Immunohistochemical Characterization of Hemangiopericytomas and Other Spindle Cell Tumors in the Dog

Transcription

1 Vet Pathol 33: (1996) Immunohistochemical Characterization of Hemangiopericytomas and Other Spindle Cell Tumors in the Dog J. PEREZ, M. J. BAUTISTA, E. ROLLON, F. CHACON-M. DE LARA, L. CARRASCO, AND J. MART~N DE LAS MULAS Departamento de Anatomia y Anatomia Patol6gica Comparadas, Facultad de Veterinaria, Universidad de Cbrdoba, Spain Abstract. The immunohistochemical expression of muscle actin has been studied in 45 canine hemangiopericytomas (CHP) using a monoclonal antibody (HHF35) and formalin-fixed, paraffin-embedded specimens. The distribution of vimentin, desmin, cytokeratins, lysozyme, factor VIII-related antigen, S- 100 protein, and glial fibrillary acidic protein was studied both in CHP and in some canine soft-tissue neoplasms (seven fibrosarcomas, seven benign schwannomas, seven benign fibrous histiocytomas, and six leiomyosarcomas) used as controls for differential diagnosis. All CHP and control tumors expressed vimentin. Twenty-three CHP expressed muscle actin, whereas all control tumors analyzed were muscle actin-negative, with the exception of leiomyosarcomas. Among muscle actin- and vimentin-positive CHP, one case could be reclassified as leiomyosarcoma because it was desmin-positive, two cases expressed lysozyme, and nine cases expressed S- 100 protein. Among muscle actin-negative and vimentin-positive CHP, seven expressed S- 100 protein. In addition, S- 100 protein was detected in five schwannomas. All CHP and control tumors analyzed were negative for cytokeratins, factor VIII-related antigen, and glial fibrillary acidic protein. Our results support the hypothesis of a pericytic origin of CHP, and suggest that muscle actin, desmin, vimentin, and lysozyme could be useful for the differential diagnosis of canine spindle cell tumors, but not all these neoplasms can be identified with these tumor tissue markers. Key words: mor markers. Canine hemangiopericytoma; immunohistochemistry; intermediate filaments; muscle actin; tu- Hemangiopericytomas (HP) are soft-tissue neoplasms of human beings and dogs thought to arise from pericytes. In human beings, the pericytic origin of this tumor has been established at the ultrastructural level7 but recent immunohistochemical studies do not support this hyp~thesis.~~,~ In dog, there is controversy about the histogenesis of this tumor because the origin of the tumor cells from pericytes has not been established.9,24,37-39,41 Both human and canine hemangiopericytomas (CHP) are rather uncommon cutaneous neoplasms of adult life and are more often found in the legs and hindlimbs, respectively. Human HP has also been found in noncutaneous locations, whereas CHP has been exclusively described in the ~ kin.~~s~~ In man, the metastatic rate varies from 11.7% to 56.5%, and recurrence is more ~ommon,~ whereas in the dog the recurrence rate varies from 26% to 60%, and when it occurs the tumor becomes more agressive. However, metastasis of CHP is quite ~nc0mm0n.3,9,lo,l6,23.24,26,39 39 I Microscopic features of human HP consist of tightly packed oval or spindle cells arranged around thin-walled sinusoidal vascular channels. Spindle cells are never arranged in bundles or fascicle^.^ In the dog, the typical histopathologic pattern of HP consists of layers of spindle cells arranged in a concentric fashion around a central vessel that is usually devoid of erythrocytes, referred to as fingerprint pattern.9,23,24,37-39 Because pericytes lack readily identifiable cellular features under the light microscope, the histopathologic diagnosis of CHP is based on the recognition of this fingerprint pattern, which is often lacking and may be also found in some other soft-tissue neoplasms such as fibrosarcomas, schwannomas, cutaneous fibrous histiocytomas, and leiomyosarcomas. Consequently, differential histopathologic diagnosis among these spindle cell tumors is very Immunohistochemically, human and murine pericytes have been shown to express vimentin, alpha smooth muscle actin,22,30-32,35 muscle and non-muscle

