Year : 2009 | Volume
: 52 | Issue : 2 | Page : 175--181
Immunohistochemical distinction between mesothelial and adenocarcinoma cells in serous effusions: A combination panel-based approach with a brief review of the literature
Paari Murugan1, Neelaiah Siddaraju1, Syed Habeebullah2, Debdatta Basu1,
1 Department of Pathology, Jawaharlal Institute of Postgraduate Medical Education and Research, Pondicherry, India
2 Department of Obstetrics and Gynecology, Jawaharlal Institute of Postgraduate Medical Education and Research, Pondicherry, India
8, Kesari Nagar Main Road, Adambakkam, Chennai
Background: The prognostic and therapeutic significance of differentiating adenocarcinoma (AC) from reactive mesothelium (RM) in effusions cannot be overemphasized. To avoid diagnostic errors, ancillary techniques like immunohistochemistry are employed. However, results vary and no universal standard has been accepted so far. Objective: To study the combined diagnostic efficacy of epithelial membrane antigen (EMA), carcinoembryonic antigen (CEA), E-cadherin (EC), calretinin (CAL), desmin (DES) and vimentin (VIM) in distinguishing RM from AC cells in serous effusions. Study Design: Unequivocally diagnosed cases of 39 adenocarcinomatous and 38 RM populations were studied using sections from 49 formalin-fixed, paraffin-embedded cell blocks. Materials and Methods: The immunomarkers were applied on cell block sections using the avidin-biotin peroxidase technique. The distribution/intensity of immunostaining in mesothelial and AC cells were graded semiquantitatively. Statistical Analysis Used: Fischer«SQ»s exact test was used to calculate the efficacy of individual markers and their combinations. Results: EMA was the best single marker for AC, with 100% sensitivity and 97.37% specificity. For the mesothelial cells, CAL exhibited 100% sensitivity and 92.31% specificity. DES was more specific than CAL but had a poor sensitivity of 55.26%. EC, CEA and VIM had unsatisfactory predictive values precluding their use as individual diagnostic markers. Among the combinations, two panels - EMA+ AND (CAL- OR DES-) for ACs and CAL+ AND (EMA- OR CEA-) for RM had 100% specificities and sensitivities. Conclusions: Most panel studies on fluid cytology are based on the arbitrary use of individual markers with the best statistical values, leading to a less than accurate diagnostic assessment. We believe that statistical parameters calculated in combination provide for a more practical and objective evaluation as well as allowing for meaningful comparative studies.
|How to cite this article:|
Murugan P, Siddaraju N, Habeebullah S, Basu D. Immunohistochemical distinction between mesothelial and adenocarcinoma cells in serous effusions: A combination panel-based approach with a brief review of the literature.Indian J Pathol Microbiol 2009;52:175-181
|How to cite this URL:|
Murugan P, Siddaraju N, Habeebullah S, Basu D. Immunohistochemical distinction between mesothelial and adenocarcinoma cells in serous effusions: A combination panel-based approach with a brief review of the literature. Indian J Pathol Microbiol [serial online] 2009 [cited 2020 Feb 28 ];52:175-181
Available from: http://www.ijpmonline.org/text.asp?2009/52/2/175/48910
The most common difficulty encountered by cytopathologists worldwide is the inability to separate without dispute the exfoliated atypical benign mesothelial cells from the metastatic cells of adenocarcinoma (AC) in effusions and the lack of a standard, accurate panel of immunomarkers as a diagnostic aid in solving the problem.
The conventional morphologic criteria, although standard and generally acceptable, fail far too often, not allowing a definite diagnosis in at least 15% of the cases , because the benign mesothelium undergoes myriad architectural and cellular alterations in reaction to numerous stimuli, while on the other hand, well-differentiated or borderline malignant cells can masquerade as benign ones.  It cannot be overemphasized that there is no single architectural, nuclear or cellular parameter that can reliably and reproducibly differentiate between the two. ,,
Therefore, the use of ancillary methods is mandated in all but unequivocal cases  and over the last decade it has become clear that of all the methods available, immunochemical stains are superior in the diagnostic workup of effusion cytology. , The eternal uncertainty, however, is in deciding the best panel of markers to be used. There is no dearth of literature pertaining to this problem and a consummate amount of research has been carried out, with, more often than not, the virtues of markers extolled in one study being countered in the next. Numerous panels have been suggested in the recent past in addition to the already existing ones. Unfortunately, however, none so far has been accepted as a universal standard.
