Fine-needle aspiration cytology of metaplastic carcinoma of the breast

BACKGROUND: Metaplastic carcinoma of the breast encompasses a heterogeneous group of tumours with variable components of sarcomatoid, squamous or poorly differentiated carcinomas. AIM: To review a series of 19 cytological preparations of metaplastic carcinomas to assess diagnostic cytological features. METHODS: 17 cases of fine-needle aspirates of histologically proven metaplastic carcinomas (4 monophasic spindle cell carcinomas, 4 squamous cell carcinomas and 11 biphasic tumours) were reviewed, with an emphasis on the presence of poorly differentiated carcinoma, squamous cell carcinoma, atypical spindle cells, benign stromal fragments and necrosis. RESULTS: All cases were diagnosed as malignant, with 68% of cases showing moderate to high cellularity, and 47% showing necrosis. If the tumours were analysed according to the constituting components histologically, 7, 15 and 8 cases, respectively, possess poorly differentiated carcinoma cells, sarcomatoid malignant cells and squamous carcinoma cells, whereas these components were cytologically identified in 11, 10 and 7 cases, respectively. Dual tumour populations were identified in only 5 of the 11 biphasic carcinomas in the cytological preparations; and the stromal material was cytologically identified in the only case with chondroid stroma. CONCLUSIONS: Identification of metaplastic carcinoma in cytology remains problematic. There seems to be morphological overlap between various components. The identification of dual components, unequivocal squamous carcinoma cells and chondroid stroma is helpful for diagnosis, but it is uncommon. The presence of poorly differentiated carcinoma cells with a suggestion of focal spindle morphology is another clue to the suggestion of metaplastic carcinoma.     

S-100 immunoreactivity in primary and metastatic carcinoma of the breast: a potential source of error in immunodiagnosis

Immunoreactivity for S-100 protein, typically a marker for malignant melanoma and neural-derived tumors, was observed in neoplastic cells of 57 of 68 cases (84%) of formalin-fixed, paraffin-embedded primary and/or metastatic carcinoma of the breast of various histologic types. The extent of S-100 immunoreactivity varied, with only a minor proportion of positive tumor cells noted in some cases. An awareness of this staining profile for S-100 protein, particularly in metastatic poorly differentiated neoplasms with unknown primaries, is imperative for accurate immunohistochemical interpretation. Using a panel of reagents which includes antibodies to keratin proteins and epithelial membrane antigen, the epithelial nature of S-100-positive carcinomas may be readily defined. Tumor cells in all cases of primary and metastatic carcinoma of the breast evaluated in this study exhibited strong staining for both of these tissue markers. To preclude misinterpretation of tumor type due to anomalous staining patterns for a specific antibody, eg, S-100 protein, a panel of antibodies is recommended for assessment of metastatic poorly differentiated tumors.     

In serous ovarian neoplasms the frequency of Ki-ras mutations correlates with their malignant potential

Ki-ras mutations by denaturing gradient gel electrophoresis (DGGE) and direct sequencing after microdissection. Point mutations at codon 12 were found in 7 of 20 tumours of low malignant potential (LMP) (35%) and in 2 of 6 well-differentiated carcinomas (33%). In contrast, no mutations were detected in the 11 poorly differentiated ovarian carcinoma samples or in the 7 serous cystadenomas. The frequency of Ki-ras mutations in serous ovarian tumours seems to correlate with the malignant potential of the neoplasms. The data favour the hypothesis of a de novo development of poorly differentiated ovarian carcinomas and do not support an evolution from LMP tumours or well-differentiated carcinomas. Received: 8 June 1998/Accepted: 8 October 1998     

Poorly differentiated cutaneous apocrine carcinomas: histopathological clues and immunohistochemical analysis for the diagnosis of this unusual neoplasm

