Abstract
The aim of this study is to report 46 new cases of canine T-cell lymphomas among a series of 140 lymphomas studied by immunophenotyping (incidence 32.8%). According to the updated Kiel classification adapted to the canine species, 13 were classified as low-grade and 33 as high-grade lymphomas. Among the low-grade lymphomas, five were small clear-cell lymphomas, three were pleomorphic small-cell lymphomas, and five mycosis fungoides. Among the high-grade cases, there were 11 pleomorphic mixed-, small-, and largecell lymphomas, 6 pleomorphic large-cell lymphomas, 11 lymphoblastic lymphomas, and 5 unclassifiable highgrade plasmacytoid lymphomas. The cytohistologic features were highly suggestive of a T-cell phenotype on the basis of cell morphology (irregular nuclei and clear cytoplasms) (30/46 cases), a T-cell zone pattern, and the presence of hyperplastic postcapillary venules (22/46 cases). All 46 cases were CD3+ CD79a-, and among 34 cases investigated for CD4 and CD8 expression, 13 were CD4+CD8-, 13 were CD8+CD4-, and 8 were CD4CD8 double positive or double negative. The pleomorphic mixed lymphomas were mainly CD4+CD8- (6/7) and the lymphoblastic lymphomas were double positive or double negative (6/8). The main clinical, hematologic, and biochemical features were generalized (28/46) or regional lymphadenopathy (16/46), hepatosplenomegaly (15/46), extranodal involvement (11/46), mediastinal mass (9/46), and leukemia (8/46), which were mainly present in cases of lymphoblastic lymphomas and hypercalcemia (16/46).
Non-Hodgkin's lymphomas (NHLs), according to human terminology, are common tumors in dogs, whereas Hodgkin's lymphoma has not been recognized yet in this species. Their annual incidence rate was estimated at 33 per 100,000 dogs in the last epidemiologic study of Teske. 84 They are highly aggressive, 23,28,85 and their etiology remains obscure. 36,84 In contrast with similar tumors in cats, no viral cause has been demonstrated, although it has been suggested in the past. 68,75,82,88 Moreover, a canine long-term T-cell line, established from a dog with Sézary syndrome, and a canine large granular lymphocyte leukemia, both producing retroviral particles, have been recently reported. 26,27
NHLs are also common in humans and are on a noticeable increase. 12,15 Given their rate of increase, estimated at 3–10% per annum 9,65 at the beginning of the new century, NHLs constitute a significant public health issue. However, NHLs are a group of disparate malignant diseases. The association between a particular environmental cause and the increase in the incidence is now well established in humans for the association of Helicobacter pylori with gastric lymphoma, 93,96 the Epstein-Barr virus (EBV) with some Burkitt's and nasal T-cell lymphomas, 4,32,33 and the human T-cell lymphotropic virus (HTLV-I) with the adult T-cell leukemia/lymphoma (ATL/L). 41,87,92 Thus, in order to provide a plausible explanation for the overall increase, it is important to understand which subgroups are increasing.
The dog, as a close companion of man and submitted to the same environmental influences, has often been proposed as a useful model for studying the etiology and therapy of NHLs. 35,54,69,73,83 The value of the canine model nevertheless depends on the possibility of distinguishing the main subgroups in both species using a reliable classification of canine NHLs following the schemes currently used for human lymphomas. The morphologic classifications of canine NHLs have been successively based on different human classifications, e.g., Rappaport, 78 working formulation, 67 and Kiel classifications. 53 The first morphologic studies of canine NHL cells emphasized the rarity of low-grade malignancy lymphomas and, in particular, follicular lymphomas. However, these studies sometimes differ regarding the frequency of occurrence of the different morphological subtypes. 1,7,8,19,71
More recently, the immunophenotype has been taken into account, resulting in classifications such as the updated Kiel classification, 52 which is based on the fundamental distinction between B-cell and T-cell lymphomas. 1,28,51,85 For the veterinary authors, the main difficulties were the standardization of the availability of the different markers for canine lymphoid cells. This standardization came about only recently, in July 1993, during the first Canine Leukocyte Antigen Workshop (CLAW). It demonstrated the availability of accurate T-cell markers (antibodies recognizing the equivalent of the CD3, CD5, CD4, and CD8 molecules) 10,14,21,62,80 but not of others, such as F3-20-7 85 or CA1–4G8, 61,85 not truly specific for the T-lineage. Furthermore, no B-cell markers were available except for an antibody against the CD21 molecule 61 and for surface (sIg) and cytoplasmic immunoglobulins (cIg). 80 As well, there was no true equivalent to the pan-B antibodies used against the CD19 and CD20 molecules in humans. Lately, a monoclonal antibody against the mb1 molecule (CD79a) in humans 58 was shown to cross-react with numerous mammalian species, including the dog. 43,44 Hence, it has been used as the more reliable pan-B marker in the later studies of canine NHLs. 22–24,48 This successive use of various markers, sometimes not readily available, explains the discrepancies between the former results concerning the immunophenotype of canine NHLs, which indicated a high percentage of lymphomas expressing a double (B and T) phenotype, 1,28,51 and the latter, which clearly separated B-cell from T-cell lymphomas. 23,85 However, Teske and collaborators 85 found a high percentage of T-cell lymphomas (37.9%) without any correlation between morphology and immunophenotype, whereas a smaller percentage of T-cell lymphomas (26%) with a morphology clearly suggesting a T-cell immunophenotype was found in a previous study. 23
In this article, we set out the incidence and the clinical, morphologic, and immunophenotypic presentation of 46 new cases of canine T-cell lymphomas in comparison with our previous study 23 and in the light of the current evolution of the human NHL classifications, i.e., the Revised European American classification of lymphoid neoplasms (REAL) 34 and the updated WHO classification. 74
Materials and Methods
Animals
Among a new series of 140 cases of canine NHLs referred to our laboratory between 1996 and 1999, all studied by immunophenotyping, 46 cases of T-cell lymphomas have been studied. All dogs were evaluated by complete physical, hematologic, and biochemical examination and had not been treated by chemotherapy before the diagnosis.