Additionally, and perhaps of greater significance is the fact that very few studies have evaluated the combined predictive values of the epithelial and mesothelial markers, an approach quintessential to deciding on a practicable panel suited for day-to-day diagnostic application.
The objectives of the present study were to evaluate the combined efficacy of six commonly used markers - epithelial membrane antigen (EMA), carcinoembryonic antigen (CEA), E-cadherin (EC), calretinin (CAL), desmin (DES) and vimentin (VIM) in distinguishing between the reactive mesothelial (RM) and adenocarcinoma (AC) cells in serous effusions and to arrive at a panel best suited for the purpose.
Materials and Methods
Forty-nine body cavity fluid samples from 43 patients sent to the cytopathology laboratory of the Department of Pathology constituted the material for the present study.
Samples from the peritoneal cavity were most frequent, representing 34 of the studied cases, whereas 14 were from the pleural and one from the pericardial cavity. A relevant histopathological diagnosis was available for 17 of the 31 patients with peritoneal effusions. In six of these 31 patients, the diagnosis was unknown either histopathologically or clinically whereas eight had a suspected ovarian carcinoma, although a tissue diagnosis was not made. In the patients with pleural effusion, seven out of the 11 did not have a diagnosis, as was the case with the patient with pericardial effusion. Thirteen cases had histopathologically proven metastatic AC at the site of effusion.
For each case, smears were made and stained using standard May Grunwald-Giemsa and Papanicolaou techniques. The remaining sediment was formalin fixed and processed as cell blocks using the fixed sediment or the plasma-thrombin method, the latter used as and when required.
The cases having an unequivocal diagnosis were chosen after a thorough independent evaluation by two of the authors on the smears as well as the hematoxylin and eosin (H and E)-stained cell block sections. Accordingly, of the total 49 cases, there were 39 with AC while 38 had RM cells. Twenty-nine cases had distinct groups of both populations of cells. One case, which showed only normal mesothelial cells, was excluded from the study. Compared with the smears, the cell blocks identified 16 more cases in which RM cells were present, although the two paralleled each other as far as the ACs were concerned.
Immunohistochemistry was performed on 3-µm cell block sections using the streptavidin-biotin peroxidase complex technique. The staining was performed manually and antigen retrieval was accomplished by microwave heating in 0.001 M anhydrous citric acid buffer solution, pH 6.0. The details of the antibodies used and the timing employed for the heat retrieval are summarized in [Table 1]. Appropriate positive and negative controls were also set up as recommended.
The two authors independently graded, on a semiquantitative basis, the percentage of cells stained and the intensity of staining for each of the cases. Scores of 0, 1, 2 and 3 were recorded when 0, 50 percent of the cells stained respectively and intensity scores of 0, 1, 2 and 3 were given when there was none, mild, moderate and strong staining, respectively. A sum of the percentage and intensity scores was taken as the final IHC grade of the marker and a cut-off grade of ≥4 was employed for the epithelial markers staining mesothelial cells and vice versa for the statistical analysis of the individual antibodies and their combinations. A systematic evaluation of the predictive potential of all possible two marker combinations was performed. This included the presence of two epithelial or two mesothelial markers, an epithelial and a mesothelial marker, either of two epithelial or of two mesothelial markers and either of an epithelial or a mesothelial marker. Further, three marker combinations based on the results obtained from the study of the individual markers and the two marker combinations were selected. These three marker combinations were performed using both the AND/OR variables and the best among the lot were chosen as part of the final panel. The best combinations were ranked in the order of highest specificities, with highest likelihood ratios used in case the specificities were equal and the Youden's J index employed in cases where the likelihood ratio could not be derived. A P -value of P -value for the individual IHC markers and their combinations.
The morphology on the H and E-stained cell block sections was considered as the gold standard while evaluating the IHC staining.