Primary cutaneous apocrine carcinoma (PCAC) is a rare cutaneous malignancy that is derived from apocrine glands. Histologically, these tumours can appear well-differentiated where diagnosis should be relatively straightforward. However, occasionally these tumours can exhibit high-grade features, and in such instances the diagnosis can be challenging. A retrospective analysis of 12 cases of poorly differentiated PCAC, obtained from large academic institutions, was performed, and summarised below. Immunohistochemical studies were performed in all cases with antibodies against CK7, p63, CAM 5.2, GCDFP-15, GATA3, CEA, PR, ER, HER2, calponin, SMA, androgen receptor and EMA. All 12 cases were poorly differentiated; however, there were some histopathological clues to the diagnosis of apocrine carcinoma; namely, the presence of focal glandular formation, acrosyringial involvement and the presence of single ‘pagetoid’ cells within epidermis. All tumours were consistently positive for CK7, GATA3 and GCDFP-15 and negative for p63. The tumours had variable expression of CAM5.2, CEA, ER, PR, HER2, androgen receptor and EMA. In three cases, there was a preservation of the myoepithelial cell layer (with calponin and SMA), which also confirmed the primary cutaneous origin. PCAC is a difficult neoplasm to diagnose, as it can appear identical to metastatic carcinomas. We describe 12 cases of poorly differentiated PCAC, highlighting their salient clinical, histopathological and immunohistochemical features, and discuss the potential diagnostic pitfalls in distinguishing this entity from other malignant neoplasms. Our results indicate that a combination of thorough histological inspection coupled with an adequate battery of immunohistochemical stains is necessary to support the diagnosis of PCAC.     

Extra-medullary myeloid tumour (granulocytic sarcoma) is often misdiagnosed: a study of 26 cases

AIMS: To describe the clinicopathological and immunophenotypic features of 26 cases of extra-medullary myeloid tumour (EMMT)/granulocytic sarcoma, which remains poorly recognized and is frequently confused with malignant lymphoma, and to discuss the main diagnostic problems experienced by the referring pathologist. METHODS AND RESULTS: Haematoxylin and eosin (H & E) sections of 26 cases of EMMT were re-examined. Immunostains for myeloperoxidase, lysozyme, neutrophil elastase, LCA, CD79a, CD20, CD43, CD45RO, CD3, CD30, CD15, CD68, MAC387, VS38C, MIC2, and the Leder stain for naphthol-ASD-chloroacetate esterase were performed on all cases. Clinical and follow-up data were obtained through a questionnaire to the referring pathologist or from the notes of the patients where available. In the 10 cases with known myeloproliferative disease, the initial diagnosis was correct in 10 whereas all cases presenting with EMMT without a previous history of myeloproliferative disorder had an initial incorrect diagnosis. The most common suggested diagnosis was that of a non-Hodgkin's lymphoma. The morphology of the tumours varied from well differentiated which included all stages of myeloid differentiation to poorly differentiated or blastic showing little or no evidence of myeloid differentiation. The proportion of positive cells for each stain varied. Chloroacetate esterase, myeloperoxidase and CD15 stained a large proportion of cells of the majority of the well differentiated tumours and a smaller proportion of the poorly differentiated/blastic tumours with very focal staining of some of the cases. Lysozyme and CD43 were the most sensitive of the markers staining a large proportion of cells of the majority of the tumours in both groups. Neutrophil elastase was the least sensitive of the markers of myeloid differentiation. CD79a, CD20, CD3 and CD30 were negative in all cases. CD43 was positive in all cases. CD68 stained a substantial number of cells in the majority of tumours. A smaller proportion of the tumours stained with MAC387. Four of the tumours showed positivity for MIC2. One tumour was positive for VS38C. CONCLUSION: This series documents continuing difficulties in the diagnosis of EMMT. Even well differentiated tumours are frequently mistakenly diagnosed as malignant lymphomas when they present without any history of antecedent myeloproliferative disorder. Careful evaluation of morphology for evidence of myeloid differentiation and a high index of suspicion when confronted with a less differentiated neoplasm are required to avoid this important diagnostic error. We suggest that a panel which includes chloroacetate-esterase, myeloperoxidase, lysozyme and CD43, together with other B- and T-lineage markers, in particular CD79a and CD3 should be used to confirm the diagnosis.     

KP1 (CD 68) staining of malignant melanomas

The monoclonal antibody KP1, which recognizes the CD 68 antigen on macrophages and myeloid precursors, was tested on 28 malignant (primary and metastatic) melanomas, 28 naevi, and 17 skin biopsies showing either normal (10) or hyperplastic melanocytes (7). Sixteen of 20 primary melanomas and six of eight metastatic melanomas showed variable numbers of KP1 positive tumour cells. All but five benign melanocytic proliferations (two Spitz naevi and three intradermal naevi), as well as normal and hyperplastic melanocytes were negative. These results indicate that difficulties may occur with the use of KP1 in the differential diagnosis between melanomas and neoplasms derived from histiocytes-macrophages, and that the expression of CD 68 antigen might be related to tumour progression in melanocytic cells.     