2 ~~ 392 Perez, Bautista, Rollbn, Chac6n-M. de Lara, Carrasco, and Martin de las Mulas Vet Pathol 33: Table 1. Immunohistochemical staining reactions of canine hemangiopericytomas, fibrosarcomas, schwannomas, fibrous histiocytomas, and leiomyosarcomas with different antibodies. Tumor Type Antibodies HHF35* Vimentin Desmin Lysozyme Keratin F. VIII-rat GFAP$ S-lOO Canine hemangiopericytoma 23/ /45 1/45 2/45 0/45 0/45 0/45 17/45 Fibrosarcoma 0/7 7/ /7 0/7 0/7 0/7 0/7 Schwannoma 0/7 7/7 0/7 0/ /7 0/7 5/7 Fibrous histiocytoma 0/7 7/ / /7 0/7 0/7 Leiom yosarcoma 6/6 6/6 6/ Anti-muscle actin. t Factor VIII-related antigen. * Glial fibrillary acidic protein. 9 S- 100 protein. 11 Number of positive cases/total number of cases analyzed. actin isoforms," myosin,13 and tropomyosin.12 Murine 0.1% pronase (Sigma Chemical, St. Louis, MO) for 10 minpericytes of high endothelium vessels of lymph nodes utes at room temperature. and chicken pencytes additionally express de~min.~.)~ The primary monoclonal and polyclonal antibodies em- The distribution of muscle actin and intermediate filament proteins has been studied in several series of human HP.15,22,30 The antigenic profile of canine pericytes has not been studied, and that of CHP has only been analyzed in isolated The purpose of this work was to analyze the immunohistochemical expression of muscle actin in canine pericytes and CHP using a monoclonal antibody that identifies human pericytes. In addition, the immunohistochemical expression of vimentin, desmin, cytokeratins, factor VIII-related antigen, lysozyme, S- 100 protein, and glial fibrillary acidic protein has also been analyzed in CHP and in some canine softtissue neoplasms that may exhibit the fingerprint pattern. The possible pericytic origin of CHP and the utility of tumor tissue markers for the differential diagnosis of these tumors was evaluated. Materials and Methods Tumor tissue samples that were analyzed came from the Pathology files of the Faculty of Veterinary Medicine of Dublin and the Veterinary School of Cordoba. They included 45 canine hemangiopericytomas, seven fibrosarcomas, seven benign schwannomas, seven benign cutaneous fibrous histiocytomas, and six leiomyosarcomas. All tumor specimens were fixed in 10% buffered formalin and embedded in paraffin wax. Four-micrometer-thick tissue sections were stained with the hematoxylin-eosin histochemical technique and with the silver methenamine method for basal membranes. All tumors were reexamined to confirm the histopathologic diagnose~,~~ and to classify canine hemangiopericytomas according to the histologic grade.23 For the immunohistochemical study, 3-pm-thick tissue sections were deparaffinized and rehydrated. Endogenous peroxidase activity was removed by incubation of the sections with 0.3% hydrogen peroxide in methanol for 30 minutes at room temperature. Tissue sections utilized for the detection of cytokeratins and lysozyme were treated with ployed, their dilutions, and sources were the following: mouse anti-muscle actin (HHF35), 1 : 50 (Enzo Diagnostic Inc., New York, NY); rabbit anti-vimentin, 1 : 200, rabbit anti-desmin, 1 : 50, mouse anti-cytokeratins (RCK 102), 1 : 50, and mouse anti-glial fibrillary acidic protein (GFAP), 1 : 50 (Eurodiagnostics, Appeldorn, The Netherlands); rabbit anti-factor VIII-related antigen (factor VIII-ra), 1 : 800, rabbit anti-lysozyme, 1 : 250, and rabbit anti-s100 protein, 1 : 800 (Dako, Glostrup, Denmark). Tissue sections were incubated with the primary antibodies overnight at 4 C. A prediluted biotinylated goat anti-rabbit immunoglobulin G (Vector Laboratories, Inc., Burlingame, CA) was applied for 30 minutes at room temperature as secondary reagent for polyclonal antibodies. For monoclonal antibodies, a biotinylated goat antimouse immunoglobulin G (Dako) diluted 1 : 20 was applied for 30 