The diagnosis of malignant lymphomas was based on cytohistologic analysis of fine-needle aspirates (FNA) or imprint smears and biopsy specimens of at least one enlarged lymph node, except in two cases, including one isolated pericardial effusion and one localized intrathoracic infiltration (mediastinal mass and lymph nodes), in which only fine-needle aspirate examination was available. Histologic analysis of extranodal masses, cutaneous biopsies, and cytologic analysis of bone marrow and effusions were selectively performed as well. Immunophenotype was established in all cases by the use of polyclonal antibody against CD3, as pan-T marker, and monoclonal antibody against CD79a, as pan-B marker, on fresh or frozen fine-needle aspirates and paraffin-embedded sections.
Thirty-four cases were evaluated for CD4 and CD8 expression on fresh or frozen cytological preparations (imprint smears, cytocentrifuge preparations of cell suspensions, fine-needle aspirates) and on frozen tissue sections when available.
Fine-needle aspirate and imprint smear evaluation
Cytological smears obtained by fine-needle aspiration (FNA) or imprint of an excised lymph node were air-dried, fixed, and stained by the May–Grünwald–Giemsa (MGG) technique. 90
The morphological classification criteria were based on cell size (medium, small, or large, i.e., nucleus equal to, smaller than, or larger than two red blood cells) and on the shape of the nucleus, the density of the chromatin, the presence and aspect of the nucleoli, the extension and basophilia of the cytoplasm, and the mitotic index. The mitotic rate was estimated in cytologic specimens by scanning five fields at 500× and counting mitotic figures. A low mitotic index was defined as 0–1 mitosis/five fields, medium as 2–4/five fields, and high as ≥5/five fields. However, regarding the irregular distribution of cell layers on cytologic preparations, the level of mitoses on histology was retained to establish the mitotic rate.
Biopsy specimen evaluation
Biopsy specimens for light-microscopic examination were fixed in 10% neutral buffered formalin at room temperature for 48 hours and processed for routine paraffin-wax embedding. Sections (1.5 µm) were cut and stained with hematoxylin and eosin (HE) and Giemsa. The pattern of the lymphomatous proliferation was noted (diffuse or not) as well as the presence of starry sky, hyperplastic postcapillary venules, and the composition of the background population. The mitotic rate was estimated in histologic specimens by scanning 10 fields at 400× and counting mitotic figures. A low mitotic index was defined as 0–2 mitoses/field, medium as 3–5/field, and high as ≥6/field. 8
Immunophenotyping
For immunophenotyping by CD3 and CD79a, a standard avidin-biotin immunoperoxidase procedure with Biomeda reagents (AutoProbe II 08–802X; Biomeda, Foster City, CA, USA) was performed. After pretreatment, the slides were covered for 10 minutes with a blocking reagent (tissue conditioner) and then incubated with primary antibodies under the dilution and incubation conditions listed in Table 1. After two rinses in phosphate-buffered saline (PBS), a biotinylated secondary antibody (reagent 1A) preincubated with normal dog serum was applied to each slide for 20 minutes at room temperature. After two washes, streptavidin-peroxidase (reagent 2) was applied for 20 minutes at room temperature. After two rinses in PBS, 3-3′diaminobenzidine was applied for 15 minutes at room temperature as a chromogen (reagent 3). The sections were then washed with distilled water and counterstained with Mayer's haematoxylin. Nonimmune serum was used as a negative control for each specimen. A normal canine lymphoid tissue was used as a positive control. For immunophenotyping by CD4 and CD8, cytological preparations were immediately used or stored at −70C and the lymph node biopsies were frozen in nitrogen-cooled isopentane for the preparation of frozen sections. Immunolabeling was then carried out according to the standard avidin-biotin immunoperoxidase procedure described above but without any pretreatment.
Antibodies used for immunohistochemical staining of canine lymphomas.
∗ Microwave oven pretreatment consisted of immersion of the slides in 10 mM sodium citrate buffer pH 6.0 and irradiation twice for 5 minutes in a 1470-W microwave oven. Pretreatment was not performed for cytological preparations.
† Cobbold, Oxford, UK.
Classification
The 46 cases were classified according to the updated Kiel classification with some adaptations, taking into account morphologic types not included in the human classification as previously described. 23
Results
Incidence, morphology, and immunophenotyping
The total incidence of T-cell lymphomas over this period of 4 years was 32.8% (46/140). The respective incidence of the various subtypes of the updated Kiel classification adapted to the canine species is listed in Table 2. Thirteen of these lymphomas (28.3%) were classified as low-grade lymphomas and 33 (71.7%) as high-grade lymphomas.
