Morphological Distinction of Histiocytic Sarcoma from Other Tumor Types in Bernese Mountain Dogs and Flatcoated Retrievers

Discussion

In the present study, a large series of histiocytic sarcomas and its differential diagnoses in BMD and FCR were selected based on their initial diagnosis during a routine examination by a clinical or anatomic pathologist. It was found that a large proportion of these initial diagnoses, 47.5% and 46.3% for histology and cytology, respectively, were changed by the revision process to different diagnoses, with the help of a scoring system on several standardized morphological criteria combined with IHC. In the group of sarcomas, it was not always possible in this study to further characterize them after being excluded as of histiocytic origin. In an earlier comparison by Regan et al. of first and second opinion histopathology of 430 tumors in dogs, 30% of diagnoses disagreed, with 17% having an impact on therapy or prognosis, with an especially high percentage among mesenchymal tumors (18). In their study, four tumors (two sarcomas, one round cell tumor and one plasma cell tumor became HS and two HS cases became another diagnosis (one T-cell lymphoma and one melanoma). This stresses the need for high and uniform certification standards and standardized diagnostic criteria and workup for veterinary clinical and anatomic pathologists, and that review and (randomized) second opinion pathology/cytology and revision is recommended. Complete clinical information could further help to increase diagnostic accuracy (18). In human medicine, second-opinion tumor histopathology is already recommended as standard practice, before treatment is requested, to avoid unnecessary or inappropriate treatments. Although in humans the discrepancy rate is relatively small (1.4-5.8% changes) (19,20), it is high enough that the Association of Directors of Anatomic Surgical Pathology has recommended institutional consultation as a standard practice (21).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

A: Fine-needle aspirate (FNA) of inguinal lymph node of a Bernese Mountain Dog (BMD). Atypical mononuclear and multinucleated giant cells are present, with pathological mitosis, marked pleomorphism, anisokaryosis and vacuolated cytoplasm. May-Grünwald Giemsa (MGG), ×500. B: FNA of spleen of BMD. Hemophagocytic histiocytic sarcoma, with a clear phagocytosis of mononuclear cells and erythrocytes. MGG, ×1000. C: FNA of lymph node of a Flatcoated Retriever (FCR). A clear anisokaryosis, as well as erythrophagocytosis are present. MGG, ×500. D: FNA of liver of BMD. Cytoplasmic extensions are visible as well as multinuclearity with aberrant nuclear shapes. MGG, ×1,000. E: Cerebrum of FCR. Multifocally hemorrhage and spindle cells are present within the tumor mass. Hematoxylin and eosin (H&E), ×100. F: Spleen of BMD. In addition to the histiocytic tumor cells there are spindle cells with whorling present. H&E, ×400. G: Liver of BMD. Tumor cells expand the liver sinusoids and portal vessels. H&E, ×100. H: Lung of BMD. Bizarre multinucleated cell showing phagocytosis. H&E, ×400. I: Lung of BMD. The tumor cells, mononuclear as well as multinuclear, show strong expression of CD18. H&E counterstain, ×100.

Archiving of samples and good accessibility and maintenance of the archive, besides digital documentation of demographic, clinical and pathological information, linked with the samples, both easily traceable with a search engine, turned out to be paramount for enabling the contribution of samples to research. In this context, the great advantage of histology over cytology is that the original samples can remain in the possession of the laboratory, while many sections of the tissue blocks can be cut and used in research projects. In this study, histological needle core biopsy (Trucut) and other biopsies with a small width (<2 mm) were only assessable for a limited number of the morphological features. This resulted in a less reliable morphological diagnosis, while the performance of IHC is often less reliable in such small samples due to false-positive edge staining. Therefore taking such small biopsies is discouraged and surgical biopsies are highly recommended, which was also already concluded in human pathology (22).

Although the differential diagnosis of a sample suspected of being a malignant tumor, with a possibility of HS, can already be narrowed down with morphological examination with strict differentiating morphological criteria only, all histological samples were also stained and evaluated with the CD18 antibody, as an extra confirmation of the leukocyte origin of tumor cells. CD3 and CD79a should be added, as in this study morphology alone was also insufficient to differentiate HS from ML. A fair proportion of ML and HS switched from diagnosis between first and second evaluation. In order to indicate the macrophage origin of the HHS subtype, CD11d staining was helpful.

Some cases of HS in our study had weak or negative staining for CD18. Absence, weak or variable IHC staining intensity can be caused by the formaldehyde of the FFPE procedure. Formaldehyde makes recognition of antigens difficult, and overfixation with this preserving agent can result in detection failure (or only weak reaction) (23). Differences in the degree of differentiation of the histiocytic tumor cells can result in different intensity of staining within and between cases (17). Variability in staining intensity for lysozyme in MH is related to the degree of anaplasia of the neoplasia (7, 17).