minutes at room temperature. After two 5-minute rinses with phosphate-buffered saline (PBS), tissue sections were incubated with the avidin-biotin-peroxidase complex (ABC) (Vector Laboratories, Inc., Burlingame, CA) diluted 1 : 50 in PBS for 1 hour at room temperature. Then, tissue sections were rinsed three times with Tris-buffered saline (TBS), incubated with the chromogen 3,3'-diaminobenzidine tetrahydrochloride (Sigma Chemical) diluted 0.035% in TBS for 1 minute, rinsed in tap water, counterstained with Mayer's hematoxylin, dehydrated, and mounted. Specific primary antibodies substituted with PBS or nonimmune sera were used as negative controls in immunohistochemical techniques. Isotype nonspecific antibodies were used as negative control to standarize the staining reaction of monoclonal antibodies in canine tissues exclusively. Tissue sections of normal canine skin were used as positive control. Results Microscopic examination of tumors previously diagnosed as canine hemangiopericytomas (CHP) revealed different histological patterns. Thus, in addition to typical fingerprint areas, tumor cells arranged in short interlacing bundles, palisades, or diffuse and myxoid patterns were seen in different areas of the

3 Vet Pathol 33:4, 1996 Immunohistochemical Characterization of Hemangiopericytomas 393 same tumor and in different tumors. The histological grade of CHP was determined following the morphologic criteria (cellular pleomorphism, mitotic index, whorling, vascularity, capsulation, intratumoral necrosis, and hemorrhages) previously reported.23 There were 22 grade I, 12 grade 11, 8 grade 111, and 3 grade IV CHP. Soft tissue neoplasms used as controls (seven fibrosarcomas, seven benign schwannomas, seven cutaneous benign fibrous histiocytomas, and six leiomyosarcomas) were all well differentiated. With the silver methenamine histochemical technique, basal membranes were observed around spindle tumor cells in the fingerprint areas of CHP. Similar results were obtained in schwannomas and canine fibrous histiocytomas, whereas neoplastic cells of diffuse areas of CHP showed a poor development of basal membranes. Immunohistochemical staining results are summarized in Table 1. Immunostaining was not found in any of the tissue sections run as negative controls for either monoclonal (Fig. 1) or polyclonal antibodies. HHF35 anti-muscle actin monoclonal antibody (MoAb) reacted with canine pericytes from arterioles, venules, and capillaries located both within the tumor masses and in the surrounding connective tissues (Fig. 2). Pericytes of arterioles and venules showed a more intense staining reaction for muscle actin than those of capillaries. In addition, HHF35 MoAb also stained vascular smooth muscle cells, skeletal muscle fibers, and myoepithelial cells of apocrine glands. Muscle actin (HHF3 5 MoAb) immunoreactive t5- mor cells were observed in 23 CHP (ten with grade I, eight grade 11, and five grade 111). The positive reaction was always seen as a diffuse staining of the cytoplasm. The proportion of positive cells varied from 20% to 80%, and they were located in both fingerprint and solid areas (Fig. 3). All six leiomyosarcomas expressed muscle actin in the majority of the tumor cells. Fibrosarcomas, schwannomas, and cutaneous fibrous histiocytomas did not react with this antibody. All CHP and control tumors reacted with the vimentin polyclonal antibody (PoAb) in a diffuse and intense fashion. Immunoreactivity was observed in the majority of tumor cells. On the contrary, cytokeratins, factor VIII-related antigen (factor VIII-ra), and glial fibrillary acidic protein (GFAP) were not detected in the neoplastic cells of either CHP or leiomyosarcomas, fibrosarcomas, schwannomas, and cutaneous fibrous histiocytomas. Internal controls (epidermis for RCK 102 MoAb, endothelial cells for factor VIII-ra, and nerve endings for GFAP PoAbs) were always positive. In CHP, the endothelium of one or more vessels was stained within some