Updated Kiel classification of the 46 canine T-cell lymphomas: possible correlations with the REAL classification.
∗ Entity without established counterpart in updated Kiel and REAL classifications.
Among the low-grade lymphomas, eight (17.4%) belonged to the category of small-cell lymphomas, 5 (10.9%) of these being small, clear-cell lymphomas and three (6.5%) being pleomorphic small-cell lymphomas. Five (10.9%) were mycosis fungoides.
Among the 33 lymphomas classified as high grade, 11 (23.9%) were pleomorphic mixed-, small-, and large-cell lymphomas, 6 (13.0%) were pleomorphic large cells, 11 (23.9%) were lymphoblastic, and 5 (10.9%) were unclassifiable high-grade plasmacytoid lymphomas.
All 46 cases expressed a CD3+ CD79a− phenotype. Among the 34 cases evaluated for CD4 and CD8 expression, 13 were CD8+CD4−, 13 were CD4+ CD8−, 5 were CD4+ CD8+, and 3 were CD4− CD8−.
The main cytologic, histologic, and immunologic features of each type (Tables 3, 4) were as follows:
Mycosis fungoides (MF): Four cases were characterized by both lymph node and cutaneous infiltration and one by nodular cutaneous infiltration alone. Among the five cases, two were of the large granular lymphocyte type with extended pale cytoplasm containing azurophilic granules and slightly irregular nuclei (Fig. 1), two had cells with cerebriform, convoluted, or lobated nuclei, and one was of monocytoid appearance with fine chromatin, less irregular nuclei, and large, pale cytoplasm. Epitheliotropism (epidermotropism and/or folliculotropism) was more or less obvious, depending on the stage of the disease and on the presence or absence of epidermis (Fig. 2). Hyperplastic postcapillary venules were present in two cases. A heavy background of eosinophils was present in one case of large granular lymphocyte type. The mitotic index was low to medium except in one case, which had focal areas of high-grade lymphoma with numerous mitoses and a starry sky pattern. All cases (5/5) were CD8+ CD4−. Small clear-cell type: The five cases were mainly characterized by a small cell size and an extended, frequently unipolar, pale cytoplasm, sometimes containing fine azurophilic granules clearly visible on cytologic preparations. The round or slightly irregular nuclei had a more or less clumped chromatin and prominent nucleoli (Figs. 3, 4). The infiltrates were characteristically present in the paracortical areas (T-zone pattern) (Fig. 5) with preservation of multiple normal lymphoid B-follicles in all cases. Hyperplastic postcapillary venules were present in all cases (5/5) (Fig. 6), and the mitotic index was constantly low. Two of two cases evaluated for CD4/CD8 immunolabeling were CD4+CD8−. Pleomorphic small-cell type: Cytologically, the three cases consisted of a population of small cells with irregularly indented or deeply cleft nuclei with a more or less clumped chromatin and a scant, poorly visible pale cytoplasm (Figs. 7, 8). In all cases, there was diffuse lymph node involvement, hyperplasia of postcapillary venules, and a low mitotic index. The only case evaluated for CD4/CD8 immunolabeling was CD8+CD4− (1/1). Pleomorphic mixed-, small-, and large-cell type: These 11 lymphomas had diffuse architectures composed of atypical small, medium-, and large cells with irregular nuclei, fine chromatin, and pale or moderately basophilic cytoplasm. Hyperplastic postcapillary venules were present in four cases. The mitotic index was medium to high (Figs. 9, 10). Most cases (6/7) were CD4+ CD8−. Pleomorphic large-cell type: These six lymphomas had features similar to the previous type but with a strong predominance of large cells and a constantly high mitotic index (Figs. 11, 12). Hyperplastic postcapillary venules were marked in four cases. Three of six cases were CD4+CD8− and 3/6 were CD8+CD4−. Lymphoblastic type: Cytologically, all 11 cases had a monotoneous proliferation of medium-sized cells, with round or slightly convoluted nuclei, finely dispersed dusty chromatin, inconspicuous nucleoli, and scant moderately basophilic cytoplasm. Sometimes the cells appeared slightly elongated and bulb shaped on FNA. Architecturally, the infiltrate was always diffuse and had numerous mitoses (Figs. 13, 14). In the lymphoblastic lymphomas, we did not observe any hyperplasia of postcapillary venules except slightly in one case and no starry sky pattern except strongly in another case. These lymphoblastic lymphomas were double positive (CD4+ CD8+) or double negative (CD4− CD8−) in the majority of cases (6/8). Unclassifiable, high-grade, plasmacytoid type: Five lymphomas were classified as this particular type, as previously described.
23
They were all characterized by a proliferation of small- to medium-sized or less frequently large cells with round or slightly irregular nuclei, fine or irregularly clumped chromatin, irregularly visible multiple nucleoli, and relatively abundant highly basophilic cytoplasm. This abundant highly basophilic cytoplasm, with eccentric nuclei, occasional binucleation, and juxtanuclear pale cytoplasmic area, provided a striking plasmacytoid appearance (Figs. 15–18). These lymphomas were always diffuse, had an extremely high mitotic index, and showed in all cases a starry sky pattern. Hyperplastic venules were observed in three cases (3/5). These unclassifiable high-grade plasmacytoid lymphomas were CD8+ CD4− (3/5) or CD8+ CD4+ (2/5) but never CD4+ CD8−.
