Malignant mesothelium as a diagnostic aid in canine pericardial disease
Milne E., Martinez Pereira Y., Muir C., Scase T., Shaw D.J., McGregor G., Oldroyd L., Scurrell E., Martin M., Devine C., Hodgkiss-Geere H.Immunohistochemical differentiation of reactive from malignant mesothelium as a diagnostic aid in canine pericardial disease.J Small Anim Pract. 2018 May;59(5):261-271.
OBJECTIVES: To develop a provisional immunohistochemistry panel for distinguishing reactive pericardium, atypical mesothelial proliferation and mesothelioma in dogs.
MATERIALS AND METHODS: Archived pericardial biopsies were subject to haematoxylin and eosin staining, immunohistochemistry for cytokeratin, vimentin, insulin-like growth factor II mRNA-binding protein 3, glucose transporter 1 and desmin. Samples were scored for intensity and number of cells stained.
RESULTS: Ten biopsies of reactive mesothelium, 17 of atypical mesothelial proliferation, 26 of mesothelioma and five of normal pericardium were identified on the basis of haematoxylin and eosin staining. Cytokeratin and vimentin were expressed in all biopsies, confirming mesothelial origin. Normal pericardial samples had the lowest scores for insulin-like growth factor II mRNA-binding protein 3, glucose transporter 1 and desmin. Mesothelioma and atypical proliferative samples were similar to each other, with higher scores for insulin-like growth factor II mRNA-binding protein 3 and glucose transporter 1 than the reactive samples. Desmin staining was variable. Insulin-like growth factor II mRNA-binding protein 3 was the best to distinguish between disease groups.
CLINICAL SIGNIFICANCE: An immunohistochemistry panel of cytokeratin, vimentin, insulin-like growth factor II mRNA-binding protein 3 and glucose transporter 1 could provide superior information compared with haematoxylin and eosin staining alone in the diagnosis of cases of mesothelial proliferation in canine pericardium, but further validation is warranted.
Pericardial effusion is common in dogs and can lead to cardiac tamponade, collapse and death. The most common causes in dogs in the UK are idiopathic pericarditis (IP) and neoplastic effusion (Stafford Johnson et al. 2004, MacDonald et al. 2009). Malignant mesothelioma of the pericardial lining is the third most common neoplastic cause of pericardial effusion, after haemangiosarcoma and chemodectoma.
Surgical intervention with pericardiectomy in patients with recurrent idiopathic pericarditis is often successful in ameliorating clinical signs, whereas in dogs with mesothelioma, there is typically a persistent and recurrent pleural effusion, and the prognosis is poorer (Stepien et al. 2000). Canine mesothelioma is a neoplasm of mesodermal origin affecting the mesothelial lining of the serosal surface of the coelomic cavities (WHO 2007) and is usually considered to be malignant (Uzal et al. 2016).
There is a clear association with aerosol exposure to asbestos in people, but causal factors of naturally occurring mesothelioma have not been well established in dogs, although an association with chronic idiopathic pericarditis (Machida et al. 2004) and asbestos exposure has been proposed (Glickman et al. 1983, Harbison & Godleski 1983). In dogs, the tumour most commonly affects the pleura, followed by the peritoneum and pericardium, and is characterised by pale, pedunculated or sessile nodules, plaques or villous projections (Uzal et al. 2016). Reactive mesothelial cells can closely mimic both mesothelioma and carcinoma on histopathological and cytological examination of tissues and effusions, respectively, and it is widely recognised in human and veterinary medicine that distinguishing reactive from neoplastic mesothelium can be very challenging. This is particularly problematic in the pericardium because mesothelial cell reactivity is more marked in this site, presumably due to the increased local friction caused by the beating heart (Baker & Lumsden 2000). In dogs, an intermediate category of “atypical mesothelial proliferation” has also been suggested (Cagle & Churg 2005), as described in people (Churg & Galateau-Salle 2012).
Numerous studies have investigated the value of immunohistochemical (IHC) panels to aid in the distinction of reactive from neoplastic mesothelium in human patients. Guidelines for the pathological diagnosis of malignant mesothelioma in people recommend a combination of histopathological features and specific IHC panels (Husain et al. 2013), e.g. desmin, p53, epithelial membrane antigen (EMA), glucose transporter 1 (GLUT1) and insulin-like growth factor II mRNA-binding protein 3 (IMP3) (Galateau-Salle et al. 2015). A recent study on human tissue reported a sensitivity of 100% and specificity of 95% for the diagnosis of mesothelioma versus reactive mesothelial cells using immunohistochemical methods for GLUT1 and IMP3 in formalin-fixed, paraffin wax-embedded tissues, with haematoxylin and eosin (HE) staining as the gold standard for comparison (Minato et al. 2014).
Strong expression to both antibodies was consistent with mesothelioma and negative reactions with benign proliferations in their study. However, it is well recognised that, while IHC may provide supporting evidence, this is based on statistical probability and may not always be useful in an individual case (Husain 2014). Others have found the relationship to be less definitive (Husain et al. 2013, Lee et al. 2013). Currently, the diagnosis of canine pericardial mesothelioma is based on histopathological examination of haematoxylin and eosin-stained sections of tissue obtained at surgery. However, more reliable differentiation of mesothelioma from atypical mesothelial proliferation and reactive mesothelium is required. This would enable pathologists to provide a more accurate definitive diagnosis and consequently help clinicians to make better decisions with respect to treatment and prognostic advice. This preliminary study investigated an immunohistochemical panel of cytokeratin, vimentin, desmin, GLUT1 and IMP3 in differentiating reactive mesothelium, atypical mesothelial proliferation and mesothelioma in dogs. The pericardium was chosen for the study because idiopathic pericarditis presents a common diagnostic conundrum and because the marked reactivity shown by pericardial mesothelium makes distinction from mesothelioma particularly challenging.
