Early diagnosis of malignant mesothelioma (MM) is urgently needed because life expectancies and treatment options are limited in advanced stages of the disease. Malignant mesothelioma often presents with recurrent hemorrhagic or inflammatory effusions, which might mask the incipient stages of the disease and thereby delay the diagnosis. Despite difficulties in recognizing the malignant cells present in those early effusions, they are often the first available biologic material for diagnosis. Therefore, awareness is needed, and efforts should be made to distinguish the malignant cells by well-defined morphological criteria, combined with ancillary methods.
To summarize the diagnostic criteria for the cytopathologic diagnosis of MM based on recently published guidelines and to evaluate the clinical utility of those criteria in clinical practice.
The guidelines for the cytopathologic diagnosis of epithelioid and mixed-type MM and data in recent literature constitute the sources of this review. Eighty-five epithelioid or mixed-type MMs diagnosed between 2004 and 2013 at the Department of Clinical Pathology and Cytology, Karolinska University Hospital, Huddinge (Stockholm, Sweden), were evaluated to determine the clinical utility of the criteria defined by the guidelines.
A conclusive diagnosis of MM can be obtained based on the criteria defined by the guidelines with high positive predictive value. When diagnosed in this way, subsequent therapy should be initiated without further delay. With the earlier diagnosis obtained by cytology, a better effect of chemotherapy can be expected, as shown by the longer overall survival in these patients compared with those with a histopathologic diagnosis.
Malignant mesothelioma (MM) is mechanistically linked to previous asbestos exposure,1,2 and it develops several decades after the initial exposure to asbestos fibers. Despite a ban on asbestos in Western countries, MM is still a global concern because of the widespread use of asbestos-like mineral fibers in the developing countries, and the long latency period between asbestos exposure and diagnosis.
ASBESTOS-INDUCED MOLECULAR CHANGES
The molecular oncogenic mechanisms elicited by asbestos on a cellular level are related to the physical properties of the fibers, which deform the cytoskeleton of mesothelial cells,3 interact with the mitotic spindle,4 and can interfere with chromosome segregation, leading to aneuploidy.5 Importantly, asbestos fibers induce chronic inflammation, and the release of reactive oxygen species generated by asbestos can damage DNA double strands.6,7 In addition, the hypoxic microenvironment of mesotheliomas8 may alter the DNA damage-repair pathways9 and reprogram the metabolism by switching from oxidative phosphorylation to anaerobic glycolysis.10
Malignant mesothelioma is characterized by chromosomal losses more frequently than gains, leading to alterations in several tumor-suppressor genes. Consequently, the major feature of MM is the loss of tumor-suppressor genes. Those genes are inactivated by deletions and mutations or by epigenetic changes.11–13 The accumulated loss and/or inactivation of multiple tumor-suppressor genes at chromosome bands 3p, 9p, and 22q has a critical role in the pathogenesis of MM, leading to loss of p16INK4A-p14ARF, located at 9p21; neurofibromatosis type 2 (NF2), at 22q12; and BRCA1-associated protein 1 (BAP1) at 3p21.31-p21.2. Asbestos can also induce the proto-oncogenes c-fos and c-jun, and in turn, those transcription factors may enhance cellular proliferation and could render cells more susceptible to subsequent mutations. Several investigators found jun located at chromosome band 1p32 to be amplified in MM.12–14
Massively parallel sequencing and genome-wide screening15–19 verified the earlier identified genetic alterations in MM15,20 but also added novel mutations to the spectrum of the mesothelioma-related molecular signature. Recently, 116 different aberrations were identified simultaneously in 42 patients with mesothelioma, of which, a median of 3 were potentially actionable in each patient.21 In mesothelioma, no single-driver mutation has, however, been found, but each patient has a unique setting of alterations, motivating an individualized choice of treatment.
EARLY DIAGNOSIS—A CLINICAL CHALLENGE
The prognosis of MM is poor, with a median survival ranging from 4 to 12 months, depending on histologic subtype.22 The earliest manifestation of MM is often an effusion with admixture of blood and inflammatory cells, which can mask the malignant cells and, consequently, delay the diagnosis. To reach a conclusive diagnosis, the cytopathologist must be aware of the spectrum of cellular and molecular alterations and recognize the malignant cells present in those early effusions, which are often the first available biologic material for diagnosis.