Morphology of the various subtypes of canine T-cell lymphomas.
∗ PCV = postcapillary venules.
Incidence of expression of CD4 and CD8 markers in canine lymphomas with morphological subtypes.
Mycosis fungoides. Large granular lymphocyte type. FNA. Note cytoplasmic azurophilic granules and some cleaved, more or less convoluted nuclei. MGG. Bar = 10 µm.
Mycosis fungoides (patch stage). Tissue section. Epidermis is infiltrated by neoplastic cells with nuclei varying from round to reniform and abundant, pale eosonophilic cytoplasm with poorly defined cell boundaries (Pautrier's microabcesses). HE. Bar = 40 µm.
Small clear cell type. FNA. Small cells with round or sometimes irregular nuclei, irregularly aggregated chromatin, prominent nucleoli, and projecting, uni- or bipolar, clear cytoplasm. MGG. Bar = 10 µm.
Small clear cell type. Tissue section. Note the round nuclei with irregularly aggregated chromatin. The extended pale cytoplasm is visible but less obvious than on FNA. HE. Bar = 10 µm.
Small clear cell type. Tissue section. Characteristic paracortical or T-zone pattern with numerous normal residual B-follicles. HE. Bar = 400 µm.
Small clear cell type. Tissue section. Hyperplastic postcapillary venules. HE. Bar = 25 µm.
Pleomorphic small-cell type. FNA. Small cells with irregular nuclear outlines. Note the scant, pale cytoplasm. MGG. Bar = 10 µm.
Pleomorphic small-cell type. Tissue section. The histological aspect is very similar to the cytological one, especially the size and the very irregular nuclear outlines of the lymphomatous cells. Nucleoli are more obvious than on FNA. HE. Bar = 10 µm.
Mixed pleomorphic, small- and large-cell type. FNA. Mixing of small, medium-sized, and large cells, all characterized by irregular outlines and pale cytoplasm. Note the nucleoli more obvious in the largest cells. MGG. Bar = 10 µm.
Mixed pleomorphic, small- and large-cell type. Tissue section. Note the striking irregular nuclear outlines, the abundant pale cytoplasm, and the various sizes. HE. Bar = 10 µm.
Pleomorphic large-cell type. FNA. Note the striking irregular nuclear outlines, fine chromatin, frequent visible nucleoli, and abundant pale cytoplasm. MGG. Bar = 10 µm.
Pleomorphic large-cell type. Tissue section. Irregular nuclear outlines, numerous nucleoli, and irregularly distributed pale cytoplasm are remarkable in the largest cells. HE. Bar = 10 µm.
Lymphoblastic type. FNA. Characteristic aspect of medium-sized cells with fine, dusty chromatin, poorly visible nucleoli, and small amount of moderately basophilic cytoplasm. Two bulb-shaped cells and a cell in metaphase. MGG. Bar = 10 µm.
Lymphoblastic lymphoma. Tissue section. Homogeneous proliferation of medium-sized cells with round or convoluted nuclei, dusty chromatin, and inconspicuous nucleoli. Giemsa. Bar = 10 µm.
Unclassifiable high-grade plasmacytoid type. FNA. Small, medium-sized, and large cells, with extended, deeply basophilic cytoplasm. One binucleated cell in the field. In this example, the lymphomatous cells exhibit fine chromatin and prominent nucleoli. MGG. Bar = 10 µm.
Unclassifiable high-grade plasmacytoid type. FNA. Note the irregularly clumped chromatin and the less visible nucleoli. One binucleated lymphomatous cell and a normal residual plasma cell in the center of the field. MGG. Bar = 10 µm.
Unclassifiable high-grade plasmacytoid type. Tissue section. Same case as in Fig. 16. Note the high mitotic index and the binucleated cell in the center of the field. HE. Bar = 25 µm.
Unclassified high-grade plasmacytoid type. Tissue section. Higher magnification of Fig. 17. HE. Bar = 10 µm.
Regarding all subtypes, the most frequent and striking morphologic criteria were constituted by hyperplastic postcapillary venules (22/46) except in lymphoblastic lymphomas, irregular nuclei and pale cytoplasms (30/46) except in unclassifiable high-grade plasmacytoid lymphomas, and frequent residual plasma cells in the background displaying the normal residual B maturation.
Epidemiologic, clinical, and biological features
The main results are listed in Table 5.
Clinical and biological features of canine T-cell lymphomas according to the different morphological subtypes.
T-cell lymphomas occurred in animals from 1 to 12 years old, with a median age of 6.9. The sex ratio of males in relation to females was 1.37 for a reference population of 1.04. A breed predilection was found for boxers only (15/46).
The animals were referred for lethargy, weakness, weight loss, peripheral lymph node enlargement, cutaneous disease, and polyuria-polydipsia of variable duration. Time before establishment of diagnosis was longest (several months) for MF and small clear cell lymphoma cases. At the time of diagnosis, the most common finding was a generalized lymphadenopathy (28/46). A regional lymphadenopathy was found in 16/46 cases.