MATERIALS AND METHODS
Cases This was a multicentre, retrospective study, carried out with approval of the local Veterinary Ethical Review Committee (approval no. 06 14). The authors’ clinical and pathological records were searched for cases of IP, atypical mesothelial proliferation and mesothelioma; then, the archives of their four histopathology laboratories were searched for paraffin wax-embedded tissue blocks from biopsies and post mortem examinations. Based on the diagnosis from routine haematoxylin and eosin-stained sections, they were classified as having simple hyperplasia [reactive pericarditis (idiopathic pericarditis), group R], atypical mesothelial proliferation (group A) and mesothelioma (group M). Those from the university laboratory were cases referred to the university’s hospital for small animals.
In addition, pericardium from clinically normal dogs, euthanased for behavioural reasons, was collected by the university laboratory for use as controls. Haematoxylin and eosin sections were reviewed, and any cases in which the tissue was of poor quality were excluded. In all cases, attempts were made to determine clinical history, clinical findings and the outcome from patient records where available and by contacting the owners via the primary clinicians. The information available was variable but included the main presenting signs; the presence, site and volume of cavitary effusions; and survival times. Tissue processing and staining Routine histopathology All tissues had been fixed in 10% phosphate-buffered formalin and embedded in paraffin wax. Sections (4μm) were cut and stained with haematoxylin and eosin. In cases for which more than one tissue block was available, the highest-quality block based on the haematoxylin and eosin sections was chosen for immunohistochemical.
Immunohistochemistry All IHC was conducted in the university laboratory. Sections (4 μm) of formalin-fixed, paraffin wax-embedded tissue were placed on SuperFrost® Plus-coated slides (Thermo Electron Ltd.), dewaxed, hydrated and rinsed in distilled water. To block non-specific endogenous peroxidase, sections were treated with a blocking agent (REAL blocking agent S202386; Dako Ltd.).
All antibodies were diluted in antibody diluent (S0809; Dako). For cytokeratin (CK), mouse monoclonal anti-cytokeratin antibody clone MNF116 (M0821; Dako) was diluted 1/50 and incubated for 30minutes at room temperature (RT) following antigen retrieval using proteinase K (S3020; Dako) for 30minutes at room temperature (RT). For vimentin, mouse monoclonal anti-vimentin antibody clone V9 (NCL-L-VIMV9; Novocastra Laboratories) was diluted 1/400 and incubated for 30minutes at RT following antigen retrieval using high-pH antigen-unmasking buffer (H-3300; Vector Laboratories Ltd.) at 110°C for 5 minutes. For desmin, mouse monoclonal anti-desmin antibody clone D33 (M0760; Dako) was diluted 1/50 and incubated for 30minutes at RT following antigen retrieval using high-pH antigen-unmasking buffer (H-3300; Vector) at 110°C for 5 minutes. For GLUT1, rabbit polyclonal anti-GLUT1 antibody (15,309; Abcam Ltd.) was diluted 1/500 and incubated overnight at 4°C following antigen retrieval using 0•01M citrate buffer with pH 6•0 at 110°C for 5 minutes. For IMP3, mouse monoclonal anti-IMP3 antibody clone 69•1 (M3626; Dako) was diluted 1/50 and incubated for 30minutes at RT followed by antigen retrieval using high-pH antigen-unmasking buffer (H-3300; Vector) at 97°C for 50minutes. Following incubation with the primary antibody, the sections were incubated with secondary antibody [Envision anti-mouse HRP (K4007) or Envision anti-rabbit HRP (K4011); Dako, as appropriate) and labelled with DAB+ chromogen (K3468; Dako). Sections were counterstained with Harris haematoxylin. All heat-induced epitope retrievals were carried out using the Histos 5 microwave histoprocessor (Milestone SRL), and all washings between steps were carried out using Tris-buffered saline plus Tween (TA999-TT; Thermo Fisher Scientific Ltd.). Positive controls were canine tissues known to express the relevant antibody (GLUT1: oesophagus, IMP3: stomach, CK: skin, liver, intestine and kidney, vimentin and desmin: intestine), and negative controls were processed without the primary antibody.
CK, vimentin and desmin immunohistochemical are well established diagnostic techniques in veterinary laboratories, and preliminary studies in the laboratory of one of the authors showed a cross-reaction of the GLUT1 and IMP3 antibodies used with canine brain and squamous cell carcinoma, respectively (data not shown). Assessment of sections The original histopathology reports were used for preliminary categorisation of the diagnoses, which were then refined by two pathologists who reached a consensus diagnosis. The pathologists were blinded to the original diagnosis. The sections were categorised as either normal, simple hyperplasia (reactive), atypical mesothelial proliferation or mesothelioma based on the descriptions of Cagle & Churg (2005).
Briefly, they state that normal mesothelium is composed of flat, inconspicuous cells, and simple hyperplasia results in a more conspicuous but bland layer of cuboidal cells, sometimes with distinct nucleoli. Atypical mesothelial hyperplasia is described as a more florid mesothelial proliferation that varies from a single layer of cells with more prominent atypia to focal, raised accumulations of cells with more marked anisocytosis and anisokaryosis and prominent nucleoli. Small tubule-like structures may sometimes be present. In mesothelioma, there is expected to be an invasion of underlying tissues, cellular nodules with expansion of the stroma, lack of the normal layers of the pleura and atypical cells within the full-thickness, severe cellular atypia and areas of necrosis (Cagle & Churg 2005).
