The differentiation between malignant mesothelioma and adenocarcinoma based on morphology alone can be a diagnostic challenge. The majority of the available antibodies recognize molecules expressed by adenocarcinoma whereas to the authors' knowledge specific markers for mesothelial cells are lacking. Calretinin, a calcium-binding protein, has been reported to be a selective marker for mesothelioma and largely is absent from adenocarcinoma on histologic material. The results with cytologic preparations have been inconsistent.
To evaluate the specificity of calretinin in differentiating mesothelioma from adenocarcinoma in cytologic preparations, 21 paraffin embedded cells blocks of serous effusions from 15 patients with metastatic adenocarcinoma and 16 cell blocks from 9 patients with malignant mesothelioma were stained with a monoclonal antibody against calretinin. The immunoreactivity was evaluated blindly by two observers. Positive staining was defined as nuclear and cytoplasmic staining with or without intense membranous decoration. The former resulted in a characteristic “fried egg” appearance.
Calretinin staining was positive in all but 2 cases of mesothelioma (14 of 16 cases; 87.5%). The latter contained predominantly spindle-shaped neoplastic mesothelial cells in the cell block preparations. All adenocarcinoma specimens were classified as negative for calretinin staining; 9 (42.9%) lacked any immunoreactivity and 12 (57.1%) showed weak, sparse, coarse, granular cytoplasmic staining without nuclear or membranous staining. Benign reactive mesothelial cells, when observed in association with adenocarcinoma, also showed the characteristic “fried egg” appearance. The difference in the staining pattern of calretinin between cells of mesothelial origin and adenocarcinoma cells was statistically significant.
A serous effusion may be the first manifestation of metastatic adenocarcinoma or malignant mesothelioma. Cytologic analysis of the effusion may allow a definite diagnosis; however, the distinction between these two entities by morphologic criteria alone may be difficult. Various ancillary studies have been used to increase the diagnostic accuracy of cytology.1–8 Immunocytochemistry is the most commonly employed “special” technique and often involves the use of a panel of antibodies such as carcinoembryonic antigen (CEA), epithelial membrane antigen, B72.3, Ber-Ep4, and LeuM1.9–13 Most of the commonly used antibodies recognize molecules commonly expressed by adenocarcinomas but not by mesotheliomas. Markers that recognize molecules expressed by mesothelial cells and mesotheliomas are scarce and rarely used for diagnostic purposes because of their limited specificity or sensitivity.14–21
Calretinin is a 29-kilodalton (kD) calcium-binding protein that is expressed normally in neurons of the central and peripheral nervous system.22 Recently, various investigators have shown that calretinin is a reliable marker for differentiating malignant mesothelioma from metastatic adenocarcinoma in histologic preparations.23–27 The results with cytologic preparations have been conflicting.23, 28, 29 In a study by Barberis et al., staining with anticalretinin antibody was identified in 100% of all malignant mesotheliomas but in only 23% of metastatic carcinoma in serous effusions.28 Conversely, Simsir et al. showed that 27% of reactive mesothelial cells, 35% of malignant melanoma, and 48% of adenocarcinomas demonstrated staining with anticalretinin antibody.29 These conflicting results prompted us to examine the diagnostic reliability of calretinin as a marker of mesothelial cells in cytologic preparations. We also evaluated the staining pattern of calretinin in different cell types, which to our knowledge has not been emphasized previously.
MATERIALS AND METHODS
Formalin fixed, paraffin embedded cell blocks from 37 cytologic specimens of pleural or peritoneal fluids were retrieved from the files of the Department of Cytopathology of New York University Medical Center/Bellevue Hospital Center for the period 1990–1997. These cytologic specimens were comprised of 16 malignant mesotheliomas from 9 patients and 21 metastatic adenocarcinomas from 15 patients. Three patients had two specimens each and four patients had three specimens each; all were collected at different time intervals. The primary sites of origin of the metastatic adenocarcinomas were the gastrointestinal tract (1 patient), breast (8 patients), and lung (12 patients). The malignant mesotheliomas were confirmed by histology (seven patients) or by ultrastructural and immunocytochemical examination of the cytologic specimens (two patients). The subclassification of the malignant mesotheliomas based on histology included five epithelial and two biphasic variants. Hematoxylin and eosin stained sections of each cell block were evaluated for the presence of tumor.