The clinical outcome of patients depends largely on the phenotype of the tumor. In pleural effusions, the sarcomatoid tumor components are not exfoliated. Modern diagnostic approaches, however, make an accurate diagnosis possible in most epithelioid and mixed-type MMs, involving a spectrum of adjuvant diagnostic methods. The positive predictive value of such a cytologic diagnosis is as high as of that obtained by histology23 and, therefore, provides a sufficient basis for initiation of treatment. As described in the recently published international guidelines,24,25 the diagnosis of MM cannot be based on routine morphology only but necessitates the use of ancillary techniques.
CYTOMORPHOLOGIC FEATURES OF MALIGNANT EFFUSIONS
The cytomorphologic features of a mesothelioma effusion are described in detail in the “Guidelines for the cytopathologic diagnosis of epithelioid and mixed-type malignant mesothelioma: complementary statement from the International Mesothelioma Interest Group, also endorsed by the International Academy of Cytology and the Papanicolaou Society of Cytopathology,”24 with a short summary below.
Highly cellular effusions should always raise the suspicion of a malignant process on routinely stained material. The diagnostic criteria include the presence of numerous papillary cell groups of varying size (Figure 1, a and b) combined with common cytomorphologic criteria, such as pleomorphic cells with enlarged nuclei, the presence of macronucleoli, and variability in structure of the nuclear chromatin.26–29 Some samples show clearly malignant characteristics, whereas, in other cases, it is difficult to distinguish those cells from reactive mesothelial proliferations. In still other cases, there is a substantial admixture of inflammatory cells, mainly lymphocytes, or peripheral blood, which might then mask the malignant cells. In those cases, it is often difficult to recognize the malignant condition and to reach a conclusive diagnosis without adjuvant methods or hemolyzing the sample. Sarcomatoid MMs will not exfoliate sufficient diagnostic cells, and it is not possible to diagnose them by effusion cytology. In cases in which malignancy cannot be established by morphology, the diagnosis should be achieved with ancillary analyses.
The routinely stained, clearly malignant specimen may be indicative of mesothelial tumor phenotype, but the definitive diagnosis of a MM necessitates the use of ancillary analyses, such as immunocytochemistry. A liberal use of immunocytochemistry is, therefore, recommended, whenever the sample is rich in mesothelial cells and cell groups.
The main cytologic findings that indicate a mesothelial origin for malignant cells have been long known30,31 and have recently been reevaluated.24,25,32 Typically, the cell groups show less cohesiveness, which creates openings or “windows,” some containing amorphous cores with an acidophilic staining reaction. Cellular findings include protrusions through the cell membrane (“blebbing”), the presence of organophilic squamoid cells when stained according to Papanocolaou, and a fine vacuolization in the perinuclear area when a May-Grünwald-Giemsa stain is used. The samples show frequent cell-in-cell configurations. In material stained with May-Grünwald-Giemsa, there is sometimes a reddish, granular background, and in those cases, it is also possible to see a red haze around some of the tumor cells. That haze corresponds to the presence of hyaluronan, a biomarker synthesized in the cell membrane of MM cells.
Whenever the effusion is rich in mesothelial cells or is suspicious for malignancy, the possibility of a MM should be explored by ancillary diagnostic measures. Those methods can differentiate malignancy from a reactive condition in highly cellular samples and demonstrate the mesothelial phenotype in less-differentiated malignant effusions. There are mainly 2 diagnostic challenges with a malignant effusion: (1) establishing the malignant condition, and (2) establishing the mesothelial phenotype of the tumor cells (Figure 2).
Malignant Versus Reactive Mesothelial Proliferation
The nuclear atypia in many MMs is quite bland, whereas, in reactive mesothelium—the so-called mesotheliosis—there can be considerable nuclear atypia. There are mainly 3 different adjuvant analyses that are helpful in establishing a malignant condition: immunocytochemistry, ploidy analysis by fluorescence in situ hybridization (FISH), and electron microscopy (EM).
Immunocytochemistry is often the first choice in this situation with the same antibodies recommended for histology.33,34 The analysis should preferably include 2 epitopes in favor of malignancy, as well as 2 epitopes excluding such a condition. Antibodies that have been suggested to indicate malignant proliferations are epithelial membrane antigen (EMA; Figure 1, d [red]), CD146, Imp3, and Glut1. The diagnostic value of different EMA preparations varies, with the E29 clone (DAKO, Copenhagen, Denmark) having superior characteristics. Although EMA reactivity that is accentuated in the cell membrane may also indicate a malignant mesothelial proliferation, EMA is merely an antibody of malignancy because it also labels other cancers.