The immunohistochemistry slides were assessed by two pathologists who were blinded to the histopathological diagnosis. The sections were scored semi-quantitatively for each antibody for each section using shared example images as a guide: Percentage of mesothelium positive: 0=completely negative, 1=1 to 25% of cells positive, 2=26 to 50% positive, 3=51 to 75% positive and 4=greater than 75% positive. Intensity of staining: negative (0), weakly positive (1), moderately positive (2) and strongly positive (3). The pattern of staining within the cells was also recorded. All antibodies stained the cytoplasm and/or cell membrane, and the pattern was recorded as membranous (predominantly on the cell membrane) and/or diffuse (all around the nucleus), focal (random), focal (basal) or focal (apical). Focal (basal) and focal (apical) categories could only be determined where the polarity of the cells was evident. Statistical analysis For each pathologist’s scores, the statistical analysis of the comparison between groups for number of cells stained and intensity of staining was carried out separately, i.e.
the scores for the two pathologists were not averaged. A two-pronged approach to the analyses was adopted. First, whether combinations of scores from the different stains were able to distinguish between groups in an expected way, i.e. a gradation from group C to R to A to M, was evaluated by the production of heatmaps using the package gplots (v 3•0.1) in R (v 3•4.0© 2017; The R Foundation for Statistical Computing), allowing the creation of hierarchical dendrograms to determine the closeness or distance of groups and stains from each other as represented by a false image matrix. Only the average number of cells and intensity scores for IMP3, GLUT1 and desmin were included. CK and vimentin were not included in this comparison as they were only used to identify the cells as being mesothelial in origin (positive for CK and vimentin) and to aid the subtyping of the mesothelioma cases. Second, classification tree-based analysis involving partitioning of the groups via binary recursive partitioning using maximum likelihood was used (Clark & Pregibon 1997).
This technique is of particular use with multivariable analyses where the degree of imbalance in group sizes makes more standard linear analysis approaches difficult (McCann et al. 2007). Its use here was an attempt to create an algorithm for the identification of cases of R, A and M using intensity of staining and number of cells stained for IMP3, GLUT1 and desmin, and the technique was carried out using the package tree (v 1•03-37).
Cases The tissues dated from 2000 to 2016 and comprised mainly pericardium with or without myocardium and attached epicardium. Based on histopathological examination, there were 10 cases of simple mesothelial hyperplasia (group R), 17 cases with atypical mesothelial proliferation (group A) and 26 cases considered to have mesothelioma (group M). There were five controls that consisted of young, pure or crossbred, adult Staffordshire bull terriers from a pet shelter that were euthanased for behavioural reasons.
Table 1. Signalment and selected history and clinical features of cases of idiopathic pericarditis (reactive), atypical mesothelial proliferation and mesothelioma included in the study
Exact ages were not available for the control dogs. The signalment and selected clinical details of the cases in groups R, A and M are shown in Table 1.
A range of breeds were represented with a marked predominance of large dogs in all three groups. Golden retrievers were the most common breed in all disease groups. In all three groups, the age at onset and survival time from first presentation were similar, with mainly middle-aged and elderly dogs affected. However, there was little available survival data. The dogs in each group showed a range of similar clinical signs.
The volume of pericardial fluid obtained on pericardiocentesis was large in all three groups, and the time to relapse of signs after pericardiocentesis was very variable and overlapped between groups. Haematoxylin and eosin staining Examples of the histopathological patterns of staining, classified according to the method of Cagle & Churg (2005), are shown in Fig 1. Normal mesothelial cells formed a single flat cell layer lining the pericardium. These cells contained a small nucleus with an indistinct nucleolus.
The deeper adipose and fibrovascular connective tissues were unremarkable. Reactive mesothelial cells were cuboidal and formed a variably intact, single cell layer lining on the pericardial surface. Nuclei were oval and contained a small prominent nucleolus and dispersed chromatin. Atypical mesothelial cells were polygonal and formed short to elongated exophytic proliferations from the pericardial surface. The proliferations were supported by variable amounts of fibrovascular connective tissue. Nuclei were oval and contained multiple, variably sized nucleoli and dispersed chromatin and occasional mitotic figures were identified. The supporting stroma was infiltrated by variable numbers of lymphocytes, plasma cells and neutrophils. Atypical mesothelial cells were occasionally identified within subserosal pericardial lymphatic vessels and there was occasional entrapment of the mesothelial cells by layers of fibrous connective tissue within the superficial subserosal stroma. In cases classified as mesothelioma, the cells had an appearance similar to atypical mesothelial cells, but there was extensive invasion of the underlying subserosal adipose tissue. If invasion could not be identified clearly, then the case was assigned to group A. Lymphatic invasion was occasionally identified, but this was not used as a criterion of malignancy given that reactive mesothelial cells can enter subserosal pericardial lymphatics. Additional findings included occasional marked pleomorphism, atypical mitoses and areas of necrosis.
Mesotheliomas were either polygonal (epithelial pattern) or spindle-shaped (sarcomatous pattern), and both morphologies were sometimes identified within the same mesothelium (biphasic pattern). Of the mesothelioma cases, 15 had a tubular or tubulopapillary epithelial pattern, three were sarcomatous, and eight were biphasic (Table S1, Supporting Information). Intravascular mesothelial cells were evident in the subserosal layers in one case from group R (10%), three from group A (17•6%) and nine from group M (34•6%).