Considering the fact that a large proportion of diagnoses were changed during the revision process, it is important to identify the cyto-/histomorphological characteristics of histiocytic sarcomas. Therefore, in the present study these aspects of histiocytic sarcomas were studied in 654 unique cases of two different predisposed breeds, BMD (n=365) and FCR (n=289). The systematic scoring of the morphological characteristics was performed on both histology and cytology slides, when both available, or on one of them. Although the tumors in both breeds had several common characteristics, several significant differences of HS morphology were found between BMD and FCR (Table III).

When looking at the morphology of the tumor cells the majority of cells were either round-oval or polygonal (BMD) or polygonal only (FCR). Round-polygonal cells have been reported before in canine HS (9) and are also the main cell type in human HS (10). Polyhedral, as used in the canine literature by Nakayawa et al. (13), seems to be a less accepted term and polygonal is preferable in the context of uniform terminology. Spindle cells were seen in BMD but significantly more often in FCR. Most likely this is related to the more common localized HS found in FCR. Both in our study, as well as a earlier report (24), spindle cells were more often seen in this localized form compared to the disseminated HS. Spindle cells have been reported in canine HS before, but only in combination with round-oval cells (1, 24-28). This makes this study the first report of HS cases with spindle cells only. Spindle cells are also reported in human histiocytic tumors. When spindle cells are present in humans, most cases turns out to be interdigitating dendritic cell sarcoma (IDCS) (29,30) or FDCS (31), especially when the presence of spindle cells is prominent and the formation of whorls is seen. Whorling is also reported in human IDCS (29,32) and FDCS (31), but never before in canine HS although the frequency was low and it was only seen in BMD. These canine cases with spindle cells (54 BMDs, 68 FCRs) may denote a yet unknown subtype of canine histiocytic tumors, possibly resembling IDCS and FDCS in humans. Unfortunately, no canine immunohistochemical markers are available yet to definitively characterize those possible subtypes. Further research is warranted.

Phagocytosis of erythrocytes and leukocytes by the tumor cells was significantly more often seen in BMD than in FCR, but this could have been related to the more frequent presence in disseminated HS. This has been reported before in canine HS (1, 6, 7, 28, 33-39) and is also often seen in human HS, sometimes simultaneously with spindle cells (40, 41). In addition, extramedullary hematopoiesis was seen more often in BMD than in FCR. This could well have been related to the higher amount of erythrophagocytosis seen in HS in BMD. As hematological data were missing in this study, a correlation with anemia could not be established. Extramedullary hematopoiesis is also reported in canine HHS (8). Vacuolization of the cytoplasm of tumor cells was seen in some 14% of the cases, equally distributed among the BMD and FCR. In earlier studies, the percentage of tumor cells showing vacuolization in both BMD and FCR was higher (1, 7, 33, 34, 38, 39, 42). However, in human HS, this is reported at a much lower percentage, comparable to the results in our study.

Multinucleated giant cells are common in canine HS (15, 33), especially in the disseminated form. In humans, multinucleated giant cells are less frequently present in HS (43), as well as in IDCS (13), FDCS (44) and indeterminate DCS (45), which may imply a more malignant nature of HS in canines. The characteristic kidney-shaped nuclei have also been reported before in canine HS (33, 39, 46) and in human HS (47) and IDCS (14). Indentations of the nucleus have been reported in the canine literature (1, 7, 24, 26, 28) as well as in human HS (41) and IDCS (48), and a combination of indentations, grooves or folds in human HS (43, 49), IDCS (14) and LCH (50). This indicates that a canine subtype comparable to human LCH/LCS may also exist. An eccentrically located nucleus was reported in canine HS (6, 37, 51) and in human HS (41) and IDCS (14). Further research will shed light on possible similarities among human and canine histiocytic tumor subtypes.

An inflammatory infiltrate of lymphocytes was seen in 21.8% (FCR) to 29.3% (BMD) of the cases. The infiltration of lymphocytes only has been reported before in canine HS (1, 6, 20, 24, 28, 39, 52), but inflammatory patterns of neutrophils only (34), a combination of lymphocytes and plasma cells (25), of lymphocytes and neutrophils (7), of neutrophils and eosinophils (33), and of lymphocytes, plasma cells and neutrophils (1, 26, 25, 51) have also been reported. The difference between BMD and FCR with regard to inflammatory pattern, with BMD having more often small lymphocytes, has not been reported before as far as we are aware.