All cases of AC stained positive for EMA with a grade of 6 [Figure 1]. Of the 38 cases of RM, two were positive for the marker, with one showing only weak focal staining. The staining for EMA in AC was predominantly membranous and cytoplasmic, with 74% of the cases showing this pattern. However, the membranous staining was relatively accentuated in some cases. In two of the cases, a cytoplasmic staining was noticed in the absence of membranous stain.
Ninety-seven percent of the AC cases stained for EC, with 24/38 (63%) cases exhibiting maximum staining scores. As many as 34% of the RM cases were also positive, with 12/13 cases exhibiting ≥grade 4 staining [Figure 2]. EC stained remarkably well around individual cell membranes, producing a net-like appearance in both the malignant and the reactive cells. Eighteen percent of the AC cases exhibited exclusively cytoplasmic staining whereas 53% stained for both cytoplasm and membrane. Cytoplasmic staining was also noticed in two of the 13 RM cases while the rest had a membranous pattern.
Ninety-two percent of the AC cases stained positive for CEA, with 19 of the 36 cases showing grade 6 staining. CEA marked 12 out of the 38 RM cases, with nine showing staining stronger than 3. The polyclonal marker also showed intense positivity in the macrophages and neutrophils [Figure 3]. The staining pattern of CEA was both cytoplasmic and membranous in 56% of the ACs whereas 25% of the cases showed a membranous marking. The cases of RM which took up the stain also followed similar patterns.
Thus, EMA proved to be the best single marker for AC, with 100% sensitivity and 97.37% specificity. EC and CEA were hampered by poor specificity, although sensitivity was quite impressive. The likelihood ratio of EMA was 38 and stood apart from the other two by a considerable margin.
CAL stained positive for all cases of RM, with 82% showing strong grade 6 positivity [Figure 4]. Only one case had a weak staining of grade 2. Eighteen percent of AC cases stained for the marker but only three were stronger than grade 4 [Figure 5]. Most cases of RM took up stain for CAL in the typical nuclear and cytoplasmic pattern, with membranous accentuation producing what is popularly known as the "fried egg" appearance [Figure 4]. Some cases did not reveal the membranous accentuation although nuclear and cytoplasmic staining was clearly made out. Five of the seven AC cases showed a cytoplasmic positivity while two stained similar to the mesothelium, with nuclear and cytoplasmic positivity.
VIM stained 74% of the RM cases with grade 6 staining seen in 20 of the 28 (71%) cases. However, 28% of the AC cases also took up the marker, of which eight were strong enough to be counted significant. VIM also had a tendency for strongly staining the lymphocytes.
DES stained positive for only 55% of the RM cases but proved to be the most specific of the trio of mesothelial markers with a mere 10% of AC cases taking up stain. Furthermore, when a cut-off of ≥4 was applied, the specificity doubled. VIM and DES predominantly stained the cytoplasm in the RM and the positive cases of AC. However, interestingly, 2/11 and 1/4 positive cases of AC stained exclusively on the membrane for VIM and DES, respectively. One case of AC and two cases of RM stained for VIM on both the membrane and the cytoplasm.
Thus, CAL was the best single marker for the mesothelial cells, exhibiting 100% sensitivity and 92.31% specificity. DES proved to be more specific than CAL but had a poor sensitivity of 55.26%. However, its likelihood ratio was almost as good as that of CAL. VIM had mediocre predictive values, with sensitivity and specificity in the low seventies.
[Table 2] shows the specificity and sensitivity of the individual markers.
Of the two marker combinations, EMA+AND CEA+ proved to be 100% specific for AC. The combination of EMA+ AND EC+ was also highly efficient, with a specificity of 97.37% and a sensitivity of 97.44%. Of the positive epithelial and negative mesothelial group, the combination of EMA+ AND DES- showed the best specificity (100%) and sensitivity (94.87%) for AC. EMA+ AND CAL- AND EC+ AND CAL- showed nearly similar effectiveness, with 100% specificity and 92.31% sensitivity. Hundred percent specificity was also achieved with CEA+ AND CAL-.
