered a controversial issue, although recent studies have suggested that implementation of ancillary techniques has led to improved To date, immunocytochemistry has contributed to the positive identification of adenocarcinoma cells in serous fluids by using markers for tumour-associated glycoproteins such as CEA. CEA has been widely applied in cytology owing to its ability to distinguish a substantial proportion of adenocarcinomas from reac1,14915*20 tive and malignant mesothelial Malignant mesotheliomas rarely show focal, occasionally granular, cytoplasmic reactivity for CEA. Some authors attributed this phenomenon to cross-reactivity with epitopes on non-CEA glycoproteins such as blood group antigens.8321 In our study, only 1 out of 24 malignant mesotheliomas showed immunoreactivity with anti-CEA, while 14 of the 3 1 metastatic adenocarcinomas were positive. Most of the metastatic adenocarcinomas of the digestive tract and lung in this study were CEApositive, but most breast and ovarian carcinoma cells lacked this marker. This finding is in agreement with previous Several other markers have been described which positively identify adenocarcinoma cells and lack reactivity with benign or malignant mesothelial cells, e.g., gastrointestinal carcinoma antigen, stagespecific embryonic antigen, Leu MI, surfactant apoproteins, Lewis blood group antigens, and Tn antigen.20’2k26 As single markers, none of these antibodies is very useful and an impractically large number need to be incorporated in a panel in order to obtain specificity. Depending on the epitope specificity of the antibody directed against human milk fat globules (HMFG) or epithelial membrane antigen (EMA), reactivity with benign mesothelial cells can be observed.” Although the antibody used in our studies generally does not react with reactive mesothelial cells, it must be emphasized that in both paraffin-embedded tissue sections of inflammatory pleura and in cytological smear preparations of reactive serous fluids we did occasionally observe sporadic EMA-positive mesothelial cells. Similarly, in a study using a polyclonal antiserum EMApositive cells with normal morphology were found in a few cases without known malignancy.28In contrast to the rare expression of EMA by reactive mesothelial cells, data from the literature indicate
Fig. I-(A) Small groups of mainly membranously OV632 stained cells of malignant mesothelioma. Reactive mesothelial cells in background (case M12). (B) Strong membranous staining of malignant mesothelioma cells with MOC31. Note some MOC31negative reactive mesothelial cells in the background (case M 12). (C) This case represents mainly solitary cells and some small groups of metastatic ovarian carcinoma cells. Most tumour cells show intense membranous and cytoplasmic staining with OV632 (case C4). (D) MOC3 1-positive tumour cells in a case of pulmonary adenocarcinoma (case C13)
141
IMMUNOHISTOCHEMISTRY OF MESOTHELIOMA
Table 11-Immunocytochemical
results in 3 1 metastatic adenocarcinomas
Patient
Sex
pl/as*
Primary tumour
c1 c2 c3 C4 C5 C6 c7 C8
f f f f f f f f
as as as as as as as
PI
Ovary Ovary Ovary Ovary Ovary Ovary Ovary Ovary
c9 C10 c11 c12 C13 C14 C15 C16 C17 C18
m m m m m m m m m
PI PI PI PI Pl PI PI PI PI PI
Lung Lung Lung Lung Lung Lung Lung Lung Lung Lung
C19 c20 c 21 c22 C23 C24
m m m m m m
as Pl PI as PI as
Digestive tract Digestive tract Digestive tract Digestive tract Digestive tract Digestive tract
C25 C26 C27 C28 C29 C30 C3 1
f f f f f f f
Pl PI PI PI PI as PI
Breast Breast Breast Breast Breast Breast Breast
m
OV632 MOC31 CEA EMA
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
*pl =pleural effusion; as = ascitic fluid.
that 50-95 per cent of malignant mesotheliomas express detectable amounts of EMA.8,1@’2,20 In keeping with these reports, we found that 20 of 24 cases of malignant mesotheliomas and all adenocarcinomas contained variable proportions of EMA-positive tumour cells. Some authors suggest a membranous staining pattern of malignant mesothelial cells with anti-EMA, whereas adenocarcinoma cells more often display cytoplasmic Although we could substantiate this claim, the practical use has to be established. Anti-EMA antibody thus appears to be a sensitive marker for epithelial and mesothelial malignancy, but additional markers are required for the definite identification of malignancy.
The newly described epithelial marker MOC3 1 is reported to react with squamous cell carcinomas, adenocarcinomas, and small cell carcinomas of the lung.30i31 As 2 of 24 malignant mesotheliomas and only 18 of 3 1 adenocarcinomas contained MOC3 1 positive tumour cells, this marker seems to be of limited use for the distinction of adenocarcinomas from malignant mesotheliomas. We did not demonstrate MOC31-positive cells in our small series of reactive effusions, which suggests that MOC3 1 expression by pleural or ascitic fluid cells may be used as a marker for malignancy. Obviously, a larger number of serous effusions need to be screened for MOC3 1-positive cells before a final conclusion can be drawn.