Invasive behavior of tumor cells, both in blood and lymph vessels and in liver and lymph node sinusoids, was sometimes noted (up to 7.1%) in BMD, but was almost absent from FCR. This invasive behavior was exclusively found in disseminated HS. A sinusoidal pattern has been recognized before in canine HS in lymph nodes (37) and in the liver including invasion of the portal areas (28), and has been associated with the HHS subtype (8). It is also reported in human HS (54, 55). Red pulp and intravascular invasion with phagocytosis is recognized in canine HS (7) and HHS (8), and in human HS (55). Necrosis was more often seen in FCR than in the BMD and may indicate a higher malignancy/faster growing tumor than in the BMD.

Different morphological characteristics in different organs within one case were seen in this study and have been reported before (56). Metastatic tumor cells seem to first disseminate from the early primary lesion and then acquire additional genetic defects and often exhibit properties different from the cells of the primary lesion (57). Macrophage-dendritic cell-HS-hybrids are possible in dogs, but tissue infiltration is discrete by subtype (8). Further research is needed.

It can be concluded that canine HS, including its subtypes, can be recognized morphologically by a mostly easily harvested dense pleomorphic cell proliferation of large round-oval to polygonal, sometimes spindle-shaped, cells with abundant, often vacuolated, cytoplasm. Anisocytosis and anisokaryosis can be marked and the tumor cells have large nuclei, which are round, indented, lobulated, doughnut- or kidney-shaped, with one or several nucleoli, and often have abnormal mitotic figures. Neoplastic bi- or bizarre multinucleated giant cells, up to 160 μm, often accompany the tumor. Vacuolization and phagocytosis of erythrocytes and other (tumor) cells is often seen, but the intensity may vary. A mostly sparse infiltration of inflammatory cells is possible, consisting of small lymphocytes, plasma cells, neutrophils, eosinophils and/or reactive macrophages. Additionally, in the histological samples, a poorly demarcated mass can be seen, with invasion of surrounding tissue and of blood and lymph vessels, sinuses of lymph nodes or sinusoids of the liver, which is easily missed. Tumors with spindle cells can show whorling and the tumor is often accompanied by variable amounts of necrosis, hemorrhage and extramedullary hematopoiesis and sometimes thrombi.

The large variation in morphological features that has become visible in the population of BMDs and FCRs in this study, makes the existence of several subtypes of canine HS very likely. These subtypes assume there are different cells of origin, different routes of differentiation or different involved genetic causes. This underlines the importance of further research on subtypes of HS because those subtypes likely entail different prognoses and different therapies. From this comparison with canine and human literature about histiocytic tumors, it can be concluded that it is likely that canine counterparts exist of human (macrophage) HS, FDCS, IDCS and LCH/LCS.

Different percentages of cellular features in the BMD and FCR (Table III) may indicate a different mixture of histiocytic cell type origins in FCR and BMD, resulting possibly from partial genetic or environmental differences at the onset of HS (4, 58). The breed-associated clinical features, such as differences in median age at diagnosis and localization of primary tumor, and a small number of DNA copy number aberrations between BMD and FCR also indicate differences in disease subtype between the two breeds (4). Different authors use different names to describe cell types. In order to come to a uniform report on HS, a uniform terminology of cell types has to be accepted.

HHS represents a distinct entity in the dog with distinctive clinicopathological features (15). Clinical and pathological findings which can indicate HHS are phagocytosing histiocytes, a diffuse pattern of tissue involvement, with preferably intra-sinusoidal (liver or lymph nodes) and intravascular spread and involvement of splenic red pulp, while encroaching on the white pulp with only sparse (T-cell) lymphocytic infiltration, extramedullary hematopoiesis, especially when the phagocytosis is prominent. Immunohistochemically, HHS cells typically have abundant and strong CD18 positivity (but this can be less intense in less differentiated tumors) (8) and, if tested, CD11d positivity and Coombs negativity as a requisite, while responsive anemia, thrombocytopenia and hypoalbuminemia are further supportive. When HS cases show these findings there is a high possibility that these are of the HHS subtype (15). In the present study, we found the percentage of hemophagocytic HS in BMDs to be at least 12.3% (45 out of 365) and in the FCR at least 2.4% (7 out of 289). The use of CD11d could be helpful for the diagnosis of HHS, but seems not to be essential for diagnosis because morphology and clinical symptoms of HHS seem to be characteristic, in other words they can differentiate HHS cases from classic HS cases. In addition, CD11d is not specific for HHS. When looking at the human HS cases for which phagocytosis was explicitly reported, the morphological characteristics of canine HHS were also found in the human counterpart (55).

Morphological and IHC differentiation of HS and other malignant tumors and histiocytic diseases is possible and feasible on routine FFPE histological and cytological samples. Therefore strict and specific criteria and a standardized workup are essential. Differences in cellular features between humans and dogs and between BMD and FCR may indicate pathophysiological species differences and a different mixture of histiocytic cell type origins in FCR and BMD possibly results from partial genetic or environmental differences at the onset of HS. Furthermore, the implementation of revision and consensus of pathological diagnoses is recommended.