Several antibodies have been proposed to indicate benign/reactive conditions and to distinguish them from malignant proliferations,35,36 among which, desmin and BAP1 are most often used. Mesothelial cells lose desmin expression (Figure 1, d [brown]) early during oncogenesis. Thus, the presence of desmin corresponds to reactive mesothelial proliferation, and desmin staining is negative in MM. More recently, nuclear reactivity to the BAP1 epitope has been recommended for a similar purpose because the corresponding gene is often being dysfunctional in MM. Various sensitivities have been published for those epitopes. Desmin reactivity is a sensitive marker when applied on unfixed cells. Formalin fixation is, however, harmful to the desmin epitope,37 and the use of BAP1 can be recommended when paraffin-embedded cellblocks are used.
At our laboratory, a double staining of alcohol-fixed cytospin material with EMA and desmin is often helpful when distinguishing a malignant condition from reactive mesothelial proliferation. A dominance of cells with distinct EMA positivity indicates malignancy, whereas preserved desmin reactivity is associated with benign mesothelial cells (Figure 1, d). Scattered reactive mesothelial cells can, however, bind the EMA antibody, and a MM effusion may also contain some benign, reactive mesothelial cells with preserved desmin reactivity. The diagnosis then depends on the dominant staining pattern, and when it is difficult to establish malignancy in that way, further adjuvant analyses are necessary.
An accurate way of establishing malignancy based on an effusion is the analysis of ploidy with FISH (Figure 3, a through c). The commercially available UroVysion test (Abbot, Wiesbaden, Germany), adopted for pleural effusions, is useful in most MM cases.38,39 The test contains centromeric probes for chromosomes 3, 7, and 17, together with a probe hybridizing to band 9p21, where genes for p14 and p16 are located. Two algorithms for interpretation are used: either (1) gains of at least 2 centromeric signals in each of four cells, or (2) loss of both 9p21 signals in 12 cells. Either of those findings indicates aneuploidy associated with a malignant condition. Care is needed, however, so that tetraploidization during reactive proliferation is not misinterpreted. When there are 3 to 4 signals from all 3 centromeric probes, then the 9p21 signals may be helpful: if those probes also show 3 to 4 signals, the cells are probably tetraploid, whereas when only the usual 2 signals are seen in the latter probe, the centromeric probes demonstrate chromosomal gain. When validating the analysis, using these criteria on mesothelioma effusions, aneuploidy can be demonstrated with a sensitivity of around 95%.38 A homozygous loss of the 9p21 signals may point toward a MM, although that is not specific for this tumor.
Other analyses, such as EM and biomarker analyses can also be used to distinguish benign and malignant conditions. This will be further described below.
Malignant Mesothelioma Versus Metastatic Adenocarcinoma
Once the malignant nature of the effusion is established, the next step in a diagnosis of MM is to demonstrate the mesothelial phenotype of the tumor cells. Immunocytochemistry is the first choice of diagnostic means, using the same epitopes as used for histologic sections.33,34 Antibodies indicating mesothelial lineage are calretinin (Figure 1, c and j [red]), podoplanin (D2-40; Figure 1, e), HBME1 (Figure 1, f), mesothelin (Figure 1, g), Wilms tumor protein 1 (WT1), and cytokeratin 5. EMA (Figure 1, d [red, with cell membrane accentuation]) indicates a malignant condition, whereas BerEp4 (Figure 1, c and j [brown]), carcinoembryonic antigen (CEA; Figure 1, h), MOC31, thyroid transcription factor 1 (TTF1; Figure 1, i), CD15, and sialyl-TN are usually negative in MM and positive in metastatic adenocarcinoma. Among different CEA antibodies, the DAKO monoclonal preparation (No. A 0115, Copenhagen, Denmark) has virtually no cross-reactivity to MM cells. The BerEp4 and MOC31 antibodies label different epitopes on the epithelial cell adhesion molecule complex. BerEp4 seems slightly more sensitive in unfixed material, whereas MOC31 is preferred after formalin fixation. The performance of different TTF-1 clones seem to vary. The clone 8G7G3/1 (DAKO, No. M3375), was used throughout the present study (Figure 1, i and k [brown nuclear staining]). TTF-1 specifically labels lung and thyroid cancers, whereas MMs are invariably negative. Ovarian origin is indicated by PAX8 and CA 125 double reactivity (brown and red, respectively; Figure 1, l).
DIAGNOSTIC ANTIBODY PANELS
Because no single antibody is sufficiently specific alone, diagnostic antibody panels are recommended. Positive markers indicating MM, and negative markers excluding the diagnosis, should be used in combination. The use of a minimum of 4 immunocytochemical markers—2 in favor of, and 2 against, the diagnosis—is recommended; but some of the immunoreactivities might show atypical reaction patterns. In those cases, the use of additional markers will provide the correct diagnosis, leaving only a smaller proportion of inconclusive cases.
