Aims Peritoneal malignant mesothelioma (PMM) is an uncommon tumour, accounting for only 7–9% of all mesotheliomas in Japan. Differential diagnosis between PMM and primary peritoneal serous carcinoma (PPSC), a high-grade serous carcinoma, may be difficult, and separating reactive mesothelial hyperplasia (RMH) from PMM can be even more challenging.
Methods To help differentiate PMM from PPSC and RMH, we used immunohistochemistry to examine mesothelial-associated markers (calretinin, AE1/AE3, CK5/6, CAM5.2, D2-40, WT-1, HBME1, thrombomodulin), adenocarcinoma-associated markers (CEA, BerEP4, MOC31, ER (estrogen receptor), PgR, TTF-1, Claudin-4, Pax8), and malignant-related and benign-related markers (epithelial membrane antigen (EMA), desmin, GLUT-1, CD146 and IMP3), and FISH to examine for homozygous deletion of 9p21. We used formalin-fixed, paraffin-embedded blocks from 22 PMMs (M:F=18:4; subtypes: 16 epithelioid, 6 biphasic), 11 PPSCs and 23 RMHs.
Results Seventeen of the mesotheliomas (four PMM from women) were classified as diffuse, while five were localised. Calretinin was 91% positive in PMM, but negative in PPSC (specificity, 100%). BerEP4, Claudin-4 and PAX8 were all 100% positive in PPSC (specificities, 100%, 95% and 95%, respectively, for excluding PMM). For distinguishing PMM and RMH, sensitivity for EMA in mesothelioma was 68%, while for IMP3 and GLUT-1 it was 64% and 50%, respectively, all with high specificities. FISH analysis revealed homozygous deletion of the 9p21 locus in 11/13 PMMs, but in 0/11 RMHs.
Conclusions Calretinin and BerEP4 may be the best positive markers for differentiating PMM from PPSC. EMA, in combination with IMP3 and desmin, is useful, and homozygous deletion of 9p21 may be helpful, for differentiating PMM from RMH.
- ASBESTOS DISEASE
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Malignant mesothelioma is a highly malignant neoplasm. Although it is still an uncommon tumour, it has a slowly increasing incidence in Japan. According to the Japanese Ministry of Health, Labour and Welfare, the reported number of deaths from peritoneal malignant mesothelioma (PMM) was 98 in 2014.1 It has accounted recently for only 7–9% of all mesotheliomas in our country. Actually, the increase in the total number of mesotheliomas is due largely to the increase of pleural mesothelioma, which makes up about 79% of mesotheliomas.
The reported numbers of PMM were larger in the past than recently. One reason might be the possible inclusion of some malignancies of the ovary and peritoneum in the case of women. Takeshima et al, using pathological materials, reviewed 382 mesothelioma-death cases for 2003–2005 in Japan. The percentage of ‘probable/definite’ pleural cases in women was 59.2%, but for peritoneal cases in women it was only 22.2%.2 Histologically, it is difficult to distinguish PMM from primary peritoneal serous carcinoma (PPSC), a primary peritoneal tumour that resembles high-grade serous carcinoma of the ovary, but without origin in the ovary or extraovarian organs. Ordóñez3 concluded that Ber-EP4 and MOC-31 are the best negative mesothelioma markers for differentiating between epithelioid mesotheliomas and serous carcinomas, and that the best discriminators among the positive markers for mesotheliomas are D2-40 and calretinin. Recently, transcription factor PAX8 has been found to stain in almost all serous ovarian carcinomas, but to be positive in only 9% of PMMs, which displayed only focal and/or weak staining.4