The comparison between the clinical diagnosis, original histological diagnosis at the time the sample was submitted to the respective laboratory as a diagnostic sample and the consensus diagnosis as part of the study is shown in Table 2. For group R, the original pathologist’s diagnosis was nine of 10 classed as reactive and one of 10 as mesothelioma. In group A, the original diagnosis was 11 of 17 reactive, five of 17 mesothelioma and one of 17 atypical proliferation. For group M, the original diagnosis was 19 of 26 mesothelioma, five of 26 reactive and two of 26 atypical proliferation. In a majority of cases in groups R and M, there was good agreement between the clinical and original histological diagnoses but not for group A; this was considered to be a result of a lack of use of the group A category by clinicians and the pathologists who originally examined the samples.
Similarly, clinical diagnosis and the consensus histological diagnosis for this study were in agreement for seven of nine cases in group R, for which a clinical diagnosis was available, and 11 of 13 for group M but only four of 11 for group A. IHC staining The immunohistochemical slides were scored independently by two pathologists, both of whom were considered to be able to score consistently; one tended to score slightly higher than the other for the intensity of staining and the number of cells stained, but the results were qualitatively similar.
To avoid excessive complexity, the results for only one pathologist, who had the most experience of scoring, are presented, although the individual scores for each FIG 1. Examples of histological findings in different groups: (A) low-grade reactive mesothelium with oedema and mild inflammation; (B) atypical mesothelial proliferation with severe inflammation and haemosiderin-laden macrophages; (C) mesothelioma, epithelial type; (D) mesothelioma, sarcomatous type. H&E. x10, scale bars 200μm E. Milne et al. 6 Journal of Small Animal Practice • © 2018 British Small Animal Veterinary Association dog by both pathologists are available in Table S1.
Clinical diagnosis, original histological diagnosis and consensus diagnosis by the study pathologists in cases of reactive mesothelium
The sections in the controls and all three disease groups showed strong staining for CK and vimentin in most of the mesothelial cells, confirming mesothelial origin (Fig 2).
Intracellular staining pattern CK and vimentin were present intracytoplasmically in the mesothelial cells. Desmin staining was more variable and was predominantly cytoplasmic in a diffuse location around the nucleus, although occasionally more focal [three (30%) group R, one (5•9%) group A, five (19•2%) group M]. In addition, mesenchymal cells (fibroblasts and myocytes) were positive for vimentin, and myocytes were immunopositive for desmin, but staining patterns were not further assessed in these cell types.
FIG 2. (A–E) Atypical mesothelium with the expected staining pattern: (A) cytokeratin positive x40, (B) vimentin positive x40, (C) desmin positive x40, (D) GLUT1-low staining x40, (E) IMP3 negative x40. (F–J) Mesothelioma with the expected staining pattern: (F) cytokeratin positive x40, (G) vimentin positive x40, (H) desmin negative x40, (I) GLUT1 positive x40 and (J) IMP3 positive x40. Scale bars, 50μm
GLUT1 staining was both cytoplasmic and membranous. IMP3 was diffusely cytoplasmic and less often focal [four (23•5%) of group A and seven (26•9%) of group M (two randomly focal, three apical and two basal)]. Only three cases in group R stained positively for IMP3, and in each case, the staining was focal, predominantly at the basal pole of the cells. Examples of sections stained for GLUT1, IMP3 and desmin are shown in Fig 2. IHC scoring for IMP3 and GLUT1 and desmin Examples of staining for IMP3, GLUT1 and desmin are shown in Fig 2. The results of the IHC scoring for intensity of staining and number of cells stained for IMP3, GLUT1 and desmin are shown in the heatmap in Fig 3. Using a panel of all three antibodies, group C was distinguishable from the disease groups as having the lowest staining for all three and the lowest numbers of cells stained. Of the disease groups, M and A were similar to each other, while group R differed from groups M and A. IMP3 showed the greatest difference in staining between disease groups, with a gradation of increasing intensity and number of cells stained from R to A to M, with group R indistinguishable from group C (Fig 3). For GLUT1, group R had less staining than groups M and A, which were similar to each other.
In a small number of cases of mesothelioma, GLUT1 conferred an advantage over IMP3 alone, e.g. of four cases in group M with low IMP3 staining, three were high for GLUT1. Group C had the lowest intensity and number of cells stained for desmin. Intensity of staining for desmin was not useful in distinguishing groups A and M, which had the greatest intensity; group R was an intermediate between control and A/M. For the number of cells staining for desmin, groups A and R had the highest number, with group M intermediate between groups A/R and C (Fig 3). An algorithm using the above variables with the aim of creating a decision tree for individual cases confirmed the trends shown by the heatmaps but was not of additional value in making a definitive diagnosis (data not shown). There were no consistent differences between mesothelioma subtypes in intensity of staining or numbers of cells stained for any of the five antibodies, although the numbers of sarcomatous and biphasic cases were small (data not shown).
FIG 3. Heatmap of a row and column dendrogram of the average number of cells and intensity of the three stains (desmin, GLUT1 and IMP3) as observed in the four groups
The difficulty in establishing a definitive diagnosis in cases of pericardial effusion is well recognised, even when using diagnostic imaging, cytological examination of effusions and pericardial biopsy. In addition, clinical signs, history and diagnostic imaging including radiography and echocardiography are also similar in IP and mesothelioma (Smith & Hill 1989, McDonough et al. 1992, Stepien et al. 2000). Indeed, in one study, masses were suspected on diagnostic imaging in some cases of IP and were not evident in some dogs with mesothelioma (Stepien et al. 2000). In the present study, similar difficulties were encountered.