142
M. DELAHAYE ET AL.
Table 111-Sensitivity and specificity of immunocytochemistry for the diagnosis of malignant mesothelioma
Sensitivity Specificity Predictive value of a positive result Predictive value of a negative result
OV632 alone
All four markers*
92% (91%)t 80% (94%)
71% (70%) 94% (loo%o)
69% (9 1%)
85% (100%)
95% (94%)
87% (82%)
*Expected staining pattern is positive for OV632 and EMA combined, with negative staining for CEA and MOC31. ?In parentheses: only male patients.
Inclusion of OV632 in our panel proved to be most useful, since (1) this antibody is highly specific in its ability to distinguish malignant from reactive mesothelial cells and (2) the majority of malignant mesotheliomas and ovarian adenocarcinomas show immunoreactivity with OV632 in contrast to adenocarcinomas of other origin. The original publication on the immunohistochemical reactivity of OV632 on frozen tissue sections reported that benign, borderline, and malignant ovarian tumours were frequently stained with this antibody in contrast to adenocarcinomas of other origin." Remarkably, in this study none of 17 pulmonary adenocarcinomas was stained with OV632. Koelma et a l . 1 9 demonstrated that in 97 per cent of malignant effusions of patients with ovarian carcinomas OV632-positive tumour cells were found. Like us, these authors did not find OV632-positive cells in 20 reactive exudates. Thus, EMA + OV632 can now be applied to distinguish malignant from reactive mesothelial cells in cytological preparations. The use of OV632 alone in this group of selected patients gave high specificity for the diagnosis of malignant mesothelioma, particularly in male patients (Table 111). With a panel of four markers this specificity is increased. ACKNOWLEDGEMENTS
We wish to thank Mrs P. Delfos for photography, Mrs M. Hanegraaff for secretarial assistance, and Mr R. Teygeman for excellent immunostaining of the slides. REFERENCES 1. Suzuki Y. Pathology of human malignant mesothelioma. Semin Oncol
1980; 8: 268-282. 2. Whitaker D, Shilkin KB. Diagnosisofpleural malignantmesothelioma in life-a practical approach. J Pathol1984; 143: 147-175.
3. Tao LC. The cytopathology of mesothelioma. Acta Cytol 1979; 2 3 209-2 13. 4. Sherman ME, Mark EJ. Effusion cytology in the diagnosis of malignant epitheloid and biphasic pleural mesothelioma. Arch Pathol Lab Med 1990; 1 1 4 845-851. 5. Trio1JH, Conston AS, Chandler SV. Malignant mesothelioma. Cytopathology of 75 cases seen in a New Jersey Community Hospital. Acta Cyiol1984; 28: 3 7 4 5 . 6. Delahaye M, De Jong AAW, Versnel MA, Hoogsteden HC, Teeling P, Van der Kwast ThH. Cytopathology of malignant mesothelioma. Reappraisal of the diagnostic value of collagen cores. Cytopathology 1990; 1: 137-145. 7. Boon ME, Veldhuizen RW, Ruinaard C, Snieders MW, Kwee WS. Qualitative distinctive differences between the vacuoles of mesothelioma cells and of cells from metastatic carcinoma exfoliated in pleural fluid. Acta Cytol1984; 28: 4 4 3 4 4 9 . 8. Cibas ES, Corson JM, Pinkus GS. The distinction of adenocarcinoma from malignant mesothelioma in cell blocks of effusions: the role of routine mucin histochemistry and immunohistochemical assessment of carcinoembryonic antigen, keratin proteins, epithelial membrane antigen, and milk fat globule-derived antigen. Hum Pathol1987; 18: 67-74. 9. Dewar A, Valente M, Ring NP, Corrin B. Pleural mesothelioma of epithelial type and pulmonary adenocarcinoma: an ultrastructural and cytochemical comparison. J Pathol1987; 154: 309-316. 10. Kawai T, Greenberg SD, Truong LD, Mattioli CA, Klima M, Titus JL. Differences in lectin binding of malignant pleural mesothelioma and adenocarcinoma of the lung. Am J Puthol1988; 1 3 0 4 0 1 4 1 0 . 11. Tron V, Wright JL, Churg A. Carcinoembryonic antigen and milkfat globule protein staining of malignant mesothelioma and adenocarcinoma of the lung. Arch Pathol Lab Med 1987; 111: 291-293. 12. Van der Kwast ThH, Versnel MA, Delahaye M, De Jong AAW, Zondervan PE, Hoogsteden HC. Expression of epithelial membrane antigen on malignant mesothelioma cells. Actu Cyfol 1988; 3 2 169-1 74. 13. Kwee WS, Veldhuizen RW, Alons CA, Morawetz F, Boon ME. Quantitative and qualitativedifferencesbetween benign and malignant mesothelial cells in pleural fluid. Acta Cytol1982; 2 6 4 0 1 4 0 6 . 14. Kwee WS, Veldhuizen RW, Golding UP, et al. Histologic distinction between malignant mesothelioma, benign pleural lesion and carcinoma metastasis. Evaluation of the application of morphometry combined with histochemistry and immunostaining. Virchows Arch [Pathol Anat] 1982; 397: 287-299. 15. O H a r a CJ, Corson JM, Pinkus GS, Stahel RA. ME 1. A monoclonal antibody that distinguishes epithelial-type malignant mesothelioma and extrapulmonary malignancies. Am J Pathol1990; 136: 4 2 1 4 2 8 . 16. Singh G, Whiteside TL, Dekker A. Immunodiagnosis of mesothelioma. Use of a mesothelial cell serum in an indirect immunofluorescence assay. Cancer 1979; 43: 2288-2296. 17. Donna A, Betta PG, Bellingeri D, Marchesini A. New marker for mesothelioma: an immunoperoxidase study. J CIin Pathol 1986; 39: 961-968.