Five-μm sections were cut and deparaffinized. Heat-induced epitope retrieval was accomplished by microwave heating for 10 minutes in 0.01 M citrate buffer (pH 6.0). Immunostaining was performed using a modified avidin-biotin-peroxidase technique. An undiluted anticalretinin monoclonal antibody (MoAb) (Zymed, San Francisco, CA) was used with overnight incubation at room temperature. The immunostaining was performed with a semiautomated staining machine (Ventana Medical System, Tucson, AZ). Paraffin sections of normal human cerebral cortex were used as positive controls. A negative control was incubated without the MoAb.
The stained sections were evaluated independently by two observers who had no prior knowledge of the original cytologic diagnosis for the following: the staining intensity (absent, weak, or strong), the staining pattern (cytoplasmic, membranous, or nuclear), and the percentage of immunostained cells (< 10%, 10–50%, or > 50%). Positive staining was defined as diffuse nuclear and cytoplasmic staining with or without intense membranous decoration. The former resulted in a characteristic “fried egg” appearance.
The results of calretinin immunoreactivity of the 37 cell block preparations are summarized in Tables 1 and 2. All cases of malignant mesothelioma except 2 (14 of 16 cases; 87.5%) reacted positively with anticalretinin (Fig. 1). The immunoreactivity was strong and diffuse (> 50% of the cell population). Staining was noted in both the cytoplasm and the nucleus. In 9 cases (56.3%) intense membranous decoration was noted in addition to the diffuse cytoplasmic and nuclear staining, thus resulting in a “fried egg” appearance. The two malignant mesotheliomas that were negative for calretinin were from two different patients and contained predominantly spindle neoplastic cells in the cell block preparations (Fig. 2). One was classified as a biphasic variant histologically. No histologic follow-up was available in the other case.
Table 1. Summary of Cases in Current Study
In addition to the diffuse cytoplasmic and nuclear staining, intense membranous decoration was noted thus resulting in a “fried egg” appearance.
Nine adenocarcinomas (four from the breast and five from the lung) were negative for anticalretinin. The remaining 12 adenocarcinomas (1 from the gastrointestinal tract, 4 from the breast, and 7 from the lung) demonstrated focal (< 50%) and weak immunoreactivity to anticalretinin. The staining was confined to spotty, coarse granular cytoplasmic staining without nuclear or membranous decoration (Fig. 3). In five cases of metastatic adenocarcinoma, the accompanying benign reactive mesothelial cells showed the characteristic “fried egg” staining pattern whereas the neoplastic adenocarcinoma cells were negative (Fig. 4).
The differences in the calretinin staining pattern between mesothelioma and adenocarcinoma were statistically significant (chi-square test, P < 0.0001). Both the sensitivity and specificity of calretinin as a positive marker for epithelial mesothelioma was 100%.
