Gastrointestinal Stromal Tumors

Transcription

1 Gastrointestinal Stromal Tumors An Immunohistochemical Study of Cellular Differentiation MARKKU MIETTINEN, M.D. Forty-five benign and 11 malignant gastrointestinal stromal tumors (GIST) were immunohistochemically studied for the presence of desmin, muscle actins (MA) and S-1 protein. To facilitate the analysis, the tumors were divided into four groups by light microscopy: (1) typical leiomyomas comparable to peripheral leiomyomas (n = 9); (2) cellular spindle cell tumors (n = 29); (3) round cell tumors ("leiomyoblastomas" n = 7); and (4) sarcomas (n = 11). The typical leiomyomas were desminand MA-positive throughout, and showed well-differentiated smooth muscle cells by electron microscopy, similar to the normal gastric smooth muscle cells. All esophageal leiomyomas belonged to this group. Nineteen of 29 of the Group 2 tumors showed desmin positivity and 2 of 29 showed MA positivity, but usually only in less than 1% of the tumor cells, and in many instances it was very difficult to determine whether the positive cells were real tumor cells or entrapped muscle cells. Only 5 of 29 of Group 2 tumors showed widespread desmin positivity and 11 of 29 showed similar MA positivity. Of round cell tumors, only 1 of 7 showed desminpositive cells and 3 of 7 MA-positive cells. None of the sarcomas showed desmin, while MA positivity was found in 6 of 11 cases, often in a large number of tumor cells. Seven tumors showed a significant number of S-1 positive tumor cells, but four of these also showed a high number of desmin- and MApositive cells, suggesting that these tumors represented complex proliferations of muscle and Schwann cell elements. Two purely S-1 positive benign probably Schwann cell-like tumors were found, both in the small bowel. Small number of S-1 positive cells were commonly found in GIST, and these probably represented entrapped Schwann cells, because many tumors showed simultaneous proliferation of non-neoplastic nerves. (Key words: ; Neurofibroma; Gastric stromal tumor; Desmin; Muscle actin; S-1 protein; Immunohistochemistry; Infrastructure) Am J Clin Pathol 1988;89: MOST gastrointestinal stromal tumors (GIST) of spindle cell or epithelioid appearance have been traditionally viewed as smooth muscle cell tumors: leiomyomas, epithelioid leiomyomas or leiomyosarcomas. 17 Schwann cell instead of smooth muscle cell differentiation has been suggested to be present in a portion of gastric Received July 2, 1987; accepted for publication August 7, Supported by the Finnish Cancer Research Fund and the Paulo Foundation. Address reprint requests to Dr. Miettinen: Department of Pathology, University of Helsinki, Haartmaninkatu 3, SF-29 Helsinki 29, Finland. Department of Pathology, University of Helsinki, Helsinki, Finland "leiomyomas" based on the S-1 protein positivity 16 and the ultrastructural features suggesting Schwann cell differentiation. 416 Further series have shown different results pertaining the frequency of S-1 protein positivity in GIST: there has been consistent S-1 protein positivity in all small bowel stromal tumors, 2 positivity in 5% of GIST, 12 or all tumors have been S-1 protein negative. 21 Immunohistochemistry for desmin, a marker for muscle cell differentiation, has also showed varying results. Most cases of GIST have been reported as desmin-negative, 912 or desmin has been found in all or most of these tumors. 6 ' 21 This study was performed to evaluate the ideas pertaining the differentiation status ("histogenesis") of GIST. Immunohistochemistry and markers relevant to smooth muscle cell (desmin, muscle actins) and Schwann cell differentiation (S-1 protein) were used for this purpose. A monoclonal antibody to muscle actins (HHF35) has been recently introduced and shown to be useful in the distinction between muscle and nonmuscle cell tumors. 23,24 Materials and Methods Fifty-six gastrointestinal stromal tumors were obtained from the files of the Department of Pathology, University of Helsinki. There were 45 benign and 11 malignant tumors. Five of the benign tumors were from the esophagus, 29 from the stomach, and 11 from the small bowel. Of the 11 malignant tumors, two were from the esophagus, two from the stomach, five from the small bowel, and two were intra-abdominal metastases from gastrointestinal tumors. The clinical data of all cases have been summarized in Table 1. There was a high number of incidental tumors which had been found in abdominal operations performed because of other reasons. In many cases, the tumors were found in investigations prompted by gastrointestinal hemorrhage or anemia caused by the tumor. Normal gastrointestinal 61 Downloaded from on 2 February 218