Immunohistochemistry of ovarian granulosa cell tumours

Summary Paraffin-embedded material of 47 ovarian tumours primarily diagnosed as granulosa cell tumours, including 2 cases of juvenile granulosa cell tumour, were studied immunohistochemically for the presence of intermediate filament proteins, epithelial membrane antigen and tumour markers. Forty-one lesions, including the 2 juvenile granulosa cell tumours, were vimentin positive, while keratin and epithelial membrane antigen expression could not be detected. Six tumours primarily diagnosed as poorly differentiated malignant granulosa cell tumours were vimentin negative, showed a mild to moderate positivity for keratin and intense positivity with the anti-epithelial membrane antigen antibody. These latter tumours were therefore classified as undifferentiated ovarian carcinomas, corresponding to their significantly poorer prognosis and shorter survival when compared with the granulosa cell tumours. Two of these six tumours were positive for carcino-embryonic antigen. Two small cell carcinomas of the ovary studied in addition expressed keratin in a proportion of tumour cells while no epithelial membrane antigen or vimentin was detectable. None of the tumours tested for alpha-fetoprotein, human chorionic gonadotrophin, human placental alkaline phosphatase and human placental lactogen, were positive.The data indicate the value of antibodies directed against intermediate filament proteins and epithelial membrane antigen to distinguish granulosa cell tumours from poorly differentiated carcinomas, a worthwhile distinction considering the much better prognosis of granulosa cell tumours.     

Galectin-3, a marker of well-differentiated thyroid carcinoma, is expressed in thyroid nodules with cytological atypia

AIMS: The distribution of galectin-3, a widely recognized marker of well-differentiated thyroid carcinoma, was investigated in 95 thyroid lesions including nodules with foci of cytoarchitectural atypia. METHODS AND RESULTS: Twenty-eight papillary carcinomas, five follicular carcinomas, one Hurthle cell carcinoma, three poorly differentiated carcinomas, one anaplastic carcinoma, 25 nodular hyperplasias and 27 follicular adenomas, including nodules with atypical features, three neoplasms of undetermined malignant potential and two thyroiditis cases were examined. By immunohistochemistry, galectin-3 was consistently found in carcinomas; otherwise benign nodules exhibited galectin-3-positive clusters of cells with poorly developed features of differentiated carcinoma (mainly of papillary type) such as nuclear chromatin clearing, nuclear clefting, pseudoinclusions, which, in each case, were not histologically sufficient to warrant a definitive diagnosis of malignancy. In other nodules galectin-3 staining was negative. The latter were either clearly benign or showed constantly a minor degree of chromatin clearing and of other atypical features when compared with galectin-3-positive cases. CONCLUSIONS: Galectin-3, a reliable marker of differentiated thyroid carcinoma as confirmed in our series of malignant neoplasms, appears expressed in nodules with an overall benign appearance but with focal areas suspicious for malignancy. The significance of such findings needs to be further investigated.     

Angiomyolipoma of the kidney. Immunoreactivity with HMB-45. Light- and electron-microscopic findings

Immunoreactivity with HMB-45 has recently been described in renal angiomyolipoma, a tumour of smooth muscle cells. HMB-45 is a monoclonal antibody that reacts specifically with melanosomes. In order to determine whether the tumour cells contain melanosomes and synthesize melanin, seven tumours were studied by light microscopy and immunohistochemically with the antibodies HMB-45, KP1 (CD68), PG-M1 (CD68), Ki-M1P, anti-lysozyme, anti-smooth-muscle actin, anti-vimentin, anti-S100 protein and KL1 (anti-keratin). Two tumours were also studied by electronmicroscopy and one by immuno-electronmicroscopy. Histochemical investigation for dopa oxidase was performed on cryostat sections. The tumours contained varying numbers of HMB-45-positive muscle cells. Reactivity was noted in lysosomal granules and rough endoplasmic reticulum. Typical premelanosomes were found in the tumour cells by electronmicroscopy. Groups of tumour cells stained for dopa oxidase. The tumour cells were not reactive for lysozyme, but reacted with KP1, PG-M1 and Ki-M1P. Immuno-electronmicrosopy showed that reactivity for KP1 was located within lysosomal granules. The findings show that the tumour cells of renal angiomyolipoma contain premelanosomes and that they are able to synthesize melanin, because they contain dopa oxidase. Immunoreactivity with KP1, PG-M1 and Ki-M1P can be attributed, in the absence of staining for lysozyme, to the large number of lysosomal granules. The tumour cells were not found to be related to macrophages or myeloid cells.     