Survival time has been reported to be longer in IP than mesothelioma (Stepien et al. 2000), but others report similar survival times (Dunning et al. 1998). In the present study, the survival times were similar between all three groups, although survival data were not available for all patients. This aspect would be better addressed in a prospective study, but the similar survival times were not considered to preclude a correct histological diagnosis (Dunning et al. 1998). In group A, there were many discrepancies between the clinical diagnosis, the original diagnosis by the laboratory at the time of submission of the material and the consensus diagnosis used in this study.
This is likely to be mainly due to the atypical category not being adopted by all clinicians and pathologists, and some samples predated its description (Cagle & Churg 2005). However, for groups R and M, the agreement between the clinical, original histological and consensus histological diagnoses was much closer, supporting the accuracy of the consensus histological diagnosis used in this study. However, the possibility that some histological diagnoses were incorrect is not ruled out, a common limitation of studies in which the gold-standard diagnostic method may be suboptimal. The distinction between reactive and neoplastic mesothelial cells, and in some cases other types of neoplasia, in cytological and histopathological samples is a common and challenging dilemma in veterinary pathology (MacDonald et al. 2009, Cagle et al. 2014). This has significant implications for patient management and prognosis. Simple methods such as effusion pH (Fine et al. 2003) or biochemical analysis (Laforcade et al. 2005) have not proven to be of value in this respect. However, there is currently an interest in both veterinary (Höinghaus et al. 2008, Przeździecki & Sapierzyński 2014, D’Angelo et al. 2014,) and human pathology (reviewed by Churg et al. 2016) in the use of immunological techniques on cytological and biopsy specimens to distinguish reactive and neoplastic processes affecting serosal surfaces. The aim of the present study was to develop a preliminary immunohistochemical panel, based on the one recommended for use in human pathology, as a potential aid to definitive diagnosis on pericardial biopsies.
The age at presentation was similar to that reported previously for idiopathic pericarditis and mesothelioma (Stepien et al. 2000, Stafford Johnson et al. 2004), although in our cases series, males predominated, especially in groups R and A. The reason for the apparent difference is unclear but has been reported previously for IP (Gibbs et al. 1982). Idiopathic pericarditis is known to occur predominantly in large-breed dogs, with a predominance in golden retrievers (Aronsohn & Carpenter 1999, Stepien et al. 2000, Stafford Johnson et al. 2004, Cagle et al. 2014), although mesothelioma is reported to be more common in small and medium-sized breeds (Stepien et al. 2000). In contrast to these reports, large breeds also predominated in group M in our study, and golden retrievers were FIG 3.
Heatmap of a row and column dendrogram of the average number of cells and intensity of the three stains (desmin, GLUT1 and IMP3) as observed in the four groups Immunohistochemistry of canine mesothelioma Journal of Small Animal Practice • © 2018 British Small Animal Veterinary Association 9 the most common breed in groups R, A and M, suggesting that the three histological categories could represent different stages of the same pathological process, at least in this breed. Progression from reactive to neoplastic mesothelium was also suggested by Machida et al. (2004), who described the development of lesions they considered to be consistent with mesothelioma in five golden retrievers with a long-term history of idiopathic pericarditis. However, the breed prevalence in our study could not be compared with those of the geographically dispersed study populations. The histopathological lesions in routine HE-stained sections in canine idiopathic pericarditis and mesothelioma have been previously described.
Regarding idiopathic pericarditis, there have been reports of diffuse and sometimes nodular fibrosis, neovascularisation, congestion, variable haemorrhage and inflammation and areas of mesothelial hyperplasia and loss (Aronsohn & Carpenter 1999, Day & Martin 2002, Peters et al. 2003). Expansion of the subserosal stroma by loose connective and granulation tissue, with a denser expansion of connective tissue more deeply, may be present, together with neutrophilic to mixed inflammatory infiltrates and haemosiderin-laden macrophages. (Day & Martin 2002). More proliferative lesions in the present study were designated as atypical mesothelial proliferations (group A) as they had features intermediate between simple hyperplasia and mesothelioma, particularly the exophytic appearance and entrapment of mesothelial cells in the subserosal stroma (Cagle & Churg 2005). Previous studies of canine mesothelioma have described multiple nodules of large cells in nests, cords and acini surrounded by fibrovascular stroma (McDonough et al. 1992). In addition, epithelioid, fibrous (sarcomatous) and biphasic types of canine mesothelioma have been described (WHO 2007). This description formed the basis for our sub-categorisation in this study.
Interestingly, the presence of embolising non-neoplastic mesothelial cells in the lumen of submesothelial lymphatics in the pericardium and mediastinal lymph nodes is reported in dogs with reactivity of the pericardial mesothelium (Peters et al. 2003, Goupil et al. 2012). This feature was also observed in several of our cases in all three disease groups; the highest frequency was in group M, but clearly, this is not a reliable feature in distinguishing mesothelioma from benign processes. The cause of embolisation is unclear, but it is also recognised in people, and it has been postulated that desquamated mesothelial cells can pass into lymphatics through stomata that are enlarged by serosal injury (Suárez Vilela & Izquierdo García 1998). Recognition of this phenomenon is important to avoid an erroneous diagnosis of metastatic malignancy. In human pathology, numerous studies on IHC have been undertaken to improve diagnostic confidence in cases of mesothelial proliferation.
