The detection of neoplastic invasion remains the linchpin for a clear diagnosis of malignant mesothelioma. Cytology-only diagnosis of epithelioid mesothelioma on aspirated effusion fluid remains controversial. A major problem is poor sensitivity, although cytodiagnosis is achievable in many cases at a high order of specificity, especially when a large volume of effusion fluid is submitted for cytological evaluation, enabling the preparation of cell-block sections for immunohistochemical investigation and when the cytological findings can be correlated with imaging studies to assess the anatomical distribution of the lesion and evidence of nodularity of the pleural disorder and, in some cases, to demonstrate evidence of invasion. Although ‘positive’ and ‘negative’ immunohistochemical markers have proved remarkably effective in distinguishing between epithelioid mesothelioma and secondary carcinoma and other malignant tumours metastatic to serosal membranes, no mesothelial marker has 100% sensitivity and specificity for mesothelioma diagnosis, so that panels of ‘positive’ antibodies and markers with negative predictive value are required. At present, no tissue or serum marker (including the molecular detection of p16/CDKN2A) has been proved to have sufficient specificity, consistency and reproducibility that it can replace evidence of invasion as the decisive marker for diagnosis when there is any uncertainty concerning a diagnosis of epithelioid mesothelioma and in the case of atypical fibrous lesions of the pleura (especially collagen-rich lesions, namely fibrous pleuritis vs desmoplastic mesothelioma), in which even the assessment of invasion can be problematical as illustrated in part 2 of this review.
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.
Despite bans on the importation and use of asbestos in many industrialised nations such as the European communities and Australia, the incidence of malignant mesothelioma (MM) continues to rise, for example in both the UK1 ,2 and in Australia3 ,4—in both men and women.1 ,3 During recent times, the proportions of asbestos exposure that underpin this increasing incidence have changed in Australia, from occupational exposures to an increasing proportion of non-occupational exposures related mainly to home maintenance or renovation exposures. In Western Australia (WA) during the period 2005–8 inclusive, Olsen et al5 found that home renovation exposures accounted for 8.4% of all men with MM and 35.7% of all women diagnosed with MM and, after adjustments for sex, and the year and age of diagnosis, the latency period for ‘home renovation’ MM appeared to be significantly shorter than for other exposure groups.
Many comprehensive accounts of the pathology of MM have been published, as well as detailed accounts of the uses of immunohistochemistry.6–11 Yet despite the advent of immunohistochemistry as an essential adjunct to the pathological diagnosis of MM, consistent and reliable diagnosis continues to be problematical in some cases, especially for small biopsies—and many anatomical pathologists emphasise the identification of neoplastic invasion as the definitive criterion for definitive diagnosis of MM as opposed to reactive mesothelial hyperplasia.10–15 Accordingly, Churg and Galateau-Sallé15 stated in 2012 that, after allowance for pseudoinvasive findings such as en face sections and benign entrapment of mesothelioma as a consequence of fibro-inflammatory processes:
‘… invasion of the stroma continues to be by far the most reliable criterion for separating benign from malignant mesothelial proliferations. Fat is the stroma most frequently encountered and the finding of mesothelial cells in fat makes the proliferation malignant unless there is an extraordinarily good reason to believe otherwise.i The same comment applies [even more so] to invasion of muscle or invasion of lung or another organ.’
This review (part 1) selectively addresses the assessment of cytology-only diagnosis, types of biopsy for MM diagnosis, unusual types of mesothelioma, and the role of immunohistochemistry for diagnosis and distinction between benign reactive mesothelial proliferations (BMP) and epithelioid MM. Consideration of well differentiated papillary mesothelioma and multicystic mesothelioma has been excluded from discussion. We have also excluded consideration of MM in situ and pleural epithelioid haemangioendothelioma. The discussion is not intended to be encyclopaedic, but instead is selective and concentrates on issues that have been highlighted in the recent literature on MM and its diagnosis. Unless stated otherwise, the discussion focuses on pleural MM in particular and on diffuse MM, although localised pleural MM are now well documented.16 We have avoided any discussion of the medicolegal ramifications of MM diagnosis, partly because of different requirements concerning the certainty of clinical versus pathological diagnosis for compensation in different nations—for example, definite or near-definite diagnosis (or ‘reasonable medical certainty’), as opposed to probabilistic diagnosis (MM ‘on the balance of probabilities’, MM ‘more likely than not’).