Neuronal and glial markers in tumours of neuroblastic origin

Summary The presence and distribution of different neural markers in 30 neuroblastic tumours (neuroblastomas, ganglioneuroblastomas) and 6 non-neuroblastic tumours were investigated by immunocytochemistry. Neuron-specific enolase (NSE), S-100 protein, tyrosine hydroxylase, neurofilaments and glial fibrillary acidic protein (GFAP) were localized in 3 undifferentiated neuroblastic tumours (group A), 12 poorly differentiated tumours (group B) and 15 well differentiated neuroblastic tumours (group C). Non-neuroblastic tumours (3 lymphomas and 3 Ewing sarcomas) showed no immunoreactivity.Tyrosine hydroxylase and, in particular, NSE were found in mature ganglion cells and developing neuroblasts of poorly and well differentiated tumours (groups B and C). S-100 was localised in neuroblasts with slender cytoplasmic processes in the same groups. Neurofilaments were detected in ganglion cells and differentiated neuroblasts (groups B and C) while GFAP was localised in immature neuroblasts of undifferentiated and poorly differentiated tumours (groups A and B). Thus, there are differences in the neural proteins found in neuroblastic tumours and a wide panel of antibodies against neural markers may be a useful tool in the histological assessment of nervous system neoplasms.     

Carcinomas of the breast showing myoepithelial cell differentiation

 Myoepithelial cells are normally located between the epithelial cells and the basal lamina of secretory elements of exocrine glands. Their role in the histogenesis of breast tumours has been studied extensively, and a definite differentiation towards myoepithelial cells has been demonstrated in adenoid cystic carcinoma, adenomyoepithelioma, low-grade adenosquamous (syringomatous) carcinoma, pure malignant myoepithelioma and poorly differentiated myoepithelial-rich breast carcinoma. All these tumours are of low malignancy, with the exception of malignant myoepithelioma and poorly differentiated myoepithelial-rich carcinoma. When a low-grade tumour is associated with a spindle cell component, distant metastases must be expected. Pure malignant myoepithelioma shows morphological and clinical features similar to those of monophasic sarcomatoid carcinomas, and it is possible that this last tumour is linked histogenetically to sarcomatoid carcinomas. Received: 8 April 1997 / Accepted: 10 November 1997     

Epithelial antigens in malignant mixed Müllerian tumours of endometrium

Immunohistochemical staining of formalin fixed paraffin sections with the mouse monoclonal antibodies E29 and CAM 5.2 effectively distinguish epithelium (antigen positive) from stroma (antigen negative) in normal endometrium. These antibodies were used as epithelial markers in the study of eight malignant mixed Müllerian tumours (MMT) of endometrium and gave normal reactions with well differentiated neoplastic glands; in contrast, negative or very weak staining was observed in poorly differentiated epithelial cells, present in large numbers in three cases. Abnormal antigen-containing cells were observed in the stroma of seven MMT. In some cases these are probably due to stromal invasion by malignant epithelium but in others they may represent an early stage of epithelial differentiation in mesenchymal cells of the malignant stroma.     

Comparison of keratin monoclonal antibodies MAK-6, AE1:AE3, and CAM-5.2

Two routinely used antikeratin monoclonal antibodies, AE1:AE3 (Hybritech Inc., La Jolla, CA) and CAM-5.2 (Becton-Dickinson, Mountain View, CA), were compared with a new antikeratin monoclonal antibody mixture, MAK-6 (Triton Biosciences, Inc., Alameda, CA). Various poorly differentiated epithelial neoplasms, lymphomas, melanomas, and sarcomas were studied with the use of the avidin-biotin-peroxidase technic. All three antibodies had similar staining profiles, however, some differences were noted. The MAK-6 and CAM-5.2 antibodies illustrated stronger staining for some grade III transitional cell carcinomas, whereas the AE1:AE3 was variably positive to negative in all cases. Three renal cell carcinomas were positive with all three antibodies, two were negative with all three, and one had scattered positive cells with MAK-6 only. All lymphomas and plasmacytomas were negative with MAK-6 and CAM-5.2, however, AE1:AE3 faintly stained two of three plasmacytomas and two of the seven large cell lymphomas. Two out of three hepatomas evaluated were strongly positive with CAM-5.2 and MAK-6 and variably positive with AE1:AE3. The third case had both positive and negative cells with all three antibodies. In conclusion, MAK-6 antikeratin antibody is as useful as AE1:AE3 and CAM-5.2 for identification of poorly differentiated epithelial neoplasms.     