Hepatosplenomegaly occurred in 15/48 cases and mainly in the high-grade subtypes. A mediastinal mass was found in 9/46 cases, mainly in cases of lymphoblastic lymphomas (6/11) and in 3 cases (3/5) of unclassifiable high-grade plasmacytoid lymphomas. Extranodal involvement occurred in 11/46 cases. The skin was involved in six cases (5/5 mycosis fungoides, 1 pleomorphic large cell lymphoma), coelomic cavity in the form of effusion in three cases, bone in one case of unclassifiable high-grade plasmacytoid lymphoma, and the nasal cavities in one case of pleomorphic small-cell lymphoma.
Blood cytopenias, in term of anemia and/or thrombocytopenia, occurred in 16/46 cases. Leukemia, characterized by involvement of bone marrow and blood, occurred in the course of the disease in 8/46 cases, mainly in lymphoblastic lymphomas (6/11 cases) and in two cases (2/5) of unclassifiable high-grade plasmacytoid lymphomas.
Biochemical profile was generally unremarkable except for evidence of hypercalcemia in 16/46 cases without correlation either to a particular subtype (see Table 5) or to a particular clinical presentation. However, hypercalcemia was never found in mycosis fungoides and small clear-cell lymphomas.
Discussion
Determination of the immunophenotype of malignant lymphomas in humans has become an essential step in their classification because of the relation to biological behavior and response to therapy. Consequently, the updated Kiel, 52 REAL, 34 and WHO 74 classifications all differentiate lymphomas based on immunophenotype. The most recent statistical studies carried out on a high number of cases estimate the frequency of human T-cell lymphomas in Europe and in the USA at 12% of cases. 86 In Japan, this frequency is much greater (around 50%) given the particular incidence of adult T-cell lymphoma/leukemia (ATL/L) associated with HTLV-I. 87 As the standardization of the specificities of the different antibodies is relatively recent, 14 very few studies in dogs differentiate between T- and B-cell lymphomas.
Indeed, only two studies considering the immunophenotype on a significant number of cases are published, showing 37.9% 85 and 26% 23 of T-cell lymphomas. In this article, the incidence of canine T-cell lymphomas (32.8%) appears slightly higher than in our previous study (26%). 23 This percentage of 32.8% is closer to those of Teske (37.9%). 85 In that study, however, the use of antibodies such as F3.20.7 55 or CA1-4G8, 61 which are not truly specific for the T-lineage, 10,14 may be responsible for an overestimation of T-lymphomas. Nonetheless, these percentages are clearly higher than that of T-cell lymphomas in humans outside of Japan. It is obvious that several additional studies of numerous cases 91 are necessary to fully establish the percentage of T-lymphomas in the dog. If such a frequency of T-cell lymphomas in the dog is confirmed in the future, it would raise the question of a possible incidence of one or several environmental agents on T-cell neoplasia development. In human medicine, it is well established that the adult T-cell lymphoma/leukemia is associated with infection by HTLV-I 41,89 and the nasal T-cell lymphomas with the Epstein-Barr virus. The association with the Epstein-Barr virus is also suspected for angioimmunoblastic T-cell lymphomas. 33,46,95 Until now, no retrovirus has been demonstrated to be associated with lymphomas in the dog. Recently, the finding of canine T-cell neoplasias producing retroviral particles, 26,27 however, suggests the involvement of virus in canine T-cell lymphomagenesis.
Besides the global incidence of T-cell lymphomas, the relative frequency of the different subtypes is important to consider. In our present study, the most frequent subtypes are the pleomorphic mixed and the lymphoblastic lymphomas. The updated Kiel classification now separates the pleomorphic low-grade small-cell lymphomas from the high-grade mixed- and large-cell pleomorphic lymphomas. However, the pleomorphic lymphomas fall into the same category as the unspecified peripheral T-cell lymphomas of the REAL and the upcoming WHO classifications (Table 6). By gathering all the pleomorphic lymphomas in our study, they could constitute a major category within the T-cell lymphomas in the dog (43.5%). This would be in agreement with Teske et al. and with our previous study and also with current human data (around 58% of T-NHLs). 74,86
Comparison between REAL (34) and UPDATED WHO (74) classifications of human non-Hodgkin's lymphoid tumors.
The proportion of lymphoblastic lymphomas seems to be clearly higher (11/46, 23.9%) than found in our previous study (1/24, 4.2%) and in all previous veterinary studies performed with or without determination of immunophenotype 28,85 except Carter's (49/285, 17.2%). 8 It is close to the percentage of human adult T-lymphoblastic lymphomas (around 17% of the T-NHLs). 81 The average age (5.3 years) of the affected animals does not allow any comparison with the incidence of lymphoblastic lymphomas in children. In terms of incidence, among the other categories of T-cell lymphomas, the high-grade unclassifiable plasmacytoid lymphomas, the small clear-cell lymphomas and the mycosis fungoides remain less frequent, as in our previous study.