This is used particularly to distinguish reactive mesothelium, metastatic disease affecting the pleura (e.g. carcinoma) and mesothelioma because of the therapeutic, prognostic and legal implications, but its value and interpretation remains controversial (Churg et al. 2015). In the absence of single specific markers, panels are recommended. GLUT1, IMP3, p53, desmin and epithelial membrane antigen are the main markers that have been proposed for distinguishing reactive mesothelium from mesothelioma in people, with desmin expected to be positive in reactive mesothelium, and the others more likely to be positive in mesothelioma. GLUT1 is a member of the glucose transporter family, which is upregulated in some malignancies. IMP3 is a member of the insulin-like growth factor II mRNA-binding protein family and is almost absent from postembryonic tissues but is upregulated in some malignancies (King et al. 2009). Several authors recommend a combination of GLUT1 and IMP3 (Minato et al. 2014, Galateau-Salle et al. 2015), but others suggest caution because the differences may be statistically significant between groups but difficult to interpret in an individual clinical case (Husain et al. 2013, Lee et al. 2013), or they distinguish reactive from neoplastic proliferations but not the type of neoplasia (Ikeda et al. 2010).
The use of more specialised methods as an adjunct to standard IHC, such as detection of the loss of expression of the tumour-suppressor gene p16, have been advocated (Hiroshima et al. 2016), but such genetic markers have not been validated for this purpose in dogs. Previous studies of IHC in canine mesothelioma are limited. McDonough et al. (1992) found CK and vimentin to be strongly expressed in mesothelioma, while Machida et al. (2004) used CK, vimentin and HBME-1 in five mesothelioma cases with strong positive staining for CK and weak but positive staining for vimentin and HBME-1. However, at least in people, mesothelial cells from the subserosal multipotential cells express only vimentin, whereas they co-express CK when they proliferate. Immunocytochemistry has been applied more recently to cytological preparations in dogs (Przeździecki & Sapierzyński 2014), and a combination of CK, vimentin and desmin was found to be useful: all three markers were strongly expressed in normal and reactive mesothelial cells and mesothelioma; there was weak expression of vimentin in carcinomas and weak expression of CK in sarcomas, and desmin expression was negative in both carcinomas and sarcomas.
However, this would not distinguish reactive from neoplastic mesothelial cells. Using cytological preparations, Höinghaus et al. (2008) also noted CK and vimentin positivity in canine mesothelioma but found desmin staining to be variable. For these reasons, our panel was extended to include GLUT1 and IMP3 in addition to CK and vimentin. In the present study, IMP3 was found to be the most valuable in distinguishing control, R, A and M groups (Fig 3). Similar results were obtained with GLUT1. On a group basis, inclusion of GLUT1 did not confer any additional advantage, but in a small number of mesothelioma cases, IMP3 staining was low but GLUT1 was high – therefore, the inclusion of both antibodies may have advantages over IMP3 alone; this has also been reported in human pathology (Minato et al. 2014). However, the less-than-perfect sensitivity and specificity of IMP3 and GLUT1 for distinguishing benign from malignant disease of the mesothelium is also recognised in human pathology by some authors (Shi et al. 2011, Chang et al. 2014), and our findings in the dog agree with this reservation.
Desmin was variable between groups and not considered to be a useful addition to a mesothelioma panel, as previously reported in dogs (Höinghaus et al. 2008). It was of interest that, at least for IMP3, there was an increasing gradation in staining in reactive, atypical and mesothelioma cases, suggesting that atypical mesothelial proliferation may be an intermediate stage between reactive mesothelium and mesothelioma or a precancerous stage of mesothelioma. Our conclusions are similar to those in human pathology in that a combination of GLUT1 and IMP3 may be of value in distinguishing between groups of mesothelial disease categories (Minato et al. 2014, Galateau-Salle et al. 2015) but can still be difficult to interpret in an individual clinical case (Husain et al. 2013, Lee et al. 2013, Husain 2014). Nevertheless, the results may still provide additional supporting information when investigating clinical cases. The pattern of staining within cells has been examined in human tissues; strong membranous staining of GLUT1 is said to favour mesothelioma (especially the sarcomatous form) over reactive mesothelium, although no such difference in pattern was shown for IMP3 or desmin, which was mainly cytoplasmic (Minato et al. 2014). No consistent differences between the groups or between subtypes of mesothelioma in staining pattern within cells were observed in our study. The present study has a number of limitations. In common with many such studies, it was difficult to be certain that all cases were accurately assigned to the disease groups using the suboptimal existing gold-standard method of haematoxylin and eosin -stained sections.
However, there was good agreement between the clinical diagnosis, original histological diagnosis and consensus diagnosis by the study pathologists for groups R and M, suggesting that, in most cases, the diagnosis was likely to be correct. Even allowing for the fact that some cases might have been assigned to the wrong group, some differences in immunohistochemical were observed. We were unable to obtain follow-up information from every case, and post mortem examinations were not performed on all dogs. In addition, it became evident that the two pathologists scoring the IHC scored them quantitatively slightly differently, although the trends were similar.
However, determining a consensus or an average score between the two was not undertaken as this was considered likely to complicate the interpretation and would not be representative of standard practice in a diagnostic laboratory. The inherent subjectivity of immunohistochemical scoring is a limitation of our study, as is the case in all pathological studies based on this method. For example, in some instances, a low positive score was allocated by one pathologist and a negative score by the other for the same section, and such discrepancies are difficult to avoid with this commonly used scoring method.