INTEGRATIVE APPROACH WITH MOLECULAR ANALYSES AND OTHER ANCILLARY METHODS
In cases in which routine cytology, together with immunocytochemistry, is inconclusive, additional ancillary methods can be helpful, such as biomarkers detected in the effusion supernatant, either as a secreted product or from tumor-cell decay. Two such soluble biomarkers have been proven to provide diagnostic information: hyaluronan and mesothelin.40–45
Hyaluronan is a high–molecular-weight polysaccharide, which is synthesized in large amounts by MMs. It is a marker for MM, and diagnostic levels are found in about 60% of MM effusions.45,46 When demonstrated with a chromatographic process, its specificity for MM was very high. Currently, however, the analysis is often performed as an enzyme-linked immunosorbent assay with reagents that occasionally may cross-react to give false-positive results, particularly in effusions from bacterial pleuritis. It is, therefore, advisable to analyze the concentrations of both hyaluronan and mesothelin, and then combine the 2 measures in a simple logistic model,45 which will give the probability of the combination representing MM.
Mesothelin is a protein bound to the cell membrane. It is often produced in large amounts by MM cells. After its synthesis, mesothelin is cleaved in 2 fragments: the cell-bound C-ERC is the receptor for CA 125, and the secreted N-ERC fragment acts as a macrophage-potentiating factor. Both of these ERCs can be recovered in the effusion supernatant, reflecting the synthesis of mesothelin. Mesothelin is not specific for mesothelioma but may act as a marker for malignancy; the hyaluronan content shows the mesothelial phenotype of the tumor.
Malignant mesothelioma can also be diagnosed by EM, which was long considered the gold standard for this diagnosis. The ultrastructural findings are well known, and cells groups recovered from a MM effusion are suitable for this examination, providing the material is glutaraldehyde fixed as soon as it arrives at the laboratory. Typically, the MM cells form polymorphic groups, with highly irregular nuclei, pronounced nucleoli, and desmosomes (Figure 4, a through f). The nuclei tend to be lobulated in a way not visible in light microscopy. The most prominent finding distinguishing MM from adenocarcinoma is the long slender microvilli, devoid of glycocalyx, found in apical membranes. When such microvilli and membrane protrusions are found on the basolateral surface of the cells or in cytoplasmic vacuoles in so-called neo-lumina, the findings strongly indicate malignancy. Other findings indicating mesothelial lineage of the malignant cells are filaments, which often surround the nucleus, and cytoplasmic, coarse tonofilaments, the latter correlating to the expression of cytokeratin 5 and a finding of squamoid differentiation. A major drawback with such ultrastructural examination is the long time needed for plastic embedding. Working ultrathin sections can often not be obtained within 2–3 weeks, a time that may be too long for the clinical demand.
DIAGNOSIS OF MALIGNANT MESOTHELIOMA IN CLINICAL ROUTINES IN RELATION TO THE GUIDELINES
The diagnostic principles reflected in Figure 2 have been used in clinical routines at the Department of Clinical Pathology and Cytology, Karolinska University Hospital, Huddinge (Stockholm, Sweden) since the late 1990s and are in line with the guidelines.24 Between 2004 and 2013, 85 cases were diagnosed with MM at our institution. In 76 cases (89%), an effusion was examined in the cytology laboratory. Forty-seven of those cases fulfilled the cytologic criteria for MM, as defined by the guidelines, corresponding to an absolute sensitivity of 62% which corresponded with conclusive diagnoses only. Forty-four of the 47 cases (94%) were diagnosed with immunocytochemical criteria, as described above, whereas in 3 cases (6%), the diagnosis was based on EM of the cell pellet. In another 13 of the 76 cases (17%). the report was that of a suspicion of MM. The total sensitivity, including cases diagnosed as “suspicious for malignancy”, was 79%. Three of these latter 13 cases (23%) had incomplete adjuvant analyses. In still another 3 cases (23%), the cytology had indicated malignancy without defining the mesothelial phenotype of the tumor.
There was some variation regarding the choice of antibody panels during the 10-year period studied (Table 1). In total, 7 antibodies were used to support the MM diagnosis, and 4 were used to exclude that diagnostic alternative. Except for the 3 cases in which EM replaced immunocytochemistry, 2 to 7 (average, 4.3) antibodies were used to support the diagnosis, and 2 to 5 (average, 3) were used to exclude other diagnostic alternatives. The sensitivities of the individual antibodies to support or exclude MM are shown in Table 2. Principally, their modes of action differ. Calretinin, D2-40, HBME1, WT1, and cytokeratin 5 demonstrate mesothelial lineage, whereas EMA and mesothelin merely indicate malignancy. If the 2 diagnostic outcomes—MM versus carcinoma or reactive versus malignant mesothelial proliferation—remain unresolved, various additional analyses become important. The distinction from reactive mesothelial proliferation can often be resolved with ploidy analysis by FISH, whereas the mesothelial lineage of unequivocally malignant cells can be established by additional immunocytochemistry or biomarker analysis.