Cytology-only diagnosis of mesothelioma
Cytology-only diagnosis on effusion fluids remains controversial, and cytology in this setting cannot assess invasion, but it can suggest a diagnosis of pleural epithelioid MM at varying levels of confidence.9 ,17–21 Pointers to a diagnosis of MM include the profusion of the mesothelial proliferation, morular-papillary structures comprising 50 cells or more (especially in the pleura), cytological atypia, macronucleoli, frequent cytoplasmic vacuoles, and a single population of atypical cells to aid in the exclusion of secondary carcinoma.9 ,11 However, some cases of confirmed BMP can display considerable cytological atypia, whereas confirmed invasive epithelioid MM can be deceptively bland in appearance, with cytological overlap between the two.11–13 ,15 Other findings such as scalloped borders of cell clusters, intercellular fenestrations, variation in cytoplasmic staining and its ‘density’, and low nuclear-to-cytoplasmic ratios can be found in both epithelioid MM and reactive mesothelial hyperplasias.10 ,11 These remarks concern only MM with a significant epithelial component: diagnosis of sarcomatoid and especially desmoplastic MM is rarely achievable by cytology except for fine-needle aspiration biopsy (FNAB), and even those cases require close correlation with clinical and imaging data, in the present authors’ experience.
Although a cytology-only diagnosis of MM based on effusion fluid can be made at a high order of specificity in some cases, the main problem is one of low sensitivity, reported to be in the range of about 30–75%,11 ,22 and as low as 16% for a ‘positive’ diagnosis after the exclusion of ‘suspicious’ findings.23 In a study of 162 cases of MM, effusion fluid cytology showed high specificity (∼99%) when all criteria for diagnosis were fulfilled.17 Those criteria comprised: (1) cells with mesothelial features as caricatures of reactive mesothelial cells; (2) binucleated and multinucleated cells; (3) minute pink granules in the background in Giemsa-stained smears; (4) a single population showing a transition between benign-appearing cells and cytologically malignant cells; (5) large round nuclei with abundant cytoplasm; (6) tinctorial gradation of staining within the cytoplasm and evidence of microvilli at the cell periphery; (7) vacuolated cells, described by Welker et al17 as ‘foamy’; and (8) large and numerous clusters of mesothelial cells. However, the sensitivity was only 47.5% when not all criteria were fulfilled. The sensitivity improved to 71–91% by interpreting the cytological findings together with effusion fluid hyaluronic acid concentrations. Very rarely, intracytoplasmic rod-like or cylindrical crystalloids may be found (figure 1)6 ,24 and may impart a Gaucher-like appearance of the tumour cells in histological sections;9 by electron microscopy, such crystalloids represent complex scroll-like granular crystalloids within the endoplasmic reticulum,6 ,9 presumably representing polymerised glycosaminoglycan material such as hyaluronic acid, and so far as we are aware they have been reported only in epithelioid MM.
More recently, Segal et al21 in Perth, WA, reported a 20-year ‘audit’ on effusion fluid cytology diagnosis of MM, with ‘definitive diagnosis’ of MM in 377 out of 517 cases, yielding an ‘absolute’ sensitivity of 73%, as assessed by data linkage for 815 cases in the WA cancer and mesothelioma registries, and for whom cytological examination of pleural effusion fluid had been carried out (63%). A further 66 patients were diagnosed as ‘atypical/suspicious’ of MM resulting in a ‘complete’ sensitivity of 86%. When the analysis was confined to biopsy/necropsy-confirmed cases only, the ‘absolute’ sensitivity was 68% and the ‘complete’ sensitivity was 82%. There were only two ‘false’ positive diagnoses, representing cases of secondary carcinoma (breast and renal cell carcinoma; RCC), misclassified before the advent of ‘specific’ mesothelial immunohistochemical markers, and for one patient there was an interval of 14 years between the time of cytological diagnosis of mesothelioma and the patient’s subsequent presentation with metastatic MM. In 140 cases with confirmed MM (27%), cytological assessment was considered to have been negative to atypical/suspicious, and of these, 45 (32%) were found to have had a sarcomatoid MM. Although this ‘audit’ originates from a centre with acknowledged expertise in the cytodiagnosis of MM, we have some reservations at this time about the general applicability of cytology-only diagnosis, and the consistency and reproducibility of such diagnosis in centres with less extensive experience in the cytodiagnosis of MM. For example, Segal et al21 did not specify the criteria used for the cytology diagnosis of MM (probably not possible in retrospect because the reports over the 20-year period had presumably been generated by different cytopathologists)—only that a positive cytodiagnosis showed a near-definite correlation with subsequent data for diagnosis. In addition, the ‘audit’ did not (and presumably could not) state what ancillary information might or might not have been taken into account at the time of the cytology diagnosis (eg, concurrent biopsy findings; in addition, the frequency of MM in WA is influenced by the high rate of MM in former miners/millers in the Wittenoom crocidolite industry,25 so that information about this circumstance could potentially influence the confidence index for a cytology diagnosis).