In the future, development of image analysis which is rapid and easy to use in diagnostic pathology laboratories, may increase the objectivity of scoring. A further limitation was the relatively small number of cases, and care must be taken not to over-interpret the findings, but because pericardial mesothelioma is a relatively uncommon diagnosis, obtaining a large case series is challenging. In conclusion, the use of an immunohistochemical panel including CK, vimentin, IMP3 and GLUT1 may be of some value in the diagnosis of cases of mesothelial proliferation in canine pericardium. Future prospective studies using a better characterised canine population and investigation of novel methods such as genetic markers are warranted. As pericardial biopsy is highly invasive, the successful application of similar techniques to cells in pericardial fluid could have clinical and welfare advantages for canine patients presenting with pericardial effusions.
Acknowledgements The authors are very grateful to BSAVA Petsavers for funding the study; the technical staff of The Royal (Dick) School of Veterinary Studies, Bridge Pathology Ltd., Abbey Veterinary Services and Cytopath Ltd. for their expertise; and the dog owners and clinicians who allowed storage of archive material and provided follow-up information. Conflict of interest No conflict of interest has been declared.
- Aronsohn, M. G. & Carpenter, J. L. (1999) Surgical treatment of idiopathic pericardial effusion in the dog: 25 cases (1978-1993). Journal of the American Animal Hospital Association 35, 521-525
- Baker, R. & Lumsden, J. H. (2000) Pleural and peritoneal fluids. In: Color Atlas of Cytology of the Dog and Cat. 1st edn., eds. R. Baker and J.H. Lumsden. Mosby, MO, USA. p164
- Cagle, P. T. & Churg, A. (2005) Differential diagnosis of benign and malignant mesothelial proliferations on pleural biopsies. Archives of Pathology and Laboratory Medicine 129, 1421-1427
- Cagle, L. A., Epstein, S. E., Owens, S. D., et al. (2014) Diagnostic yield of cytologic analysis of pericardial effusion in dogs. Journal of Veterinary Internal Medicine 28, 66-71
- Chang, S., Oh, M. H., Ji, S. Y., et al. (2014) Practical utility of insulin-like growth factor II mRNA-binding protein 3, glucose transporter 1, and epithelial membrane antigen for distinguishing malignant mesotheliomas from benign mesothelial proliferations. Pathology International 64, 607-612
- Churg, A. & Galateau-Salle, F. (2012) The separation of benign and malignant mesothelial proliferations. Archives of Pathology and Laboratory Medicine 136, 1217-1226
- Churg, A., Sheffield, B. S. & Galateau-Salle, F. (2016) New markers for separating benign from malignant mesothelial proliferations. Are we there yet? Archives of Pathology and Laboratory Medicine 140, 318-321
- Churg, A., Roggli, V., Galateau-Salle, F. et al. (2015) Classification of tumours of the pleura. In: WHO Classification of tumours of the lung, pleura, thymus and heart. 4th edn. Eds. W.D.
- Travis, E. Brambilla, A.P. Burke et al. International Agency for Research on Cancer, Lyon, France. pp 125-136.
- Clark, L. A. & Pregibon, D. (1997) Tree based models. In: Statistical models. Eds J. M.
- Chambers and T. J. Hastie. S.Wadsworth & Brooks/Cole Advanced Books & Software, Pacific Grove, CA, USA. pp 377-420
- D’Angelo, A. R., Di Francesco, G., Quaglione, G. R., et al. (2014) Sclerosing peritoneal mesothelioma in a dog: histopathological, histochemical and immunohistochemical investigations. Veterinaria Italiana 50, 301-305
- Day, M. J. & Martin, M. W. S. (2002) Immunohistochemical characterisation of the lesions of canine idiopathic pericarditis. Journal of Small Animal Practice 43, 382-387
- Dunning, D., Monnet, E., Orton, E. C., et al. (1998) Analysis of prognostic indicators for dogs with pericardial effusion: 46 cases (1985-1996). Journal of the American Veterinary Medical Association 212, 1276-1280
- Fine, D. M., Tobias, A. H. & Jacob, K. A. (2003) Use of pericardial fluid pH to distinguish between idiopathic and neoplastic effusions. Journal of Veterinary Internal Medicine 17, 525-529
- Galateau-Salle, F., Churg, A., Roggli, V., et al. (2015) The World Health Organisation classification of tumour of the pleura: advances since the 2004 classification. Journal of Thoracic Oncology 11, 142-154
- Gibbs, C., Gaskell, C. J., Darke, P. G. G., et al. (1982) Idiopathic pericardial haemorrhage in dogs: a review of fourteen cases. Journal of Small Animal Practice 23, 483-500< /li>
- Glickman, L. T., Domanski, L. M., Maguire, T. G., et al. (1983) Mesothelioma in pet dogs associated with exposure of their owners to asbestos. Environmental Research 32, 305-313
- Goupil, A., Bolliger, C., Lapointe, C., et al. (2012) Embolised mesothelial cells in a tracheobronchial lymph node associated with idiopathic chylopericardium in a dog. Journal of Small Animal Practice 53, 664-667
- Harbison, M. L. & Godleski, J. J. (1983) Malignant mesothelioma in urban dogs. Veterinary Pathology 20, 531-540