Further, separating reactive mesothelial hyperplasia (RMH) from PMM can be an even more challenging task. Several markers have been proposed, including epithelial membrane antigen (EMA), p53 protein, desmin and P-glycoprotein5–8 Kato et al9 noted that a member of a family of glucose transporter isoforms, GLUT-1, was expressed in 100% of malignant pleural mesothelioma, whereas no reactive mesothelium cases were positive for that isoform. Shi et al10 found that whereas insulin-like growth factor II messenger RNA-binding protein 3 (IMP3) was exhibited by 73% of mesothelioma cases, it was undetectable in any benign reactive mesothelial proliferations. In cytological smears from pleural, peritoneal or pericardial effusions, CD146, a cell-adhesion molecule, was expressed in almost all malignant pleural mesothelioma cases, but its expression was undetectable in any reactive mesothelium cases.11
One of the most powerful methods for distinguishing benign from malignant mesothelial proliferations is the detection of homozygous deletion of the 9p21 locus within a cluster of genes that includes p16/cyclin-dependent kinase inhibitor 2A (CDKN2A).12–16 Chiosea et al,17 using fluorescence in situ hybridisation (FISH) analysis, demonstrated homozygous deletion of the 9p21 locus in 67% of pleural mesothelioma and in 25% of peritoneal mesothelioma. Recently, p16 FISH was reported to show better sensitivity and specificity than the GLUT-1 immunohistochemical marker in pathology and cytology specimens.18 This analysis has been applied to almost all pleural mesotheliomas, but fewer peritoneal cases have been studied so far. In order to differentiate PMM from PPSC and RMH, we used (A) immunohistochemistry to examine for mesothelial-related markers, adenocarcinoma-related markers, and malignant-related and benign-related markers, and (B) FISH to examine for homozygous deletion of 9p21.
Material and methods
The PMM specimens were collected in 10 different hospitals across Japan (table 1). Formalin-fixed, paraffin-embedded neoplastic tissue blocks were available for all cases, together with clinical information. All cases conformed to the guidelines for the pathological diagnosis of malignant mesothelioma published in 2012.19 Also studied were 11 PPSC cases and 23 cases of RMH of the peritoneum.
To extract asbestos bodies from formalin-fixed lung tissue in cases 1–3, 5, 8, 11, and 15, a modified Smith method was used (viz. digesting tissue completely in bleaching fluid (hypobromous acid, potassium hydroxide and surfactant) and collecting the solid residue on a membrane filter).20 This method allows extraction of the asbestos fibres and bodies that are visible under the phase-contrast microscope. The results are reported as numbers of asbestos bodies per gram of dry lung tissue.
Immunohistochemical staining procedure
The polymer-peroxidase method (EnVision; Dako ChemMate, Denmark) was applied to deparaffinised sections of PMM, PPSCs and RMHs. All antibodies were incubated overnight at 4°C. The monoclonal and polyclonal antibodies used are listed in table 2, together with the antigen-retrieval conditions.21 ,22
The degree of antibody staining in tumour cells was assessed as either negative or positive (>10% of tumour cells stained).
Fluorescence in situ hybridisation
Dual-color FISH analysis was performed using the Vysis LSI CDKN2A SpectrumOrange/CEP 9 SpectrumGreen probe (Vysis, Illinois, USA). Tissue sections on slides were deparaffinised and hydrated to distilled water, and examination was performed in the same way as in our previous study.21 Only individual and well delineated cells were scored. At least 50 cells were scored for each case. Each tumour was assessed on the basis of the average score and the maximum number of copies of the CDKN2A gene to the chromosome 9 copy number (CEP-9). Homozygous deletion was considered to be present if both CDKN2A signals were lost in at least 20% of nuclei and at least one signal for the CEP-9 probe was evident in each nucleus.
The sensitivity and specificity for a combination of individual markers were calculated with the assistance of an evidence-based decision-making tool (http://ilm.medicine.arizona.edu/EBDM/DTPC/calculator.html).
We obtained surgically resected or autopsy specimens from patients with PMM (22 cases), PPSC (11 specimens) and RMH (23 specimens).