of the fingerprint areas with the factor VIII-ra PoAb. Immunoreactivity with desmin PoAb was found in a high proportion of tumor cells in all leiomyosarcomas. In addition, tumor cells of one CHP, which additionally expressed muscle actin, were also stained (Figs. 4, 5). Immunoreactivity with lysozyme PoAb was observed in two cutaneous fibrous histiocytomas and in two CHP that additionally coexpressed vimentin and muscle actin. In these two CHP, lysozyme antibody staining reaction was found in the cytoplasm of fusiform neoplastic cells (Fig. 6). S- 100 protein PoAb immunoreactivity was found in both nuclei and cytoplasm of a variable proportion (1 0% to 70%) of tumor cells of 17 CHP, which usually were pleomorphic or spindle in shape (Fig. 7). Five schwannomas were also positive for S- 100 protein, whereas the other control tumors analyzed were unreactive. Discussion The results of this study show that canine pericytes and 23 out of 45 canine hemangiopericytomas (CHP) coexpress muscle actin and vimentin, and that, with the exception of leiomyosarcomas, CHP are the only type of tumors analyzed that express muscle actin. The expression of muscle actin by canine pericytes of arterioles, venules, and capillaries of the dermis and subcutaneous tissue and the lack of desmin intermediate filament protein resembles that described for human However, canine and human hemangiopericytomas have a dissimilar expression of muscle actin. Thus, in spite of the fact that human pericytes express muscle actin, human hemangiopericytomas do not. This has been attributed either to the loss of this antigen during differentiation of neoplastic pericytes or to the origin of human hemangiopericytomas from some cell type other than true peri~ytes.~~,~~ In our study, muscle actin was expressed by tumor cells of 23 out of 45 CHP analyzed. The dissimilar expression of muscle actin in human and canine hemangiopericytomas closely reflects the already known differences in histopathology and biological behavior shown by these neoplasms, and it may indicate that the cell of origin of these tumors in human beings and in dogs is different. The expression of muscle actin in CHP showed no correlation with the degree of histological differentiation. Thus, 12 out of 22 well-differentiated (grade I) tumors were muscle actin-negative whereas five out of eight grade 111 CHP were positive. This fact, as well as the lack of smooth muscle actin in 22 CHP, could be attributed to the different degree of smooth muscle differentiation of neoplastic pericytes. It has been suggested that pericytes are pluripotential cells that express different antigens according to their location. I *,30,32 Thus, arteriolar and venular pericytes ex-

4 394 Pkrez, Bautista, Rollon, Chacon-M. de Lara, Carrasco, and Martin de las Mulas Vet Pathol 33:4, 1996 Fig. 1. Negative control for monoclonal antibodies. Note the abscense of immunostaining in a canine hemangiopericytoma with a typical fingerprint pattern. Avidin-biotin-peroxidase complex method. Bar = 35 clm. Fig. 2. Immunoreactivity to muscle actin (HHF35) antibody in smooth muscle cells (+) and normal pericytes (*) from peritumoral connective tissue. Note that the endothelial cells are unreactive (b). ABC method. Bar = 20 pm. Fig. 3. HHF35 antibody. Immunoreactivity to muscle actin is observed in the cytoplasm of some neoplastic cells arranged in a fingerprint pattern. The positive internal control (vascular smooth muscle cells) is shown at the upper left corner. Canine hemangiopericytoma. Avidin-bi- 0th-peroxidase complex method. Bar = 70 pm. Fig. 4. Intense immunoreactivity to muscle actin (HHF35 antibody) in the cytoplasm of neoplastic cells of the tumor that also expressed desmin and was reclassified as leiomyosarcoma. Avidin-biotin-peroxidase complex method. Bar = 35 pm. Fig. 5. Same case as Fig. 4. Immunostaining to desmin antibody within the cytoplasm of some tumor cells. Avidin-biotin-peroxidase complex method. Bar = 35 gm.