- Hiroshima, K., Wu, D., Hasegawa, M., et al. (2016) Cytological differential diagnosis of malignant mesothelioma and reactive mesothelial cells with FISH analysis of p16. Diagnostic Cytopathology 44, 591-598
- Höinghaus, R., Hewicker-Trautwein, M. & Mischke, R. (2008) Immunocytochemical differentiation of canine mesenchymal tumors in cytologic imprint preparations. Veterinary Clinical Pathology 37, 104-111
- Husain, A. N. (2014) Mesothelial proliferations: useful marker is not the same as a diagnostic one. American Journal of Clinical Pathology 141, 152-153
- Husain, A. N., Colby, T., Ordonez, N., et al. (2013) Guidelines for pathologic diagnosis of malignant mesothelioma: 2012 update of the consensus statement from the International mesothelioma Interest Group. Archives of Pathology and Laboratory Medicine 137, 647-667
- Ikeda, K., Tate, G., Suzuki, T., et al. (2010) Diagnostic usefulness of epithelial membrane antigen epithelial membrane antigen EMA, IMP3, and GLUT1 for the immunocytochemical distinction of malignant cells from reactive mesothelial cells in effusion cytology using cytospin preparations. Diagnostic Cytopathology 39, 395-401
- King, R. L., Pasha, T., Roullet, M. R., et al. (2009) IMP3 is differentially expressed in normal and neoplastic lymphoid tissue. Human Pathology 40, 1699-1705
- Laforcade, A. M., Freeman, L. M., Rozanski, E. A., et al. (2005) Biochemical analysis of pericardial fluid and whole blood in dogs with pericardial effusion. Journal of Veterinary Internal Medicine 19, 833-836
- Lee, A. F., Gown, A. M. & Churg, A. (2013) IMP3 and GLUT1 immunohistochemistry for distinguishing benign from malignant mesothelial proliferations. American Journal of Surgical Pathology 37, 421-426
- MacDonald, K. A., Cagney, O. & Magne, M. L. (2009) Echocardiographic and clinicopathologic characterization of pericardial effusion in dogs: 107 cases (1985-2006). Journal of the American Veterinary Medical Association 235, 1456-1461
- Machida, N., Tanaka, R., Takemura, N., et al. (2004) Development of pericardial mesothelioma in Golden retrievers with a long-term history of idiopathic haemorrhagic pericardial effusion. Journal of Comparative Pathology 131, 166-175
- McCann, T. M., Simpson, K. E., Shaw, D. J., et al. (2007) Feline diabetes mellitus in the UK: the prevalence within an insured cat population and a questionnairebased putative risk factor analysis. Journal of Feline Medicine and Surgery 9, 289-299
- McDonough, S. P., MacLachlan, N. J. & Tobias, A. H. (1992) Canine pericardial mesothelioma. Veterinary Pathology 29, 256-260
- Minato, H., Kurose, N., Fukushima, M., et al. (2014) Comparative immunohistochemical analysis of IMP3, GLUT1, epithelial membrane antigen EMA, CD146 and desmin for distinguishing malignant mesothelioma from reactive mesothelial cells. American Journal of Clinical Pathology 141, 85-93
- Peters, M., Tenhündfeld, J., Stephan, I., et al. (2003) Embolized mesothelial cells within mediastinal lymph nodes of three dogs with idiopathic haemorrhagic pericardial effusion. Journal of Comparative Pathology 128, 107-112
- Przeździecki, R. & Sapierzyński, R. (2014) Using of immunocytochemistry in differential diagnosis of neoplasms of serosal cavities in dogs. Polish Journal of Veterinary Sciences 17, 149-159
- Shi, M., Fraire, A. E., Chu, P., et al. (2011) Oncofetal protein IMP3, a new diagnostic biomarker to distinguish malignant from reactive mesothelial proliferation. American Journal of Surgical Pathology 35, 878-882
- Smith, D. A. & Hill, F. W. G. (1989) Metastatic malignant mesothelioma in a dog. Journal of Comparative Pathology 100, 97-101
- Stafford Johnson, M., Martin, M., Binns, S., et al. (2004) A retrospective study of clinical findings, treatment and outcome in 143 dogs with pericardial effusion. Journal of Small Animal Practice 45, 546-552
- Stepien, R. L., Whitley, N. T. & Dubielzig, R. R. (2000) Idiopathic or mesotheliomarelated pericardial effusion: clinical findings and survival in 17 dogs studied retrospectively. Journal of Small Animal Practice 41, 342-347
- Suárez Vilela, D. & Izquierdo García, F. M. (1998) Embolization of mesothelial cells in lymphatics: the route to mesothelial inclusions in lymph nodes? Histopathology 33, 570-575
- Travis, W. D., Brambilla, E., Burke, A. P., et al. (2015) Classification of tumours of the pleura. In: WHO Classification of Tumours of the Lung, Pleura, Thymus and Heart. International Agency for Research on Cancer, Lyon, France. pp 125-136
- Uzal, F. A., Plattner, B. L. & Hostetter, J. M. (2016) Alimentary system. In: Jubb, Kennedy and Palmer’s Pathology of Domestic Animals. 6th edn., Vol. 2. Ed M. G. Maxie. Elsevier, St. Louis, MO, USA. p 256
- WHO Histological classification of tumors of the alimentary system of domestic animals(2007) In: Tumors of Serosal Surfaces (Pleura, Pericardium, Peritoneum, and Tunica Vaginalis). Ed F. Y. Schulman. The Charles Louis Davis, DVM Foundation, Gurnee, IL, USA. pp 144-146