Regarding the classification of various morphological subtypes, it is important to notice that, as in human medicine, the classification of all lymphomas including T-cell types is constantly developing. The updated Kiel classification 52 was the first human system based on immunophenotype and morphology separating low-grade from high-grade lymphomas without full reference to the worse prognosis for T-cell lymphomas. 17 Further classifications were based on histologic, immunologic, and genetic features of the different entities, on their clinical presentation and course and postulated normal counterpart in the immune system. The different categories of T-cell neoplasias that have been defined by the REAL classification 34 and the updated WHO classification project 74 are very similar and are shown in Table 6. Despite the obvious clinicopathologic interest of these new classifications, the lack of sufficient epidemiologic, phenotypic, and genetic information in dogs causes difficulties in adapting these classifications. Thus, we have chosen to use the updated Kiel classification to first compare the morphologic characteristics of canine T-cell lymphomas to those of the various subtypes of human T-cell lymphomas and then to try a comparison with the main categories of the new classifications. In this article, the general morphologic appearance of T-cell lymphomas shows great similarities with our previous study. 23 This confirms, on the whole, the good correlation between morphology and immunophenotype. This should allow the pathologist to suspect a T-phenotype by cytohistologic analysis in dogs just as it is possible in humans, except for lymphoblastic and unclassifiable high-grade plasmacytoid lymphomas, even if the immunophenotyping appears absolutely necessary to classify the lymphoma as a B- or a T-cell one. On the architectural level, since T-cell lymphomas are generally diffuse, the T-cell zone pattern is suggestive of phenotype in small clear-cell lymphomas along with the presence of microvascular hyperplasia seen in 47.8% of all T-cell lymphoma cases in all subtypes, except in lymphoblastic lymphomas. Cytologic criteria suggestive of T-phenotype either on cytologic or histologic examination are irregular nuclear outlines, abundant pale cytoplasm sometimes containing azurophilic granules, and a background of normal plasma cells. These observations contrast with those of Teske and collaborators, who did not find any correlation between morphology and immunophenotype except for the presence of numerous plasma cells in canine T-cell lymphomas. 85 This difference may be explained by the fact that the use of some nonspecific T-cell markers led to an incorrect interpretation of the morphologic data for the whole study. These correlations between morphology and phenotype are in agreement with those observed in human medicine on the basis of very similar morphologic criteria. 20,39,40,42,52,81 Regarding the various subtypes of T-cell lymphomas, it is possible to emphasize some morphologic characteristics in order to try a comparison with the different categories of human T-cell lymphomas.
Histologic lesions of mycosis fungoide cases are similar on the whole to those described in previous veterinary studies and in the human species. 18,40,42,57,60,81 However, it is important to point out the loss of epidermotropism and the possible transformation toward a high-grade type in advanced stages, like in humans, leading to some difficulties of classification between cutaneous T-cell lymphomas and mycosis fungoides. 79,81 For this reason, the cytologic criteria, either in cytology or in histology, i.e., cerebriform or convoluted nuclei and large granular lymphocyte aspect particularly obvious on FNA, seem important to consider concomitantly. 23,25,26,57
The category of small clear-cell lymphomas, very suggestive of a T-phenotype because of its characteristic extended, frequently unipolar pale cytoplasm and its T-cell zone pattern, has no definite equivalent with any small T-cell lymphoma in humans. It can be compared, on the one hand, with the T-zone lymphoma of the Kiel classification on the basis of its histologic pattern, but it appears cytologically different from this latter, which mixes numerous small monomorphic and pleomorphic cells and T-immunoblasts. It could be similar to the T-prolymphocytic lymphoma/leukemia (T-PLL) of the Kiel, REAL, and WHO classifications on the basis of the presence in some cells of a clearly visible nucleolus and of a frequent T-cell zone pattern as in human T-PLL. However, it is very important to emphasize the noticeable difference, which is the lack of rapid and progressive leukemia in the small clear-cell lymphomas in dogs compared with the T-PLL in humans. Moreover, in our previous study, we found two cases of T-prolymphocytic lymphomas, which may constitute another rare category of small T-cell lymphomas in the dog, closer to the human T-PLL.
The category of pleomorphic small-cell lymphomas, which is in the minority, may first be compared with that of the Kiel classification, 52,81 as has been done in Teske's study 85 and in our previous study, 23 or to the small-cell types of the unspecified peripheral T-cell lymphomas of the REAL and the updated WHO classifications, which group all the pleomorphic lymphomas.
In the same way, the mixed- and large-cell pleomorphic NHLs show characteristics similar to the corresponding categories of the Kiel classification and of the unspecified peripheral T-cell lymphomas of the REAL and the updated WHO classifications, i.e., the various cellular sizes, the great nuclear irregularity, ranging from slightly indented to serrated to cerebriform, and the abundant pale cytoplasm encountered in some of these lymphomas in humans. 2,40
In theory, lymphoblastic T-cell lymphomas are morphologically difficult to distinguish from their B-equivalents. But, in practice, the presumption of a T-phenotype in lymphoblastic lymphomas is high. The reasons are that all the lymphoblastic lymphomas among the 140 cases of lymphomas that have been morphologically and phenotypically reviewed for this study have a T-phenotype, as does the only lymphoblastic lymphoma of our previous study. If future studies confirm this, it will be in agreement with human data, where 90% of lymphoblastic lymphomas have a T-phenotype. 5,66 From a morphologic point of view, canine lymphoblastic lymphomas appear very similar to their human equivalents.
Only the unclassifiable high-grade plasmacytoid lymphomas constitute a real problem for the morphologic presumption of a T-phenotype since they consist of small, medium-sized, or large blast cells with highly nucleolated nuclei and hyperbasophilic, often plasmacytoid, cytoplasm. The main criteria suggesting a T-phenotype are the presence of some irregular nuclei among the whole population and of normal plasma cells in the background and especially the lack of the various cell components of the polymorphic centroblastic B-cell lymphomas. 23 These unclassifiable high-grade plasmacytoid lymphomas cannot be morphologically compared with any particular category of human T-cell lymphomas but may constitute, in the future, a subtype of the peripheral T-cell lymphomas unspecified of the REAL and the WHO classifications.