WT-1 assists in distinguishing ovarian from uterine serous carcinoma and in distinguishing between serous and endometrioid ovarian carcinoma

AIMS: It has been suggested that WT-1 is helpful in distinguishing a primary ovarian serous carcinoma (OSC) from a primary uterine serous carcinoma (USC). Since both neoplasms are often disseminated at diagnosis and since USC often spreads to the ovary and vice versa, it may be difficult to ascertain the primary site. This is important, since adjuvant therapies for OSC and USC may differ. WT-1 staining patterns also differ between OSC and ovarian endometrioid carcinoma and so it is possible that WT-1 may assist in the distinction of these two neoplasms, which is sometimes problematic, especially with poorly differentiated tumours. This study aims to document the value of WT-1 in these settings. Cases of ovarian borderline serous tumour, primary peritoneal serous carcinoma (PPSC) and uterine endometrioid carcinoma were also studied. METHODS AND RESULTS: Cases of OSC (n = 38), USC (n = 25) (in five of these cases there was also a component of endometrioid adenocarcinoma), ovarian endometrioid carcinoma (n = 13), uterine endometrioid carcinoma (n = 7), ovarian borderline serous tumour (n = 16) and PPSC (n = 6) were stained with WT-1. Cases were scored on a scale of 0-3, depending on the percentage of positive cells. The intensity of staining was scored as weak, moderate or strong. There was positive nuclear staining of 36 of 38 (94.7%) OSC with WT-1. In most OSC (68.4%), >50% of cells stained positively and staining was usually strong. Five of 25 (20%) USC were positive with only two cases exhibiting staining of >50% of cells. All primary ovarian and uterine endometrioid carcinomas were negative. All PPSC were positive, usually with diffuse strong immunoreactivity. Fourteen of 16 borderline serous tumours exhibited positivity with WT-1. CONCLUSIONS: WT-1 is useful in distinguishing OSC (characteristically diffuse strong nuclear positivity) from USC (characteristically negative). However, rarely OSC is negative and occasional cases of USC are positive. WT-1 may also be helpful in differentiating poorly differentiated OSC from poorly differentiated ovarian endometrioid carcinoma.     

Cellular origins of ultraviolet radiation-induced corneal tumours in the grey, short-tailed South American opossum (Monodelphis domestica)

Corneal tumours were induced in almost 100% of grey, short-tailed South American opossums (Monodelphis domestica) exposed three times weekly to ultraviolet radiation (UVR) for periods of a year or more. Five tumours, representing the morphological spectrum of UVR-induced corneal tumours (two fibrosarcomas, one malignant fibrous histiocytoma, one putative haemangiosarcoma, and one squamous cell carcinoma overlying a sarcoma), were assayed immunohistochemically for reactivity with antibodies against the intermediate filaments vimentin, smooth muscle actin (alpha isoform), muscle-specific actins (alpha and gamma isoforms), desmin and cytokeratin, and with antibodies against the vascular endothelial marker von Willebrand factor. The squamous cell carcinoma was cytokeratin-positive. Other tumours were cytokeratin-negative and vimentin-positive. Three tumours had scattered individual cells and groups of cells immunoreactive with antibodies against smooth muscle actin and muscle-specific actins; two tumours (a fibrosarcoma and the malignant fibrous histiocytoma) had small numbers of desmin-positive cells. The putative haemangiosarcoma contained two populations of neoplastic cells, von Willebrand factor-positive vascular endothelial cells and smooth muscle actin-positive spindle cells. It was concluded (1) that UVR-induced corneal tumours may be composed of cells derived from resident epithelial cells, immigrant vascular endothelial cells, or fibroblast-like cells of unknown origin, and (2) that such tumours may contain more than one neoplastic cell type.     