The combination of either of the two positive epithelial markers for AC proved to be least useful while the combination of a positive epithelial or a negative mesothelial marker produced a high specificity (97.37%) and 100% sensitivity with EMA+ OR CAL-. EMA+ AND (CAL- OR DES-) proved to be the best of the triple marker combinations for AC, with 100% specificity and sensitivity. The other such combinations, most of them involving EMA and/or CAL, also showed high specificities and sensitivities with J indices of >90.
In the mesothelial category, the low sensitivity of DES and VIM brought down the efficacy of CAL. However, the combination of CAL+ AND VIM+ and CAL+ AND DES+ proved to be 100% specific.
When combining a positive mesothelial marker and a negative epithelial marker for RM, the combination of CAL+ AND EMA- had a high specificity and sensitivity of 97.37% and 100%, respectively. The other combinations in this category, CAL+ and CEA- AND VIM+ AND EMA- were highly specific although not as sensitive.
Like its epithelial counterpart, the combination of either of two positive mesothelial markers for RM was not of much use. The combination of positive mesothelial or negative epithelial markers for RM, however, revealed high (>90%) specificities and 100% sensitivities. DES+ OR EMA- with a likelihood ratio of 19.500 proved to be the best, followed by CAL+ OR EMA- and CAL+ OR EC-, which had identical likelihood ratios of 13.000.
Like EMA+ AND (CAL- OR DES-) for the ACs, the triple marker combinations for mesothelial cells yielded the combination CAL+ AND (EMA- OR CEA-), with 100% specificity and 100% sensitivity. Similar results were obtained with CAL+ AND (DES+ OR EMA-). The two other combinations in this category with a high efficacy were CAL+ AND (VIM+ OR EMA-) and CAL+ AND (EC- OR EMA-), showing a J index of 97.37 and 94.81, respectively. [Table 3] lists the best panels after analysis of the combined predictive values.
The overwhelming prognostic implications and therapeutic challenges involved when a patient is diagnosed with the presence of malignant cells in serous effusions justify the continuing need for refinement of the existing diagnostic procedures and protocol.
Immunohistochemistry has assumed an undisputed role as the most commonly used ancillary method for the purpose of differentiating benign mesothelial from malignant cells in effusions.  An extensive body of literature advocates the use of selective commercially available monoclonal antibodies in the workup of problematic serous effusions.  Bedrossian, in 1998,  had summarized the use of epithelial markers in serous effusions in 65 selected publications from the literature of 1980-1995. A review of the panels suggested after 1995 as per a MEDLINE search is summarized in [Table 4]. ,,,,,,,,,,,,,,,
As inferred by other authors, , our study showed that no single marker on its own is capable of predicting the presence or absence of either benign or malignant cells with 100% accuracy and consistency, although EMA proved to be highly efficient. In such a situation, it becomes imperative that the final interpretation should be based on the combined efficacy of the markers so as to maximize the predictive potential. Accordingly, it has been suggested that a minimum of two markers should be selected as the expression of antigens in metastatic malignancies is often heterogenous. Moreover, as we fully appreciated it during the course of our study, the increased information obtained when additional appropriate antibodies were used was much more than simply additive because the different results confirm and control each other.  Pereira et al. have quite prudently stated that the appropriate panel should include both positive and negative markers.  Further, they opine that an ideal panel would include at least two positive and two negative markers for each possible diagnosis as many neoplasms show overlapping immunoreactivity.
In our study of the double markers, the combination of EMA+ AND DES- showed the best specificity (100%) and sensitivity (94.87%) for ACs, while in the mesothelial category, the combination of CAL+ AND EMA- had a high specificity and sensitivity of 97.37 and 100%, respectively. The triple marker combinations unraveled two panels (comprising markers of both cell types), which proved to be 100% specific and sensitive. These were
1. For ACs:
1.1. EMA+ AND (CAL- OR DES-).
2. For RM:
2.1. CAL+ AND (EMA- OR CEA-).
The other triple marker combinations, most of them involving EMA and/or CAL, also showed very high specificities and sensitivities, with J indices of >90.