The patients with PMM were 18 men and 4 women (mean age, 61 years; range, 23–90 years), and 9 of these patients had had previous occupational or environmental exposure to asbestos. Other clinical information is given in table 1.
We were able to detect asbestos bodies in extracts from formalin-fixed, non-neoplastic lung tissues obtained from the seven autopsy cases (table 1).
Macroscopically, 17 cases were classified as diffuse type, and 5 cases as localised. Fifteen cases out of 17 diffuse types had died, one was alive, and one had no follow-up information (table 1). Four cases out of the five localised types (##12, 13, 15 and 16) were alive at 15–105 months. Microscopically, 16 epithelioid and 6 biphasic types were studied (table 1). We subclassified epithelioid mesothelioma according to the 2015 WHO criteria.23 The identified patterns were four tubulopapillary, four solid, two trabecular, one adenomatoid and five deciduoid (figure 1).
We performed immunohistochemistry using 21 antibodies. There was no difference in staining between the male and females cases. Tumour cells of PMM were: (A) positive for AE1/AE3 in all cases, CK5/6 in 21 cases, calretinin in 20 cases, D2-40 in 17 cases and WT-1 in 15 cases (figure 1A), but (B) negative for CEA, BerEp4, MOC31, estrogen receptor (ER), PgR and TTF-1 in all cases (figure 1B; tables 3 and 4). Only one case of epithelioid mesothelioma was positive for Claudin-4. Among mesothelial markers, calretinin was at a high positive rate (91%) in PMM, while being negative in PPSC (figure 1C), with its sensitivity (91%) and specificity (100%) being very high. Among adenocarcinoma markers, BerEP4 was at a high rate (100%) in PPSC (figure 1D), while being negative in PMM, with its sensitivity (100%) and specificity (100%) being very high. Claudin-4 and Pax8 displayed membranous and nuclear positivity in all cases of PPSC, with very high sensitivity (100%) and specificity (95%) (table 4). Concerning Claudin-4, some cases of mesothelioma exhibited cytoplasmic positivity, but we designated only membranous staining as ‘true positive’. For differentiation of the PMM from RMH of the peritoneum, the sensitivity and specificity obtained for EMA in mesothelioma were 68% and 100%, respectively (figure 1E,F). For IMP3, sensitivity was 64% and specificity 95% (table 4).
Fluorescence in situ hybridisation
In formalin-fixed, paraffin-embedded samples, we used FISH analysis to identify demonstrated homozygous deletion of the 9p21 locus in 13 PMMs, 9 PPSCs and 11 RMHs. This analysis revealed homozygous deletion of the 9p21 locus in 11/13 of PMMs, but in none of either PPSCs or RMHs (figure 2). Both of the cases (#9 and #12) without detected homozygous deletion were alive (at 37 months after biopsy and 105 months after tumour resection, respectively). Sensitivity and specificity for 9p21 homozygous deletion in reference to a diagnosis of PMM over RMH were 85% and 100%, respectively (table 4).
In a recent European (Italian) study, asbestosis and serosal plaques were shown to be more frequent in PMM than in pleural mesothelioma cases.24 In the present cases, the ratio of men to women was 4.5:1, so men suffered more frequently from PMM than women. Further, nine of our cases of epithelioid and biphasic mesotheliomas had a history of asbestos exposure, but there was no asbestos exposure in the three cases (##12, 13, 15) for which information was available among the five cases with the deciduoid type of mesothelioma (table 1).