Finally, the histologic study in this series did not reveal particular entities like angiocentric, angioimmunoblastic, and anaplastic large-cell lymphomas, which constitute separate entities in humans.
The precise determination of the immunophenotype is essential to identify the various categories of human lymphomas of the most recent classifications (REAL, updated WHO). The main markers, used in human medicine 6,11,16,29,37,38,39,49,92,94 for the precise determination of the immunophenotype of T-cell neoplasms are antibodies against pan-T antigens (CD2, CD3, CD5, CD7), which permit identification of a T-phenotype; antigens restricted to the cortical thymocytes (CD1) or present on the immature lymphoid cells, like the terminal deoxynucleotidyl transferase (TdT) expressed by the lymphoblastic lymphomas (typically CD1+ TdT+); antigens defining functional subpopulations (CD4, CD8) mono-expressed by the lymphomas originating from the peripheral T-cells (CD4+CD8− or CD8+ CD4−) and nonexpressed or, on the opposite, coexpressed by the immature cells of lymphoblastic lymphomas (CD4+CD8+ or CD4−CD8−); specific antigens of T-cell receptor, TCR (α β) or TCR (γ δ), the majority of peripheral T-lymphocytes belonging to the α β subset, whereas only a minority being γ δ T-cells (these latter antigens allow the identification of some rare entities like hepatosplenic γ δ T-cell lymphomas, 34,59,74,81 ); antigens present in the azurophilic granules of cytotoxic cells such as the perforin and the TiA1 protein expressed by the T cytotoxic cell (T CD8+) and natural killer (NK) lymphomas; and finally, activating antigens as the CD30 expressed by the anaplasic large-cell lymphomas (CD30+). In addition to these general considerations, it is important to point out that some categories of T-cell lymphomas in humans more specifically express certain markers, like the mycosis fungoides, which are essentially CD4+, the unspecified pleomorphic lymphomas, which more often have a CD4+ phenotype, than a CD8+ phenotype, and the adult T-cell lymphoma/leukemia, which are usually CD4+. Conversely, they could also lose expression of some pan-T markers.
Therefore, phenotypic heterogeneity of human T-cell lymphomas is remarkable and as such contributes to the diagnosis of T-cell lymphoma subcategories. Unfortunately, the panel of monoclonal and polyclonal antibodies available for the identification of these markers in the dog is still incomplete. Among T-markers, the equivalents of CD3, CD5, CD4, and CD8 molecules, recognized by the corresponding heterospecific or homospecific antibodies, seem to be well established.
Some antibodies against the CD1a molecule recognizing dendritic and Langerhans cells and some cortical thymocytes have been reported in the dog, 63 but their expression in canine lymphoblastic lymphomas has not been reported. In addition, we do not have any reference concerning the possible cross-reactivities between canine and human antibodies against the TdT, the CD30 molecule, and the TiA1. A cross-reactivity against the perforin of the large granular lymphoma cells has been reported in the feline species 45,50 but not, to our knowledge, in the canine species. In our study, we used the anti-CD3 and the anti-CD79a antibodies to determine the T-immunophenotype (CD3+CD79a−) expressed by all T-lymphomas, lymphoblastic lymphomas included. The expression of CD4 and CD8 markers seems to be approximately equivalent between CD8+CD4− and CD4+CD8− lymphomas. There are, however, noticeable differences between the various categories of lymphomas. Canine mycosis fungoides are all CD8+, which is in agreement with most veterinary observations 23,57,60 but differs with human data. The pleomorphic mixed-, small-, and large-cell lymphomas appear mainly CD4+CD8− (85.7%). This agrees with the mature T-cell phenotype most frequently expressed by the human unspecified peripheral T-cell lymphomas (PTCLs). The lymphoblastic lymphomas usually express an immature double negative or double positive phenotype (75%), like in the human lymphoblastic lymphomas, which are CD4+CD8+ or CD4−CD8− in 80% of cases. 5,6,49,81 Finally, the phenotype of the unclassifiable high-grade plasmacytoid lymphomas, mainly CD8+ (CD8+CD4− in 60% of cases, CD8+CD4+ in 40% of cases) but never, until now, CD4+CD8−, constitutes an additional argument to consider them as a particular entity.