It can be argued that a 27% rate of assessment of effusion fluid cytology specimens as negative to ‘atypical/suspicious’ is inadequate for prospective diagnosis in clinical practice, in the absence of other compelling evidence for this diagnosis. In such cases we routinely recommend establishment of the diagnosis by either a tissue biopsy adequate for histological and immunohistochemical assessment or, if biopsy is considered inadvisable or contraindicated, by serial imaging studies to ascertain whether the serosal (usually pleural) disease is progressive, with evidence of nodularity or extension into extraserosal tissues.26
We agree that when all of the cytological criteria for epithelioid MM are fulfilled in an effusion fluid specimen, a highly probable to near-definite diagnosis of MM can be ventured at a high level of specificity. Problematical cases are those for whom not all of the cytological criteria are fulfilled: for many such cases, the problem can be attributed to inadequacy of the effusion fluid specimen and thus low sensitivity. We recommend that, unless loculation of the fluid or other physical constraints prevent adequate sampling of the effusion fluid, at least 100 mL of effusion fluid and preferably the entire volume (after sampling of small volumes for biochemical and microbiological assessment) is submitted for cytology, to allow recovery of sufficient numbers of cells for cell block sections and immunohistochemistry studies. Even in these circumstances, we consider that detailed clinical-radiological-cytological correlation is required,26 and we continue to see many cases of MM in which effusion fluid cytology is interpreted as ‘reactive mesothelial cells’ shortly before a diagnosis of MM is established by biopsy. However, after aspiration of effusion fluid, most such cases in our institutions proceed to a video-assisted thoracoscopy (VAT) with pleural biopsy and talc pleurodesis, thereby facilitating diagnosis at a high order of confidence (this review excludes discussion of the relative merits of VAT-based talc pleurodesis vs bedside pleurodesis vs no-VAT and insertion of a pleural indwelling catheter).27–31
Choice of biopsy for MM diagnosis
Biopsy techniques for MM diagnosis have been reviewed by Walters and Maskell.23 In general, the confidence index for a biopsy diagnosis of MM is proportional to the volume of tumour sampled: Attanoos and Gibbs32 found that diagnosis of MM was attained in 75% of biopsies measuring over 10 mm in size, but only 8% when the biopsy was less than 10 mm in diameter. Several factors can affect the choice of, and priority for, different types of biopsy, including: any co-morbidities that would contraindicate procedures that are more invasive than others; radiological findings such as the presence of a pleural mass lesion amenable to a core biopsy;33–35 and established patterns of clinical practice at different medical centres.
‘Blind’ percutaneous pleural biopsy has low sensitivity for the detection of MM and is not recommended for ‘routine’ diagnosis.26 ,32 ,33 FNAB also has a low diagnostic yield (about 30%)36 and seems justifiable only when a mass lesion is present. Sterrett et al37 reported the FNAB results for 19 cases of MM, and found that a diagnosis of malignancy could be made in 17, with a diagnosis of MM in eight (42%) with a further four cases in which the findings were considered to be consistent with or suggestive of MM, and they considered that FNAB was suitable in particular for patients who did not have a pleural effusion at presentation.