Macroscopically, the present mesotheliomas of the localised types (two with biphasic and three with deciduoid growth pattern) had a much better prognosis than the diffuse ones. Microscopically, 16 epithelioid types, including 5 deciduoid ones, and 6 biphasic types were identified in our study, but no sarcomatoid ones. Actually, pure sarcomatoid tumours are very rare in PMM.25 Kadota et al26 classified malignant pleural mesotheliomas into five subtypes, according to the predominant histological pattern: trabecular, tubulopapillary, micropapillary, solid or pleomorphic. Patients in the combined trabecular and tubulopapillary subgroup had significantly better overall survival (23.3 months) than those with the solid subtype (13.7 months).26 Deciduoid mesotheliomas were divided by Ordóñez27 into two groups: high grade (loss of cohesion, marked nuclear atypia and high mitotic activity) and low grade (cohesive, less pleomorphic and low mitotic activity). The high-grade group of cases exhibited shorter survival (mean, 7 months) than the low-grade group (mean, 23 months). Ordóñez27 concluded that the differences in prognosis reported among deciduoid mesothelioma cases were due to the existence of a high-grade subgroup that exhibited highly aggressive clinical behaviour. The three of our deciduoid cases (##12, 13 and 15) still alive at 15 months, 37 months and 105 months after operation were of the localised type, with or without nuclear atypia (table 1). Borczuk et al28 noted that high mitotic rate and p16 loss were associated with increased risk of death in patients with peritoneal mesothelioma in multivariate analysis.
Many studies have attempted to differentiate between PMM and papillary serous adenocarcinoma. Attanoos et al29 observed that nuclear calretinin expression could be identified in most malignant mesotheliomas, but found no reactivity in serous ovarian and peritoneal carcinomas, giving 88% sensitivity and 100% specificity. They concluded that BerEP4 was a useful marker for distinguishing PPSC from PMM, with 95% sensitivity and 91% specificity for serous papillary ovarian adenocarcinoma. In two studies by Ordóñez:30 (1) reactivity for ER was evident in 88% of metastatic serous carcinomas of the ovary and in 86% of PPSCs, whereas none of the mesotheliomas expressed ER, and (2) BerEP4 and MOC-31 were found to be the best negative mesothelioma markers, while podoplanin, and calretinin were the best positive markers for differentiating between PMM and serous carcinomas.3 Recently, Ordóñez31 noted that PAX8 and Claudin-4 have a higher sensitivity and specificity for discriminating between peritoneal epithelioid mesotheliomas and serous carcinomas.
Here, we performed immunohistochemistry to try to differentiate PMM from PPSC. Calretinin was at a high positive rate (91%) in PMM, but negative in PPSC, with sensitivity (91%) and specificity (100%) both being very high. Among adenocarcinoma markers, BerEP4 was at a high rate (100%) in PPSC, but negative in PMM, with sensitivity (100%) and specificity (100%) both being very high. Several authors have recommended Ber-EP4 as one of the best markers for distinguishing between serous carcinoma and peritoneal mesotheliomas.3 ,29 ,32–34
For the differentiation of malignant mesothelioma from reactive mesothelial hyperplasia, malignant-related and benign- related markers have been used so far,5–10 but there has been little study of malignant mesothelioma and mesothelial hyperplasia in the peritoneum. Claudin-4 has been reported to exhibit strong reactivity in 57/58 serosal metastasis, and in 245/278 primary carcinomas, but to be negative in normal and reactive mesothelium, as well as in 82 mesotheliomas, based on biopsies and effusions.35 ,36 In our study, the sensitivity and specificity obtained for Claudin-4 were 100% and 95%, respectively. Dejmek et al8 reported in effusion cytology that EMA exhibited predominantly membrane-bound reactivity in 21 out of 36 mesotheliomas, but in only 1 out of 24 mesotheliosis. They also found EMA to be 79% positive at the cell membrane in epithelial or mixed mesotheliomas, but against 32% in adenocarcinoma metastases.37 In our immunohistochemical study, the sensitivity and specificity, for EMA in mesothelioma were 68% and 100%, respectively. For IMP3, sensitivity was 64% and specificity was 95%. Thus, EMA might be thought to be a good discriminating factor between mesothelioma and reactive mesothelial cells, except that its sensitivity is not high enough.