We consider it fallacious to give a final report based on the specificity of a panel that does not include the makers for the other group of cells because it would be disastrous to err on a positive diagnosis of AC. Therefore, as pointed out by Pereira et al. , this necessitates the mandatory use of a panel that consists of both epithelial and mesothelial markers.  Accordingly, although the combinations EMA+ AND CEA+, CAL+ AND VIM+ AND CAL+ AND DES+ were 100% specific for either ACs or RM, it would be prudent not to use these panels for diagnostic purposes. This is also true of the other combinations with high predictive values like EMA+ AND EC+. Furthermore, even though the triple combination CAL+ AND (DES+ OR EMA-) was 100% specific and sensitive for RM, it was not considered suitable because it could give rise to a double mesothelial panel with an absent epithelial marker.
Beginning 1980, till date, at least 52 reports have been published on the subject of marker panels (two or more) in effusion fluid diagnosis. Among these, a mere 10 speak of interpretation based on the evaluation of both epithelial and mesothelial markers. Moreover, significantly, only four have actually evaluated the combined predictive values of the panels while the rest of the reports, despite promulgating a panel-based approach, have not gone beyond accounting for individual marker specificity and sensitivity. The panels that have been suggested in these studies are based on the arbitrary use of individual markers with the best statistical values. We feel that this is a rather simplistic and inaccurate way of looking at the issue and the use of the term "combination of markers" is in fact misleading unless the statistical parameters are calculated in combination as has been done in this study and four others, as illustrated in [Table 5].
Lee et al.  had evaluated four markers-CEA, EMA, fibronectin and cytokeratin (CK). Even though they had not performed a combined evaluation, it was possible to extrapolate results because in their study, fibronectin had the unique distinction of a marker that was 100% specific and sensitive. This, along with the fact that neither EMA nor CEA was reported to stain the RM, confirmed, even without availability of the full data set, that the combination of EMA+ and F- had a specificity of 100% and a sensitivity of 92% for AC and the combination of F+ and EMA/CEA- had the all encompassing 100% specificity and sensitivity for RM. Thus, it appears that most authors, as exemplified in the case depicted, had either ignored or overlooked vital interpretation, which could have made these studies more meaningful and thereby useful for practical application.
Further, the only study (other than the present) to have evaluated a combination of more than two epithelial and mesothelial markers for differentiating AC from RM was by Ko et al.  This was also the only other study to have used the OR factor in the distinction, which proved to be vital in enhancing the specificity and sensitivity of the panels. These authors had derived CAL- AND (CEA+ OR B72.3+), with a specificity of 100% and a sensitivity of 79%, as the best immunopanel among all combinations in their study.
Thus, to the best of our knowledge, ours is the first study in literature to shed light on a panel of markers for both AC and RM, consisting of epithelial and mesothelial antibodies, having the ideal specificity and sensitivity in differentiating between the two cell types.
With claims of new, improved immunomarkers and related technology flooding the literature, there is a real and constant threat of the so-called "older" ones being buried under the avalanche. In the haste to move on, coupled with pressure from commercial forces promoting the expensive "latest," the untapped potential of existing antibodies is often in danger of being side stepped. While conceding that our observations need validation by other laboratories and studies on a larger scale, we do hope that the results will serve as a reminder of this fact, for, as aptly quoted by Bedrossian,  " .... what is old and what is new if not the perspective and perhaps the fancy of the beholder?"