CT-guided core biopsies have high sensitivity for the diagnosis of MM or other malignant neoplasms and low complication rates, with a diagnostic yield of up to approximately 87%,33 ,34 and are suitable for cases in which imaging studies have demonstrated pleural thickening or a nodular/mass lesion.34 ‘Standard’ VAT biopsy is suitable for other patients with a pleural effusion but no mass lesion, or patients for whom VAT-based talc pleurodesis is indicated, with a sensitivity of up to approximately 94% for the diagnosis of epithelioid MM.34 The 2010 guidelines from the European Respiratory Society (ERS) and the European Society of Thoracic Surgeons (ESTS) stated that thoracoscopy is the preferred technique and allows extensive inspection of the pleura and the taking of multiple and large biopsies that include subpleural tissue for the histological assessment of invasion.36 VAT is tolerated well in general, with a low complication rate. Flexible thoracoscopy under local analgesia or neurolept anaesthesia is used increasingly by respiratory physicians, with a diagnostic yield comparable to ‘standard’ VAT.38 In a review of seven post-2000 studies, Walters and Maskell23 found that local anaesthetic thoracoscopy gave a diagnosis of MM in 87.5–100% of cases.
Even so, open biopsy allows more accurate subtyping of mesothelioma:32 ,39 ,40 in a series of extrapleural pneumonectomy cases of MM, Kao et al41 reported that open biopsy correctly classified the subtype in 83%, in comparison to 74% for a VAT biopsy, and 44% for CT-guided biopsy. However, a 2004 WHO chapter on mesothelioma states that thoracotomy is not required for diagnosis, VAT being sufficient, and is best avoided because of the risk of ‘tumour implantation in the chest wall’.42 ‘Thoracotomy’ should probably be restricted to a small incisional biopsy into the chest wall for those cases in which the pleural space has been obliterated, so that VAT cannot be performed.
Immunohistochemical investigation for the differential diagnosis between epithelioid and biphasic mesothelioma versus non-mesothelial carcinomas
Immunohistochemistry is integral to the diagnosis of MM: at present, immunohistochemistry is considered to be the most useful ancillary procedure to establish a diagnosis of epithelioid MM or biphasic mesothelioma as opposed to non-mesothelial cancer, and has largely replaced electron microscopy for this purpose.9 A major advance in this field has been the development of antibodies with varying degrees of sensitivity and specificity for mesothelial cells, in addition to markers for carcinomas, most often adenocarcinoma and most often bronchopulmonary adenocarcinoma. Yet there is no known marker with 100% sensitivity and 100% specificity for mesothelioma (or adenocarcinoma), so that panels of mesothelial and carcinoma-related antibodies are required for either cytology cell-block sections or biopsy tissue.9 ,10 Mesothelial markers include calretinin (widely considered to represent the most useful and specific marker for mesotheliomas, provided that the result includes labelling of nuclei in addition to cytoplasm),9 Wilms’ tumour gene product, D2-40 (podoplanin antibody), mesothelin, cytokeratin 5 or cytokeratin 5/6, HBME-1 (which we continue to find valuable provided that the antibody is used at high dilution)9 and thrombomodulin.8–11 Carcinoma-related markers in widespread use include CEA, CD15, B72.3, Ber-EP4, MOC-31 and BG8, and for pleural mesothelioma in particular napsin A and/or TTF-1,9 ,11 and CDX2 for discrimination between MM versus gastrointestinal adenocarcinomas.43 The sensitivities and specificities for these markers/antibodies have been extensively reviewed,9–11 and are not revisited in this article. Most studies and reviews deal with the sensitivity/specificity of each individual antibody as opposed to clusters of antibodies using a dichotomous approach (logistic regression). In a retrospective tree-based regression analysis of 173 cases of pleural MM of epithelioid type and 27 cases of adenocarcinoma metastatic to the pleura, Klebe et al44 found that a panel of three antibodies (calretinin, BG8 and CD15) was sufficient to diagnose or to exclude epithelioid MM. Nonetheless, a panel that includes the following is widely recommended: a broad spectrum cytokeratin antibody cocktail such as AE1/AE3 or a label for low molecular weight cytokeratins such as CAM5.2 or MNF116 or cytokeratin 7; plus at least two mesothelial markers such as calretinin and WT1; plus at least two carcinoma-related markers, to include TTF-1 