According to Chiosea et al, homozygous deletion of 9p21 in peritoneal mesothelioma was only 25%.16 Therefore, we analysed homozygous deletion of 9p21 in 10 PMM cases of the epithelioid type (including 3 deciduoid types) and 3 of the biphasic type, using FISH. Eleven out of 13 (85%) cases displayed homozygous deletion. The two cases (##9 and 12) without homozygous deletion were tubulopapillary and deciduoid types, and were alive after tumour biopsy and extirpation (table 1). All those with homozygous deletion, except case 18 (which was lost to the record), were known to be dead, but it was not detected in any PPSC (9 )cases or RMH (11) cases. Sensitivity and specificity, for 9p21 homozygous deletion were 85% and 100%, respectively. One possible reason for our high rate of 9p21 homozygous deletion is that our cases were all advanced, and almost all were autopsy cases. Dacic et al38 noted that homozygous deletion of p16/CDKN2A was seen in 60% of malignant pleural mesotheliomas, and that it was more frequent in cases with shorter survival (12 months), while longer survival (46 months) was negative for deletion. Krasinskas et al39 reported that patients with CDKN2A deletions and loss of p16 protein expression had worse overall and disease-specific survivals.
Asbestos analysis was performed on extracts from formalin-fixed non-neoplastic lung tissues obtained from seven patients with epithelioid or biphasic mesotheliomas, and five patients were found to exhibit more than 1000/g (dry lung), including four patients with more than 5000/g. These patients were presumed to have had occupational exposure to asbestos.
In summary, our data suggest that calretinin and BerEP4 might be the best positive markers for differentiating PMM from PPSC of the peritoneum. Although EMA displayed high specificity, it had a lower sensitivity than previously reported for differentiating PMM from RMH. Nevertheless, EMA, in combination with IMP3 and desmin, is useful, and homozygous deletion of 9p21 may be helpful for differentiating PMM from RMH.
Take home messages
Calretinin was positive in 91% of peritoneal malignant mesothelioma (PMM), but negative in primary peritoneal serous carcinoma (PPSC). BerEP4 was 100% in PPSC, but negative in PMM. Sensitivity and specificity were very high.
For the differentiation of PMM from reactive mesothelial hyperplasia (RMH) of the peritoneum, the sensitivity and specificity for epithelial membrane antigen (EMA) in mesothelioma were 68% and 100%, respectively.
Fluorescence in situ hybridisation analysis revealed homozygous deletion of the 9p21 locus in 11/13 of PMMs, but in none of RMHs.
Calretinin and BerEP4 may be the best positive markers for differentiating PMM from PPSC. EMA, in combination with IMP3 and desmin, is useful, and homozygous deletion of 9p21 may be helpful, for differentiating PMM from RMH.
The authors thank Dr Robert Timms for correcting the English version of the manuscript. The authors also thank Dr Shotaro Maeda (LSI Medience Corporation), Dr Ken Shimizu (Saitama Medical Center), Dr Takayuki Haga (National Hospital Organization East Saitama National Hospital), Dr Shigeki Yamada (Saitama Medical Center, Jichi Medical University), Dr Hisataka Uchima (Saitama Citizens Medical Center), Dr Toshihito Shinagawa (Kawasaki Municipal Ida Hospital), Dr Seiichiro Shimizu (Showa General Hospital), Dr Shigeo Yokoyama (Oita University Faculty of Medicine), Dr Kanako Kubota (Hokkaido University Hospital) and Dr Takuji Tanaka (Kanazawa Medical University), for providing cases.
Abstract in Japanese
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Abstract in Japanese - Online abstract
This work was presented in abstract form at the 102nd annual meeting of the United States and Canadian Academy of Pathology, Baltimore, Maryland, 6 March 2013
Handling editor Cheok Soon Lee
Contributors All authors have contributed to the planning, conduct and reporting of the work described in the article.
Competing interests None declared.
Patient consent Obtained.
Ethics approval Institutional review board of the National Defense Medical College.
Provenance and peer review Not commissioned; externally peer reviewed.