From a clinical point of view, peripheral T-cell lymphomas (PTCLs) in humans are clearly distinguished from lymphoblastic lymphomas originating from precursor T-cells. The clinical presentation of the PTCLs is very polymorphous, making the diagnosis sometimes very difficult. 20 According to a recent study concerning one of the largest series of lymphomas, 2 it has been established that a majority of these affect medium-aged adults (on average, 53 years), with a slight predominance of males. 2 A generalized lymphadenopathy, sometimes with moderately enlarged lymph nodes, predominates (68%) and, to a lesser extent, an extranodal invasion (37%), the presence of a mediastinal mass (32%), and bone marrow involvement (25%). The lymphoblastic lymphomas mainly affect children and young adults but may arise at any time and twice as much in men as in women. 64,81 They are generally characterized by a mediastinal mass (57–75%) and the frequent presence of acute leukemia (40–72%). To our knowledge, no study has, up to now, established the epidemiologic and clinical particularities of the T-lymphomas in dogs. The current veterinary data concern all the lymphomas without reference to the phenotype. In our study, the average age (6.9 years) of the dogs affected by a T-cell lymphoma does not show any difference with the one (from 6.3 to 7.7 years) from the previous veterinary studies 3,31,54,70,72,84 and may be compared with the human T-lymphomas data. However, a slight difference of the average age seems to appear between low-grade and high-grade lymphomas, the lymphomas affecting the youngest animals being the unclassifiable high-grade plasmacytoid and lymphoblastic lymphomas and the oldest animals being affected by the mycosis fungoides. Contrary to human medicine but in agreement with most of the previous veterinary data, 3,31,84 there appears to be no sex predilection for the canine T-cell lymphomas in our study. Finally, with respect to the general canine lymphoma data, 3,76,77,84 the overrepresentation of boxers is in agreement with the preestablished data.
The predominating clinical presentation of generalized lymphadenopathy agrees with the general data in canine lymphomas and with the incidence of disseminated forms (stage III–IV) in human T-cell lymphomas. However, the presentation of regional lymph node enlargement, sometimes of small volume (34.8%), must be emphasized since it could constitute, as in humans, a pitfall for early diagnosis. Overall, the frequency of extranodal involvement (23.9%) appears lower than in human T-lymphomas (37.1%) and is largely due to the cutaneous involvement in mycosis fungoides cases. In our series, only one case of bone disorder and one of nasal involvement were encountered. This last case, nonangiocentric at histologic examination, may not be compared with the angiocentric nasal type of the recent REAL and upcoming WHO classifications. This extranodal involvement could be underestimated since extranodal atypical T-cell lymphomas may not be recognized by veterinary clinicians and may not be referred to diagnostic centers. The incidence of a mediastinal mass appears rather low (19.6%) in our study. It is, however, frequent in the unclassifiable high-grade plasmacytoid lymphomas (3/5, 60%) and in lymphoblastic lymphomas (6/11, 54.5%), which agrees with the observation of Carter and collaborators 8 and with the human data. 81
Leukemia is only found in the unclassifiable high-grade plasmacytoid lymphomas (2/5, 40%) and especially in the lymphoblastic lymphomas (6/11, 54.5%), which is also in line with the observation of Carter and collaborators 8 and with the human data.
The other hematologic and biochemical data are not remarkable compared with the general data concerning the canine lymphomas 56 except for the incidence of hypercalcemia (34.8%). This value comes close to the highest percentages of the previous veterinary studies (range 10–40%) 13,54 and confirms the association between hypercalcemia and T-cell lymphomas in dogs, which has already been pointed out by other authors. 28,84,92 In humans, hypercalcemia is much more common in HTLV-I-associated leukemia/lymphomas. 30,34,47
This study showed that gathering all the morphologic, phenotypic, and clinical data during this study might allow identification of new entities and comparison with the human REAL and updated WHO new classifications since these latter consider all the previous data and give up the classification of lymphomas of low and high grade, formerly carried out on the morphologic criteria alone.
According to these criteria, the precursor T-lymphoblastic lymphomas with rapid evolution, affecting dogs of a lower average age, showing frequent leukemic and mediastinal invasions, and characterized by monotonous proliferations of medium-sized cells with a fine chromatin and inconspicuous nucleoli, often double-negative or double-positive CD4CD8, may easily be compared with their human equivalents and seem to be much more frequent than their B-equivalents.
Mycosis fungoides, affecting old animals, now seem to be clearly defined in dogs on the basis of clinical, cytohistologic, and phenotypic data, i.e., skin erythema, ulcers or nodules, epitheliotropism, cleaved or convoluted nuclei, sometimes azurophilic cytoplasmic granules, and, in a majority, a CD8+ phenotype.
The small clear-cell lymphomas evolving slowly and affecting adult or old animals, characterized by an extended unipolar pale cytoplasm and a T-cell zone pattern nonassociated with leukemia, constitute for us a special category of small T-cell lymphomas.
The pleomorphic lymphomas, which predominate in this study, affect middle-aged dogs with a generalized or localized lymphadenopathy and have a diffuse infiltration of small, medium-sized, or large cells with irregular nucleus and clear cytoplasm. They express a phenotype of peripheral T-cells, more often CD4+. These lymphomas should be regrouped in a peripheral T-cell lymphoma unspecified category until categorization of specific variants occurs.
The category of unclassifiable high-grade plasmacytoid lymphomas remains mysterious and needs to be confirmed as a specific entity. In this case, it should define a category of aggressive lymphomas affecting mainly young adults with an occasionally mediastinal and leukemic presentation, a particular plasmacytoid morphology, and a mostly CD8+ phenotype.
To determine the incidence as well as the evolution of the various categories of canine T-cell lymphomas, it is essential to conduct further epidemiologic and statistical studies and to investigate the possible influence of environmental agents on the development of different categories of canine T-cell lymphomas.
Footnotes
Acknowledgements
The authors thank F. Pizzhighini and S. Balleydier for technical assistance, M. Meyer for preparing the manuscript, and M. F. Perron for correcting the manuscript.