5 Vet Pathol 33: Immunohistochemical Characterization of Hemangiopericytomas 395 press predominantly smooth muscle actin, whereas tin, muscle actin, lysozyme, and S-100 protein. Both capillary pericytes express predominantly non-smooth cases were moderately differentiated (grade 11). The muscle actin isoforms.11,22 Similar findings have been expression of lysozyme suggests histiocytic differentireported with respect to the expression of myosin13 and ation. In tissues of mesenchymal origin, lysozyme is tropomyosin. l2 considered a histiocytic marker both in human^^,^^ and The occurrence of the intermediate filament protein dogs. l8 This antigen has been detected in two canine vimentin in all mesenchymal tumors in this study cor- and four feline malignant fibrous histiocyt~mas.~~ related with similar findings by other work- However, the sensitivity of lysozyme as a marker of ers.2,4.5.~4.17,20,25,28,30,34,36 However, CHP and leiomy- canine fibrous histiocytomas was low in our study, as osarcomas were the only tumors that expressed muscle only two out of seven cases studied were positive. In actin. Thus, the expression of muscle actin seems to addition, none of these seven benign fibrous histiobe useful in the differentiation of CHP from fibrosar- cytomas expressed muscle actin. The coexpression of comas, schwannomas, and fibrous histiocytomas. The muscle actin and lysozyme supports the hypothesis expression of muscle actin by these canine soft-tissue suggested by some a~thorsl~,~~ that the cell of origin of neoplasms has not been reported previously. In human some mesenchymal tumors is a pluripotential cell that beings, muscle actin has been detected in five out of may differentiate into several cell types. ten I5 and in one out of 2229 cutaneous fibrous histio- In addition to the single case that coexpressed vicytomas. However, the expression of muscle actin was mentin, muscle actin, lysozyme, and S- 100 protein, 16 considered as a sign of muscle differentiati~n.~~,~~,~~ Hu- CHP expressed S- 100 protein with (nine) and without man fibrosarcomas did not express muscle a~tin,'~,~~ (seven) coexpression of muscle actin. Muscle actinand only one out of 18 human schwannomas expressed positive CHP can be easily differentiated from schwanthis antigen.29 Human leiomyosarcomas express mus- nomas (five out of seven reacted with S-100 protein cle actin15,33,35 but canine leiomyosarcomas have not antibody) but muscle actin-negative CHP have an imbeen analyzed with this marker. However, in a canine munophenotype comparable to that of schwannomas lingual rhabdomyoma, the expression of muscle actin and melanomas. Although S- 100 protein detection has has been considered a more reliable marker of muscle been used in the diagnosis of benign sheath nerve tudifferentiation than desmin expre~sion.~~ mors and me1anomas,5j9,25,28,33 it has been detected in In addition to vimentin, some CHP expressed additional tumor markers. Among muscle actin-positive CHP, one coexpressed desmin. The immunophenotype of this case was similar to that found in the six leiomyosarcomas used as controls. Desmin is considered a very specific marker of muscle differentia- tion.l,4,5j4j5,28 Desmin has been detected in one out of 2 CHP analyzed by Andreasen et al.,2 and in the unique case studied by Thoolen et al.34 However, in these works, desmin was also detected in lymphosarcomas and histiocytomas, and its detection was attributed to nonspecific reactions of the antibodies employed. Other authors did not detect desmin in one CHP studied." On the other hand, desmin has been detected in ma- lignant fibrous histiocytomas of dogs,34 and cats.20,34 In contrast, none of the seven benign fibrous histiocytomas analyzed in our study expressed desmin, which could be attributed to the different degree of differentiation of the tumors in both studies. One CHP of our study coexpressed vimentin, muscle actin, and lysozyme, and another coexpressed vimen- a variety of nonneural and nonmelanogenic tumors such as canine and feline malignant fibrous histiocyto ma^,^^,^^ human breast carcinomas,6 and human hemangiopencytomas.22 Our results confirm the wide distribution of S- 100 protein in neoplastic tissues, limiting its usefulness as tumor tissue marker. Neither CHP nor control tumors cells reacted with antibodies to cytokeratins, factor VIII-related antigen (factor VIII-ra), and glial fibrillary acidic protein (GFAP). The immunoreactivity of CHP with factor VIII-ra polyclonal antibody highlighted the vascular pattern of the tumors and revealed that only a small proportion of tumor cells are close to blood vessels. Thus, some fingerprint areas were centered around a blood vessel, but some others were not. Findings were similar in silver methenamine-stained sections as only some tumor cells appeared to have contact with basal membranes. This is in agreement with the poor development of basal membranes in CHP reported on electron microscopy Some authors have suggested a possible neural origin c Fig. 6. Immunoreactivity to lysozyme antibody within the cytoplasm of some fusiform neoplastic cells (4) arranged in a diffuse pattern of growth. Canine hemangiopericytoma. Avidin-biotin-peroxidase complex method. Bar = 35 pm. Fig. 7. Immunoreactivity to S-100 protein antibody in both nuclei and cytoplasms of isolated pleomorphic tumor cells (+). Canine hemangiopericytoma. Avidin-biotin-peroxidase complex method. Bar = 35 pm.