Granulomatous prostatitis and poorly differentiated prostate carcinoma. Their distinction with the use of immunohistochemical methods

Granulomatous prostatitis and poorly differentiated prostate carcinoma can mimic each other both clinically and histologically. To develop criteria useful in resolving problem cases, the authors compared the reactivities of these conditions (nine cases of granulomatous prostatitis and six cases of poorly differentiated carcinoma) with a panel of antibodies to cytokeratin (AE1/3), prostate-specific antigen (PSA), prostatic acid phosphatase (PAP), lysozyme, antimacrophage M, and leukocyte common antigen (LCA). In granulomatous prostatitis, histiocytes were not immunoreactive for PAP, PSA, or cytokeratin; however, histiocytes reacted to lysozyme in nine of nine cases, antimacrophage M in seven of nine cases, and LCA in one of nine cases. Tumor cells from all six carcinoma cases reacted with PAP, PSA, and cytokeratin; all failed to react with lysozyme, LCA, and antimacrophage M. The authors conclude that granulomatous prostatitis and poorly differentiated carcinoma can be reliably distinguished with immunohistochemical methods.     

Primary and secondary heart tumours in dogs and cats

Primary and secondary neoplasms of the canine and feline heart are uncommon. During a 2-year period, 83 dogs suffering from primary cardiac (n=11), extracardiac benign (n=6) or malignant (n=66) tumours and 30 cats with primary cardiac (n=1) or extracardiac (n=29) malignant tumours were examined. Echocardiography revealed four cases of primary cardiac neoplasms in dogs, but secondary heart tumours were not detected. After necropsy, tissue samples from the heart and tumours were examined histologically and immunohistochemically. In dogs, primary neoplasms included seven haemangiosarcomas, two chemodectomas, one rhabdomyosarcoma, and one neurofibrosarcoma. In 24 of 66 dogs examined, metastases of extracardiac neoplasms were found in the heart (15 carcinomas, six malignant lymphomas, three haemangiosarcomas). In cats, one case of primary haemangiosarcoma of the pericardium and five cases of secondary cardiac tumours (two malignant lymphomas, three carcinomas) occurred. Cardiac neoplasms in cats were not identified clinically but were detected by detailed gross sectioning of the heart (n=2) or histopathological examinations (n=3). This study showed an unexpectedly high number (36%) of dogs with cardiac metastases.     

Clusterin, a marker for anaplastic large cell lymphoma immunohistochemical profile in hematopoietic and nonhematopoietic malignant neoplasms

We evaluated the immunohistochemical staining profile of clusterin in paraffin sections of 143 neoplasms (non-Hodgkin lymphoma, 83, including 41 anaplastic large cell lymphomas [ALCLs]; Hodgkin lymphoma, 17; primary and metastatic carcinoma, 30; and other neoplasms, 13). In 40 of 41 ALCLs (34 systemic, 7 cutaneous), neoplastic cells revealed clusterin reactivity characterized by a Golgi staining pattern. The proportion of reactive cells varied with more than 25% positive cells in the majority of cases. In 7 non-Hodgkin lymphomas of other types, fine cytoplasmic (3 cases) or strong membranous reactivity (4 cases) was observed for clusterin. In Hodgkin lymphoma, rare Reed-Sternberg cells exhibited focal cytoplasmic or membranous clusterin positivity. In the nonhematopoietic neoplasms, a Golgi staining pattern was apparent in only 2 cases, 1 lobular carcinoma of the breast and 1 poorly differentiated colonic carcinoma; however, cytoplasmic reactivity was noted in 12 of 30 carcinomas and 1 of 5 neuroendocrine neoplasms. A Golgi pattern of reactivity for clusterin seems highly characteristic of ALCL among hematopoietic neoplasms, but also might be observed in rare nonhematopoietic tumors, necessitating the use of a broad immunohistochemical panel for evaluation of poorly differentiated neoplasms of indeterminate derivation.     

Immunocytological staining reactions of anti-carcinoembryonic antigen, Ca, and anti-human milk fat globule monoclonal antibodies on benign and malignant exfoliated mesothelial cells.

A panel of three monoclonal antibodies were used in an immunoalkaline phosphatase staining method on a series of serous effusion samples from cases of mesothelioma, lung carcinoma, and benign disease. The antibodies used were anti-carcinoembryonic (CEA) antigen, Ca, and anti-human milk fat globule membrane antigen. Antibodies to the Ca antigen and human milk fat globule membrane antigen stained 75% and 83% of mesothelioma and 75% of cases of lung carcinoma, respectively. The anti-CEA antibody stained most cases of lung carcinoma strongly but was negative on 11 of 12 cases of mesothelioma and showed weak staining on one case. Benign cases were negative with all three antibodies. These three antibodies may be useful in distinguishing benign and malignant mesothelial cells and lung carcinoma in serous effusions.