IMMUNOHISTOCHEMISTRY OF MESOTHELIOMA
18. Fleuren GJ, Coerkamp EG, Nap M, Van den Broek LJCM, Warnaar SO. Immunohistological characterization of a monoclonal antibody (OV632) against epithelial ovarian carcinomas. Virchows Arch [Pathol Anur] 1987; 410 481486. 19. Koelma IA, Nap M, Van Steenis GJ, Fleuren FJ. Tumor markers for ovarian cancer. A comparative immunohistochemical and immunocytochemical study of two commercial monoclonal antibodies (OV632 and OC125). Am JClin Pathol1988; 9 0 391-396. 20. Ordonez NG. The immunohistochemical diagnosis of mesothelioma. Differentiation of mesothelioma and lung carcinoma. Am J Surg Pathol1989; 13: 276291. 21. Stirling JW, Henderson DW, Spagnolo DV, Whitaker D. Unusual granular reactivity for carcinoembryonic antigen in malignant mesothelioma (Letter to the Editor). Hum Pathol 1990; 21: 678479. 22. Permanetter W, Wiesinger H. Immunohistochemical study of lysozyme, alpha,-anti-chymotrypsin, tissue polypeptide antigen, keratin and carcinoembryonic antigen in effusion sediments. Acta Cylol1987; 31: 1041 12. 23. Pinto MM, Bernstein L.H, Brogan DA, Criscuolo EM. Carcinoembryonic antigen in effusions. A diagnostic adjunct to cytology. Acta Cyrol1987;31: 113-118. 24. Ernst CS, Atkinson B, Chianese D, et al. Differential diagnosis between mesotheliomas and metastatic adenocarcinomas using monoclonal antibodies against gastrointestinal carcinoma antigen and stage-specificembryonic antigen. Appl Pathol1986; 4 115-124.
143
25. Sheibani K, Battifora H, Burke JS. Antigenic phenotypeofmalignant mesothelioma and pulmonary adenocarcinomas. An immunohistologic analysis demonstrating the value of LEU MI antigen. Am J
Pathol1986; 123: 212-219. 26. Noguchi M, Nakajima T, Hirohashi S, Akiba T, Shimosato Y. Immunohistochemical distinction of malignant mesothelioma from pulmonary adenocarcinoma with anti-surfactant apoprotein, anti-Lewis (a), and anti-Tn antibodies. Hum Pathol1990; 2 0 54-57. 27. Hilkens J, Buijs F, Hilgers J, ef al. Monoclonal antibodies against human milk-fat globule membranes detecting differentiation antigens of the mammary gland and its tumors. Int J Cancer 1984; 31: 1977206. 28. To A, Dearnaley DP, Ormerod MG, Canti G, Coleman DV. Epithelial membrane antigen. Its use in the cytodiagnosis of malignancy in serous effusions. Am JClin Pathol 1982; 7 8 214-219. 29. Leong ASY, Parkinson R, Milios J . ‘Thick’cellmembranes revealed by immunocytochemicalstaining: a c h e to thediagnosisofmesothelioma. Diagn Cytopathol1990; 6: 9-1 3. 30. Berendsen HH, De Leij L, Postmus E, el a/. Small cell lung cancer. Tumour cell phenotype detected by monoclonal antibodies and response to chemotherapy. Chest 1987; 91 Suppl. 1 lS-12S. 31. De Leij L, Postmus PE, Poppema S, Elema JD, The TH. The use of monoclonal antibodies for the pathological diagnosis of lung cancer. In: Hansen HH, ed. Lung Cancer: Basic and Clinical Aspects. Boston: Martinus Nijhoff, 1986; 3 1 4 8 .