Both malignant mesothelioma and metastatic adenocarcinoma can present with a serous effusion and their distinction has always been a diagnostic challenge to both clinicians and cytologists. A history of asbestos exposure does not necessary imply a diagnosis of malignant mesothelioma because patients with such a history also are at risk of developing lung carcinoma.30, 31 Furthermore, not all mesotheliomas are related to asbestos exposure; in some series, as many as 85% of patients diagnosed with malignant mesotheliomas had no evidence of asbestos exposure.32, 33 Various cytologic features are characteristic of, but not specific for, mesothelioma. For example, intercellular spaces (windows), commonly observed in cellular aggregates of mesothelial cells, also can be identified in 13% of cases of metastatic adenocarcinoma.34 Therefore, ancillary studies often are performed to assist in the differential diagnosis. The demonstration of neutral mucin production by cytochemistry is a special technique employed for this purpose.35, 36 However, cytochemical stains are relatively insensitive and lack specificity.37, 38 Electron microscopy also has been used in differentiating mesothelioma from adenocarcinoma by demonstrating long, thin, “bushy” microvilli in the mesotheliomas.39, 40 Availability of adequate material, sizable costs, and long turnaround times have limited the universal implementation of this technique. Furthermore, not all mesotheliomas display the characteristic ultrastructural features mentioned earlier.41
In recent years, immunocytochemistry has contributed greatly to the differentiation of mesothelioma and adenocarcinoma.9–13 Most of the currently available markers such as CEA, LeuM1, Ber-Ep4, and B72.3 recognize molecules commonly expressed by adenocarcinomas but not by mesotheliomas. A diagnosis of metastatic adenocarcinoma is favored if the neoplastic cells demonstrate positive immunoreactivity for CEA, LeuM1, B72.3, or Ber-Ep4, which are expected to be absent in the majority of mesotheliomas. Therefore, a panel of two or more antibodies often is used to confirm (or rule out) the possible diagnosis of metastatic adenocarcinoma but cannot support a positive identification of mesothelioma.42, 43
To our knowledge specific markers for mesotheliomas are scarce. Furthermore, their specificity and sensitivity are limited.14–21 Among the alleged specific mesothelial markers, the ME1 MoAb, which recognizes a molecule present on the surface of mesothelial cells, reacts with up to 40% of carcinomas.20 The K1 MoAb, generated by immunizing mice with the OV-CAR3 ovarian carcinoma cell line, reacts with epithelial and biphasic mesotheliomas as well as 70% of nonmucinous ovarian carcinomas.15, 21 Thrombomodulin, a transmembrane glycoprotein normally expressed by endothelial and mesothelial cells, was found to stain approximately 67% of the mesotheliomas, as well as > 50% of the adenocarcinomas in effusion cytology.19 HBME-1, an antigen present on the microvillus surface of the mesothelial cell, was expressed by > 50% of the adenocarcinomas.18, 19 N-cadherin, a member of the cadherin family, was reported to be immunoreactive with 77% of reactive mesothelial cells, 35% of malignant mesotheliomas, and 48% of adenocarcinomas.29
Calretinin, a calcium-binding protein of 29- kD, belongs to a large family of EF-hand proteins.22 It is expressed abundantly in central and peripheral neural tissues and is believed to play a key role in somatosensory transduction.44, 45 Recently, calretinin has been found to be a selective marker for mesothelial cells.23–27 Several authors have investigated the use of calretinin as a positive marker for mesothelial cells in cytologic specimens, with conflicting results.23, 28, 29
The results of the current study support the use of calretinin to differentiate mesotheliomas from metastatic adenocarcinomas in cytologic preparations. All mesotheliomas containing a significant epithelial component demonstrated intense nuclear and cytoplasmic immunoreactivity with anticalretinin MoAbs. Thus, the sensitivity of calretinin immunostaining of biphasic mesotheliomas with an epithelial component and pure epithelial mesothelioma is 100%. Other investigators also have described similar nuclear and cytoplasmic calretinin staining patterns in the mesothelial cells in both surgical and cytologic specimens.23, 27, 28 Barberis et al. demonstrated that immunoreactivity of calretinin observed in paraffin embedded cell blocks correlated with the findings in the corresponding cytocentrifuged preparations.28
Heat-induced epitope retrieval and incubation overnight with the primary antibody were part of the current study staining protocol with anticalretinin. Two other studies that reported a positive result and the presence of both nuclear and cytoplasmic staining with anticalretinin also described these two steps in their protocol.23, 28 However, the use of heat-induced epitope retrieval and the duration of incubation with the primary antibody were not specified in the study by Simsir et al., who reported a negative result and the presence of cytoplasmic staining alone with anticalretinin.29 In view of the results of the current study and other reports, the types of antibodies did not appear to affect the outcome because both polyclonal and monoclonal antibodies from various sources gave similar results.23, 28
Two cases of mesothelioma with negative staining contained only spindle-shaped neoplastic cells; one was a biphasic variant. Other investigators also have shown that the sarcomatoid component of biphasic mesotheliomas and of pure sarcomatoid mesotheliomas demonstrated less intense or altogether absent staining with anticalretinin.23, 24 Sarcomatoid mesotheliomas are distinguished readily from metastatic adenocarcinomas based on cytologic morphology. Immunoreactivity of sarcomatoid mesotheliomas with epithelial markers such as cytokeratin aids in the differential diagnosis with the majority of spindle cell sarcomas.46 The only exceptions are synovial sarcoma and keratin positive leiomyosarcoma. Both are differentiated from mesothelioma readily because keratin immunostaining is focal in the former but diffuse in the latter.46 Theoretically, both keratin and calretinin immunostaining are not able to distinguish a metastatic spindle cell carcinoma from a sarcomatoid mesothelioma. In such instances, a detailed clinical history and thorough radiologic imaging are helpful in the differential diagnosis.