In our material, the sensitivity of calretinin to show mesothelial lineage was 98% (42 of 43 cases), whereas EMA reactivity was 95% (40 of 42). Among less commonly used antibodies, reactivity to mesothelin was found in 28 of 30 cases (93%); HBME1, 28 of 30 (93%); WT1, 22 of 27 (81%); D2-40, 20 of 22 (91%); and cytokeratin 5, 18 of 18 (100%). Among reagents excluding the MM diagnosis, BerEp4 was found in 5 of 42 (12%), CEA in 0 of 38 (0%), desmin in 0 of 24 (0%), and TTF-1 in 0 of 26 (0%). In 2 cases, desmin reactivity had originally been reported. Upon rereview, those samples only contained a few scattered desmin-positive cells, representing benign mesothelial cells, which can also be present in a malignant effusion. Dominance of desmin-positive mesothelial cells gives a negative predictive value of 100%.
The use of 2 antibodies in favor, and 2 excluding, MM can be sufficient, providing all reactions, together with their positive and negative controls, give unequivocal results. During the present study period, however, additional antibodies were considered necessary to establish the diagnosis. Other adjuvant analyses were also performed simultaneously, often to ensure the correctness of the diagnosis.
Ploidy analysis by FISH was performed to verify malignancy in 20 of the 85 cases (24%). Aneuploidy, in accord with the manufacturer's criteria, could be demonstrated in 14 cases (70%), and 7 cases (35%) had homozygous deletion of the 9p21 band. Although careful use ensures high specificity, the sensitivity may vary. The 70% of aneuploidy cases seen in this routine use of the analysis was less than what was found when validating the test (95%),38 the difference, however, was not statistically significant.
Electron microscopy of effusion cell pellets was performed in 30 of the 76 cases (39%), 24 of them (80%) showing ultrastructural features diagnostic for MM. Biomarker analyses were performed in 44 of the 76 cases (58%). The established logistic model45 provides a probability of MM based on obtained concentrations of hyaluronan and mesothelin. Such probability levels exceeding 0.95 were obtained in 34 of those 44 cases (77%). Thus, the cytologic diagnoses were supported by typical findings in either EM or biomarker analysis in 41 of 45 cases (91%). These adjuvant analyses improved the sensitivity for detecting MM in 5 of the 45 cases (11%).
In 23 of the 76 cases (30%), the cytologic diagnoses were verified by a simultaneous or later biopsy. All patients with the diagnosis based on effusion cytology showed or developed clinical and radiologic evidence of that malignancy. When using guideline criteria, together with a liberal use of adjuvant analyses, in 10 years of clinical routine the positive predictive value of a cytologic diagnosis fulfilling those criteria was 100%. In 8 of the 76 cases (11%), the findings were suspicious of MM but not conclusive. Two of those 8 cases (25%) could not be verified as MM, so the positive predictive value of such an inconclusive diagnosis was 6 of 8 cases (75%). Thus, when a MM can be diagnosed in this way, the diagnosis is reliable and provides, on its own, sufficient grounds for initiating therapy.
Because the effusion is often the first diagnostic material available in patients with MM, cytology provides a possibility of earlier diagnosis. An earlier diagnosis, based on the first effusion, allows a better effect from chemotherapy because early treatment has been shown to improve the clinical outcome.47 The cytologic diagnosis of MM can be obtained with a high positive-predictive value, applying the criteria defined in the guidelines23 in clinical routine. When diagnosed in this way, subsequent therapy should be initiated without further delay.
In our cohort, the mean survival after a diagnosis of MM by effusion cytology was 20 months, whereas for those diagnosed only by histology, it was 12 months. It thus seems that cases in which it is possible to provide a diagnosis by cytology, including epithelioid and mixed subtypes represent a subgroup of patients with a somewhat better prognosis. In combination with the earlier diagnosis, a better effect of chemotherapy can, therefore, be expected, providing therapy is initiated without delay in cases in which effusion cytology is diagnostic.
The authors have no relevant financial interest in the products or companies described in this article.
Presented in part at the Pulmonary Pathology Society Biennial Meeting; June 13–16, 2017; Chicago, Illinois.