whenever adenocarcinoma of the lung is part of the differential diagnosis.8–11 Depending on the anatomical site of the serosal abnormality (eg, pleura vs peritoneum), gender, and the clinical circumstances and past medical history, other site-specific markers can be added, such as CD10, the RCC antibody and CD138 (syndecan-1) for a clear-cell tumour affecting the pleura, even if MM-like in distribution.11 Even so, CD10 expression occurs in up to approximately 50% of MM,11 and it does not clearly discriminate between a clear-cell epithelioid MM and RCC of clear-cell type—highlighting the importance of the clinical history and the findings on imaging studies. As another example of the diagnostic pitfalls that can be encountered, up to 15% of a subset of high-grade carcinomas of the breast, often BRAC1-associated, can express calretinin, and these carcinomas may also express cytokeratin 5/6 and lack detectable oestrogen receptor protein, with the potential for misdiagnosis of pleural metastases as epithelioid MM.45 ,46 The ERS/ESTS guidelines do not recommend the use of cytokeratin 7/cytokeratin 20 profiling for the diagnosis of mesothelioma,36 and in one study cytokeratin 20 expression by less than 10% of tumour cells was found in three of 14 cases of MM (approximately 20%).47
The International Mesothelioma Interest Group recommends that markers have sensitivity or specificity greater than 80% for the lesions in question,11 whereas the ERS/ESTS guidelines suggest a minimum sensitivity of 60–70%.36 Interpretation of positivity should take into account the localisation of the stain (eg, nuclear vs cytoplasmic labelling) and the percentage of cells stained: more than 10% has been suggested for cytoplasmic membranous markers.11
Within the framework of these recommendations, we consider that it is also important for each laboratory to decide on what panel of antibodies gives the most consistent and reliable results for discrimination between MM with an epithelioid component versus non-mesothelial neoplasms, and on variations to the panel for the circumstances of individual cases.9
Immunohistochemistry and p16/CDKN2A for the discrimination between MM and BMP
The role of immunohistochemistry for the distinction between MM with an epithelioid component versus a reactive BMP is reviewed in Husain et al,11 Klebe and Henderson14 and Churg and Galateau-Salle.15 Some investigators have found that strong circumferential immunolabelling of mesothelial cells for epithelial membrane antigen (EMA) is evidence in favour of MM as opposed to BMP,48–50 provided that the EMA antibody is based on the E29 clone.9 ,51 Other immunohistochemical markers investigated as indicators of MM include glucose transporter-1 (GLUT-1), insulin-like growth factor messenger RNA-binding protein 3 (IMP3), the X-linked inhibitor of apoptosis, p53 and bcl-2, as well as fluorescence in-situ hybridisation (FISH) for deletion of p16/CDKN2A (see table 1).9 ,15 ,49–63 In contrast, immunolabelling for desmin is claimed to be evidence in favour of a BMP.49 ,50 ,63
These immunohistochemical markers have been evaluated in effusion fluid cytology specimens in particular; apart from p16/CDKN2A, the findings in various studies are summarised in table 1.
There is evidence that homozygous deletion of p16/CDKN2A, as demonstrated by FISH, may be useful for the distinction between MM and BMP, with sensitivity and specificity in one study that were superior to immunolabelling for GLUT-1.55 For example, three studies have reported such deletions of p16/CDKN2A in 43–70% of pleural MM (mainly but not exclusively epithelioid MM), but not in reactive mesothelial hyperplasias.55 ,64 ,65 The p16 deletion was less frequent in peritoneal mesotheliomas than in pleural MM.55 ,64
From our review of the literature and our own experience, we conclude that: (1) it would be of interest to evaluate markers such as EMA, GLUT-1, IMP3 and FISH for p16/CDKN2A deletion, for distinction between MM and BMP, not as individual markers, but as a cluster using ‘tree’ regression analysis;44 and (2) there is insufficient evidence at present that these markers, either in isolation or in combination and including p16/CDKN2A, have shown sufficient specificity, consistency and reproducibility, to replace neoplastic invasion as the decisive marker for diagnosis when there is any uncertainty concerning a diagnosis of epithelioid MM or for atypical fibrous lesions of the pleura; and (3) none of these markers represents a decisive discriminator between MM and secondary carcinoma.