6 396 Ptrez, Bautista, Rollbn, Chacbn-M. de Lara, Carrasco, and Martin de las Mulas Vet Pathol 33:4, 1996 of CHP because of their common histological features with extracranial meningioma, schwannoma, and neu- In our study, normal canine pericytes from blood vessels of the dermis expressed muscle actin but we did not detect muscle actin in nerve sheaths and Schwann cells from normal peripheral nerves. Human Schwann cells also lack muscle a ~tin.~~ These results do not support the hypothesis of a possible neural origin of CHP suggested by some authors. 9,24.37,39,4 I In conclusion, our results show that immunohistochemical studies are useful in the differential diagnosis of spindle cell tumors in the dog. Acknowledgements We are greatly indebted to Prof. Dr. B. J. Sheahan, from the Department of Comparative Pathology of the Faculty of Veterinary Medicine of the Universitary College of Dublin, for the extensive revision of this work and for his kind procurement of 27 out of the 45 canine hemangiopericytomas used in this study. References 1 Andreasen CB, Mahaffey EA: Immunohistochemical demonstration of desmin in canine smooth muscle tumors. Vet Pathol 24: , Andreasen CB, Mahaffey EA, Duncan JR: Intermediate filament staining in the cytologic and histologic diagnosis ofcanine slun and soft tissue tumors. Vet Pathol , Bostock DE, Dye MT: Prognosis after surgical excision of canine fibrous connective tissue sarcomas. Vet Pathol 17: , Desnoyers MM, Haines DM, Searcy GP: Immunohistochemical detection of intermediate filament proteins in formalin fixed normal and neoplastic canine tissues. Can J Vet Res 54: , Du Bulay CEH: Immunohistochemistry of soft tissue tumours: a review. J Pathol 146:77-94, Dwarakanath S, Lee, AKC, Delellis RA, Silverman ML, Frasca L, Wolfe HJ: S-100 protein positivity in breast carcinomas: a potential pitfall in diagnostic immunohistochemistry. Hum Pathol 18: , Enzinger FM, Weiss SW: Hemangiopericytoma. In: Soft Tissue Tumors, ed. Enzinger FM and Weiss SW, 2nd ed., pp Mosby Company, St. Louis, MO, USA, Fujimoro T, Singer SJ: Immunocytochemical studies of desmin and vimentin in pericapillary cells of chicken. J Histochem Cytochem 35: , Goldschmidt MH, Shofer OFS: Canine hemangiopericytoma. In: Skin Tumors of the Dog and Cat, ed. Gold- Schmidt MH and Shofer OFS, pp Pergamon Press, Oxford, England, Graves GM, Bjorling DE, Mahaffey E: Canine hemangiopericytoma: 23 cases ( ). J Am Vet Med ASSOC 192(1):99-102, Herman IM, D Amore PA: Microvascular pericytes contain muscle and nonmuscle actins. J Cell Biol 101: 43-52, Joyce NC, Haire MF, Palade GE: Contractile proteins in pericytes. 1. Immunoperoxidase localization of tropomyosin. J Cell Biol 100: , Joyce NC, Haire MF, Palade GE: Contractile proteins in pericytes. 11. Immunocytochemical evidence for the presence of two isomyosins in graded concentrations. J Cell Biol 100: , Martin de las Mulas J, Vos JH, Van Mil FN: Desmin and vimentin immunocharacterization of feline muscle tumors. Vet Pathol 29: , Miettienen M: Antibody specific to muscle actins in the diagnosis and classification of soft tissue tumors. Am J Pathol 130(1): , Mills JHL, Nielsen SW: Canine haemangiopericytomas. A survey of 200 tumours. J Small Anim Pract 8: , Moore AS, Madewell BR, Lund JK Immunohistochemical evaluation of intermediate filament expression in canine and feline neoplasms. Am J Vet Res 50( 1):88-92, Moore PF: Utilization of cytoplasmic lysozyme immunoreactivity as a histiocytic marker in canine histiocytic disorders. Vet Pathol 23: , Nakajima T, Watanabe S, Sat0 Y, Kameya T, Hirota T, Shimosato Y: An immunoperoxidase study of S-100 protein distribution in normal and neoplastic tissues. Am J Surg Pathol 6: , Pace LW, Kreeger JM, Miller