2 62 MIETTINEN AJ.C.P.. May 1988 Table 1. Basic Clinical Data and Immunostaining Results for 56 Gastrointestinal Stromal Tumors Age/S Type Location/Size/Symptoms Des MA S /M 35/F 53/F 44/F 34/M? 6/F 71/F 44/M 87/F 67/F 49/F 78/M 63/F 25/M 49/F 58/F 62/F 62/F 72/F 58/M 19/F 46/M 39/F 63/M 67/F? 75/F 69/M 58/M 66/F 63/F 63/M 65/F 62/M 43/F 47/M 71/M 63/M 6/M 8/F 5/M 58/F 68/F 47/M 33/F 5/F 78/F 4/M 4/F 63/M 59/M 6/F 69/M 7 66/F Esophagus Esophagus Esophagus Esophagus 1.5 cm Esophagus 2 cm ST ST cardia ST serosa 1 X.5 cm, incidental Small bowel ST 3 cm ST cardia, submucous 2 cm ST ST muscular layer 2 cm ST muscular layer 2 X 2 cm ST fundus 4 X 4 cm ST serosa 11 X 1 X 9 cm ST submucous 1 cm ST subserous ST subserous 2 cm ST subserous 5 mm, incidental ST wall 4 cm Duodenum 2 cm Duodenum 3.5 X 2 cm Ileum 1 cm Jejunum 2 cm Jejunum 2.5 X 2 cm Jejunum 4 cm, intestinal obstruction ST ST 1.5 cm, pyloroplasty scar ST Submucous 2 cm ST subserous, incidental ST subserous 1 cm ST subserous 5 mm, incidental ST subserous pedunculated ST subserous, fundus Duodenum 3X2 cm, GI hemorrhage Small bowel, GI hemorrhage Small bowel 2 X 1 cm, incidental ST 3 X 3 cm, hematemesis ST 5 cm ST 5 cm, GI hemorrhage ST pylorus, 2 cm, epigastric pain ST Serosa, pedunculated ST Submucous 3X2 cm, hematemesis Small bowel, GI hemorrhage ST, liver metastasis Small bowel Esophagus, obstruction Peritoneal metastasis ST ST Small bowel Small bowel 2 X 2 cm, obstruction Small bowel Small bowel Esophagus NP NP NP NP NP NP NP NP NP NP ST - stomach; NP = nerve proliferation: GI = gastrointestinal. The immunoreactivity has been graded as follows: = negative: + = single positive cells present; = more than single positive cells present, but less than 1% of the cells positive: = 1-4% of tumor cells positive; = more than 4% of the tumor cells positive. ' neoplasm with palisading pattern. stromal tissues were also investigated for comparative purposes. Formaldehyde-fixed and paraffin-embedded tissue was mainly used for immunohistochemistry. Fresh-frozen sections were available from four cases. Material for electron microscopy was obtained from formalin in 4 cases, and taken from paraffin blocks in 12 cases. Small tissue pieces were deparaffinized and rehydrated. After Downloaded from on 2 February 218