The reported rate of incidence of calretinin positive adenocarcinomas ranged from 10–48%.23, 26, 28, 29 In the current study, 12 of the metastatic adenocarcinomas (57%) demonstrated immunostaining to an anticalretinin MoAb. Similar to the observations made by other investigators,23, 26, 28 the staining of adenocarcinoma cells was focal and less intense when compared with that of reactive and neoplastic mesothelial cells. More important is the staining pattern displayed by different cell types, which to our knowledge has not been emphasized previously. Adenocarcinomatous cells showed spotty, coarse granular cytoplasmic staining, whereas cells of mesothelial origin showed diffuse staining of both the nucleus and cytoplasm.
In 64% of mesotheliomas with positive calretinin immunostaining in the current study, intense membranous decoration also was noted, resulting in a “fried-egg” appearance. The latter observation, which to our knowledge has not been described in the literature, is found to be characteristic of cells of mesothelial origin because we also have noticed the “fried-egg” staining pattern in reactive mesothelial cells. The latter can serve as an internal positive control. The specificity of calretinin immunostaining with epithelial mesothelioma is excellent (100%) when the staining pattern is considered. Not all mesotheliomas demonstrated membranous decoration. Four patients with malignant mesothelioma had several specimens obtained at different time intervals. Although cytoplasmic and nuclear staining were noted in all specimens, membranous staining was noted in only some of the specimens from the same individual patient. We attributed the absence of membranous staining in some specimens to antigenic loss resulting from variation in fixation.
The results of the current study as well as the results of other studies demonstrate that calretinin is not effective in differentiating reactive mesothelial cells from malignant mesotheliomas because both show the same staining intensity and pattern with calretinin.23, 26, 28, 29
To our knowledge the biologic significance of calretinin expression in mesothelial cells and mesothelioma currently is unknown. The expression of calretinin in both reactive and neoplastic mesothelial cells suggests that calretinin does not function as an oncoprotein.24 Gotzos et al. have demonstrated increased expression of calretinin during the G1 phase of the cell cycle in WiDr cells.47 When treated with oligonucleotides against calretinin synthesis, these cells were blocked in the G1 phase and subsequently underwent apoptosis.48 Therefore, calretinin may play a role in the maintenance of the cell cycle of both reactive and neoplastic mesothelial cells and the prevention of apoptosis in these cells.
The findings of the current study have demonstrated that calretinin is a reliable marker for epithelial mesotheliomas with a characteristic nuclear and cytoplasmic staining pattern. Positive calretinin immunoreactivity requires the presence of diffuse and intense nuclear and cytoplasmic staining. The sensitivity and specificity of calretinin immunostaining as a marker for mesothelial cells was 100% when the staining pattern of various cell types was taken into consideration. When used in conjunction with the currently employed immunomarkers such as anti-CEA, B72.3, and Ber-Ep4, which recognize molecules expressed by adenocarcinomas, anticalretinin immunostaining is a valuable adjunct in the differential diagnosis of malignant mesotheliomas versus metastatic adenocarcinoma in cytologic preparations.