Serum biomarkers proposed for the detection or diagnosis of MM
We emphasise that the biomarkers mentioned in this review are of relevance to MM with an epithelioid component only: they appear to have no validity for the detection of sarcomatoid MM and other unusual subtypes of MM (see part 2). The serum biomarkers proposed include soluble mesothelin-related protein (SMRP),66–85 (and including SMRP in effusion fluid),70 ,76 osteopontin,74 ,80 ,81 ,86–91 megakaryocyte potentiating factor,74 ,77 ,79 ,90 ,92 ,93 CA125,9 ,71 and the recently-advocated marker fibulin-3.94–97 The usefulness of these markers is restricted by low sensitivity and specificity for MM and, in the case of fibulin-3, only a few reports have been published so far.94–97 None of these biomarkers has shown sufficient sensitivity, specificity or reproducibility either alone or in limited combination to replace existing imaging-cytology-biopsy requirements for diagnosis,14 ,84 ,98 ,99 and they do not at present fulfil the criteria for screening 'at-risk’ populations.14 ,84 ,100 Although SMRP, osteopontin and CA125 lack specificity as diagnostic markers,71 ,74 they may have value in monitoring the progression of MM or its response to treatment.14 ,99
Immunohistochemistry and the diagnosis of sarcomatoid tumours affecting serosal membranes
Immunohistochemistry plays a more restricted role for the diagnosis of sarcomatoid MM than for MM with an epithelial component, because many sarcomatoid MM express only cytokeratins in addition to vimentin and, in some cases, markers of smooth muscle differentiation.9–11 101–103 Expression of calretinin is variable (30–89%) in sarcomatoid areas of mesothelioma,8 ,9 ,42 ,104–107 and was usually focal, with labelling of less than 10% of cells in one series of 326 cases of sarcomatoid MM.107 The high percentage labelling recorded in some studies seems to be explicable by the acceptance of cytoplasmic labelling for calretinin as a positive result,106 whereas positive nuclear labelling is required in addition to any cytoplasmic staining.9 Positive labelling for calretinin (or with D2-40) in sarcomatoid tumours is most likely to be significant for a diagnosis of sarcomatoid MM when the tumour also expresses cytokeratins and when labelling of residual benign mesothelium and lymphatic endothelium can be excluded.11 Most sarcomatoid MM and desmoplastic MM show positive labelling for cytokeratins, 93% in one study of 326 cases,107 often widespread and diffuse but focal and weak in some cases,9–11 ,107 and cytokeratin-negative sarcomatoid MM do occur,9 ,107 and especially in areas of heterologous differentiation.11 Cytokeratin labelling can also highlight invasion, such as genuine invasion into subpleural fat by a desmoplastic MM9 ,15 (see part 2). The ERS/ESTS guidelines recommend the use of at least two broad-spectrum cytokeratin antibodies and two markers with negative predictive value, to support a diagnosis of sarcomatoid MM.36
An unequivocal biopsy diagnosis of MM requires the demonstration of neoplastic invasion by a mesothelial proliferation.
A cytological diagnosis of MM can be made at a high order of confidence when an atypical mesothelial proliferation correlates with the detection of invasion by imaging studies.
Both core biopsies and VAT-guided biopsies have a high yield for a diagnosis of MM, depending on the anatomical distribution of any pleural abnormalities including the presence or absence of a pleura-based mass lesion.
Immunohistochemistry is essential for cytological or histological investigation of a suspected MM with an epithelioid component and should include a cytokeratin marker and two or more mesothelial and two or more carcinoma-related markers.
Immunohistochemistry is highly effective for discrimination between MM with an epithelioid component vs secondary carcinoma.
At present, immunohistochemistry cannot replace neoplastic invasion for discrimination between MM and a BMP (including p16/CDKN2A).
The immunohistochemical repertoire of sarcomatoid MM is more restricted than for MM with an epithelioid component, and mesothelial cell markers are often negative.
Competing interests None.
Provenance and peer review Commissioned; externally peer reviewed.
↵i Such ‘extraordinarily good’ reasons include iatrogenic displacement of mesothelial cells along biopsy and other surgical wounds, and we would also comment that assessment of invasion need not always be histological: it can sometimes be identified by imaging studies.