|1||To A, Coleman DV, Dearnaley DP, Ormerod MG, Steele K, Neville AM. Use of antisera to epithelial membrane antigen for the cytodiagnosis of malignancy in serous effusions. J Clin Pathol 1981;34:1326-32.|
|2||Singh HK, Silverman JF, Berns L, Haddad MG, Park HK. Significance of epithelial membrane antigen in the work-up of problematic serous effusions. Diagn Cytopathol 1995;13:3-7.|
|3||Davidson B. Ovarian carcinoma and serous effusions: Changing views regarding tumor progression and review of current literature. Anal Cell Pathol 2001;23:107-28.|
|4||Naylor B. Pleural, peritoneal and pericardial fluids. In: Bibbo M, editor. Comprehensive Cytopathology. 2 nd ed. Philadelphia: WB Saunders Co; 1997. p. 551-621.|
|5||DeMay R. Fluids. In: DeMay R, editor. The art and science of cytopathology. Chicago: American Society of Clinical Pathologists; 1996. p. 257-325.|
|6||Pereira TC, Saad RS, Liu Y, Silverman JF. The diagnosis of malignancy in effusion cytology: A pattern recognition approach. Adv Anat Pathol 2006;13:174-84.|
|7||Saleh H, Bober P, Tabaczka P. Value of Ki67 immunostain in identification of malignancy in serous effusions. Diagn Cytopathol 1999;20:24-8.|
|8||Leiman G. The cytopathology of serous effusions: The cavalry has arrived-with reinforcements. Curr Diagn Pathol 2001;7:123-30.|
|9||Bedrossian CW. Special stains, the old and the new: The impact of immunocytochemistry in effusion cytology. Diagn Cytopathol 1998;18:141-9.|
|10||Lee JS, Nam JH, Lee MC, Park CS, Juhng SW. Immunohistochemical panel for distinguishing between carcinoma and reactive mesothelial cells in serous effusions. Acta Cytol 1996;40:631-6.|
|11||Jensen ML, Johansen P. Immunocytochemical staining of smears and corresponding cell blocks from serous effusions: A follow-up and comparative investigation. Diagn Cytopathol 1996;15:33-6.|
|12||Delahaye M, van der Ham F, van der Kwast TH. Complementary value of five carcinoma markers for the diagnosis of malignant mesothelioma, adenocarcinoma metastasis, and reactive mesothelium in serous effusions. Diagn Cytopathol 1997;17:115-20.|
|13||Schofield K, D'Aquila T, Rimm DL. E-cadherin expression is a sensitive and specific method for detection of carcinoma cells in fluid specimens. Diagn Cytopathol 2000;22:263-7.|
|14||Simsir A, Fetsch P, Mehta D, Zakowski M, Abati A. E-cadherin, N-cadherin, and calretinin in pleural effusions: The good, the bad, the worthless. Diagn Cytopathol 1999:20:125-30.|
|15||Motherby H, Nadjari B, Friegel P, Kohaus J, Ramp U, Bφcking A. Diagnostic accuracy of effusion cytology. Diagn Cytopathol 1999;20:350-7.|
|16||Davidson B, Risberg B, Kristensen G, Kvalheim G, Emilsen E, Bjεmer A, et al . Detection of cancer cells in effusions from patients diagnosed with gynaecological malignancies: Evaluation of five epithelial markers. Virchows Arch 1999;435:43-9.|
|17||Chhieng DC, Yee H, Cangiarella JF, Symmans WF, Cohen JM. Use of E-cadherin and CD44 aids in the differentiation between reactive mesothelial cells and carcinoma cells in pelvic washings. Cancer 2000;90:299-306.|
|18||Gupta RK, Kenwright DN, Fauck R, Lallu S, Naran S. The usefulness of a panel of immunostains in the diagnosis and differentiation of metastatic maqlignancies in pericardial effusions. Cytopathology 2000;11:312-21.|
|19||Ko EC, Jhala NC, Shultz JJ, Chhieng DC. Use of a Panel of Markers in the Differential Diagnosis of Adenocarcinoma and Reactive Mesothelial Cells in Fluid Cytology. Am J Clin Pathol 2001;116:709-15.|
|20||Lozano MD, Panizo A, Toledo GR, Sola JJ, Pardo-Mindαn J. Immunocytochemistry in the differential diagnosis of serous effusions: A comparative evaluation of eight monoclonal antibodies in papanicolaou stained smears. Cancer 2001;93:68 -72.|
|21||Politi E, Kandaraki C, Apostolopoulou C, Kyritsi T, Koutselini H. Immunocytochemical panel for distinguishing between carcinoma and reactive mesothelial cells in body cavity fluids. Diagn Cytopathol 2005;32:151-5.|
|22||Malle D, Valeri RM, Photiou C, Kaplanis K, Andreadis C, Tsavdaridis D, et al . Significance of Immunocytochemical expression of E-cadherin, N-Cadherin and CD 44 in serous effusions using liquid-based cytology. Acta Cytol 2005;49:11-6.|
|23||Ueda J, Iwata T, Ono M, Takahashi M. Comparison of three cytologic preparation methods and immunocytochemistries to distinguish adenocarcinoma cells from reactive mesothelial cells in serous effusion. Diagn Cytopathol 2006;34:6-10.|