MA, Turk JR, Fischer JR: Immunohistochemical staining of feline malignant fibrous histiocytomas. Vet Pathol 31: , Pantekoek JFCA, Schiefer B: Metastasising canine fibrosarcoma originally diagnosed as haemangiopericytoma. J Small Anim Pract 16: , Porter PL, Bigler SA, McNutt M, Gown AM: The immunophenotype of hemangiopericytomas and glomus tumors, with special reference to muscle protein expression: an immunohistochemical study and review of the literature. Mod Pathol4( 1):46-52, Postorino NC, Berg RJ, Powers BE, McChesney AE, Taylor RA, Withrow SJ: Prognostic variables for canine hemangiopericytoma: 50 cases ( ). J Am Anim HOSP ASOC 24~ , Pulley LT, Stannard AA: Tumors of the skin and soft tissues. In: Tumors in Domestic Animals, ed. Moulton JE, 3rd ed., pp University of California Press, Berkeley, USA, Rabanal RH, Fondevila DM, MontanC V, Doming0 M, Ferrer L: Immunocytochemical diagnosis of skin tumours of the dog with special reference to undifferentiated types. Res Vet Sci 47: , Richardson RC, Render JA, Rudd RG, Shupe RE, Carlton WW: Metastatic canine hemangiopericytoma. J Am Vet Med Assoc 182(7): , Rivera RYR, Carlton WW. Lingual rhabdomyoma in a dog. J Comp Pathol 106:83-87, Roholl PMJ, De Jong ASH, Ramaekers FCS: Application of markers in the diagnosis of soft tissue tumours. Histopathology 9: , 1985

7 Vet Pathol 33:4, 1996 Immunohistochemical Characterization of Hemangiopericytomas Roholl PJM, Elbers HRJ, Prinsen I, Claessens JAJ, Unnik JAM: Distribution of actin isoforms in sarcomas: an immunohistochemical study. Hum Pathol 21: , Schiirch W, Skalli 0, Lagaci: R, Seemayer TA, Gabbiani G: Intermediate filament proteins and actin isoforms as markers for soft-tissue tumor differentiation and origin Hemangiopericytomas and glomus tumors. Am J Pathol 136(4): , Sims DE: The pencyte-a review. Tissue Cell 18(2): , Skalli 0, Pelte MF, Peclet MC, Gabbiani G, Gugliotta P, Bussolati G Ravazzola M Orci L: Alpha-smooth muscle actin, a differentiation marker of smooth muscle cells, is present in microfilamentous bundles of pericytes. J Histochem Cytochem 37(3): , Stefansson K, Wollmann R, Jenkovic M: S- 100 protein in soft-tissue tumors derived from Schwann cells and melanocytes. Am J Pathol 106: , Thoolen RJMM, Vos JH, Van der Linde-Sipman JS, De Weger RA, Van Unnik JAM, Misdorp W, Van Dijk JE: Malignant fibrous histiocytomas in dogs and cats: an immunohistochemical study. Res Vet Sci 53: , Tsukada T, McNutt MA, Ross R, Gown AM: HHF35, a muscle-actin-specific monoclonal antibody. 11. Reactivity in normal, reactive and neoplastic human tissues. Am J Pathol 127: , Vos JH, Van den Ingh TSGAM, Misdorp W, Ramaekers FCS, Van Mil FN, De Neijs M: An immunohistochemical study of canine tissues with vimentin, desmin, glial fibrillary acidic protein, and neurofilament antisera. J Vet Med Ser A 36: , Walder EJ, Gross TL: Neoplastic diseases of the skin. Part IV. Mesenchymal tumors. In: Veterinary Dermatopathology, ed. Gross TL, Ihrke PJ, and Walder EJ, pp Mosby Company, St. Louis, MO, USA, Weiss E: Tumours of the soft (mesenchymal) tissues. Bull WHO 5O:lOl-110, Wilcock BP: Neoplastic diseases of skin and mammary gland. In: Pathology of Domestic Animals, ed. Jubb KVF, Kennedy PC, and Palmer N, 4th ed., vol. 1, pp Academic Press, Inc, San Diego, CA, USA, Xu F": Ultrastructure of canine hemangiopericytoma. Vet Pathol 23: , Yager JA, Wilcock BP: Spindle cells tumors. In: Color Atlas and Text of Surgical Pathology of the Dog and Cat. Dermatopathology and Skin Tumors, ed. Yager JA and Wilcock BP, pp Mosby Company, London, England, 1994 Request reprints from Dr. J. Perez, Departamento Anatomia y Anatomia Patologica Comparadas, Facultad de Veterinaria, Avda Medina Azahara 9, C6rdoba (Spain).