3 Vol. 89 No. 5 GASTROINTESTINAL STROMAL TUMORS 63 this step all samples were postfixed in 1.5% phosphatebuffered osmium tetroxide, dehydrated in alcohol, and embedded in LX-112 resin. The thin sections were poststained in uranyl acetate and lead citrate. For immunohistochemistry, immunoperoxidase technique and the biotin-avidin amplification system was used. 13 Frozen sections were fixed in acetone (ten minutes). Paraffin sections were deparaffinized and treated in.5% pepsin (Crude pepsin preparation, Merck & Co, Darmstadt, FRG) in HC1, ph 1.8, 2-3 minutes at +37 C, since this treatment has been empirically found to enhance the immunoreactivity of the pertinent antigens in formalin-fixed material and to make the results comparable to those obtained in frozen sections. The antibodies, their sources and dilutions used in this study have been summarized in Table 2. The endogenous peroxidase was blocked with.5% hydrogen peroxide in water (five minutes at +22 C). The primary antibody was incubated overnight at +4 C. Biotinylated rabbit antimouse immunoglobulin antiserum or biotinylated goat antirabbit immunoglobulin antiserum were used as secondary antibodies, followed by avidin combined in vitro with biotinylated horseradish peroxidase. The incubation time for both of these reagents was 3 minutes at +22 C. The color was developed with 3-amino 9-ethylcarbazole (Sigma Chemicals, St. Louis, Mo.,.2 mg/ml in.5 mol/l sodium acetate buffer, ph 5., for 2 minutes). Results Normal Gastrointestinal Tissue Both the mucosal muscular layer and the proper muscular layer showed uniform and strong immunoreactivity for desmin and muscle actins (MA) as studied with the HHF35 antibody. The muscular layers of the larger vessels were also positive for both markers. In addition, a single cell layer around many small vessels, compatible with pericytes, was positive with the antibody to MA but not with antidesmin. S-1 protein immunoreactivity was found in the submucous and intramuscular nerve plexuses, and numerous scattered thin nerve related cells were revealed within the muscular layer which itself was negative. General Classification of the Gastrointestinal Mesenchymal Tumors The main results of the 56 GISTs have been summarized in Table 1. By light microscopy, these tumors were divided into four groups to facilitate the analysis of results: (1) typical leiomyomas morphologically similar to leiomyomas of peripheral soft tissues (n = 9); (2) cellular spindle-cell tumors (n = 29); (3) round cell tumors with Table 2. Antibody Sources and Dilutions Used in This Study Antibody (reference) Source Dilution Desmin (5) Amersham Inc, London, UK 1:1 Muscle actins (23) Enzo Biochem Inc, New 1:5 York, NY (HHF35 antibody) S-1 protein Dakopatts Inc, Copenhagen, 1:3 Denmark Rabbit anttimouse Dakopatts 1:3 immunoglobulins, biotinylated Goat antirabbit Dakopatts 1:3 immunoglobulins, biotinylated Avidin + biotinylated Dakopatts 1:16 horseradish peroxidase features of leiomyoblastomas (n = 7); (4) sarcomas defined as such by the presence of pronounced mitotic activity or by the fact that the tumor was a metastasis (n = 11). None of the tumors could be classified as neurofibromas or schwannomas by the generally accepted criteria for peripheral Schwann cell tumors. 8 Group 1. Typical s. Nine typical leiomyomas were found. All esophageal benign tumors (5/5) but only 3/33 of gastric and 1/11 of benign small bowel tumors belonged to this category. Two of three of the typical gastric leiomyomas were located at the cardia adjacent to the esophagus. The typical leiomyomas showed histological characteristics comparable with leiomyomas of other tissues, and they could be best compared with solitary genital leiomyomas in the sense the the tumors were sharply demarcated and closely resembled normal smooth muscle cells. The tumor cells thus had blunt edged "cigarshaped" nuclei and yellow (picrinophilic) cytoplasm in Weigert van Gieson staining and deeply eosinophilic cytoplasm in hematoxylin and eosin staining, and the tumors had a whorled general appearance (Fig. 1). In immunostaining, these tumors showed a uniform positivity for both desmin and muscle actin (Fig. 1). Electron microscopy (six cases) showed well-differentiated smooth muscle cells with abundant bundles of microfilaments with focal densities, and these cells typically were surrounded by a prominent basal lamina, often with adjacent pinocytic vesicles at the cell membrane (Fig. 1). Group 2. Cellular Spindle Cell Tumors (n = 29). This was the largest group of gastrointestinal stromal tumors and included 19 of 29 gastric tumors and 9 of 11 of small bowel tumors. This group includes the tumors designated as cellular leiomyomas by Appelman and Hel- Downloaded from on 2 February 218

4 64 MIETTINEN Downloaded from on 2 February 218 A.J.C.P. May 19

5 Vol. 89 No. 5 GASTROINTESTINAL STROMAL TUMORS 65 FIG. 1. A (upper, left).'a gastric leiomyoma ("true leiomyoma") shows smooth muscle cells in a whorling pattern. Hematoxylin and eosin (X2). B (upper, right). The tumor cells show consistent desmin positivity (X2). C (lower, left). Electron microscopy reveals microfilaments with prominent focal densities (X185). FIG. 2 (lower, right). A highly cellular gastric spindle cell tumor has sharp edged nuclei. This tumor was negative for both muscle markers and S-1 protein Hematoxylin and eosin (X2). wig.2 These tumors had only scanty or no cytoplasmic picrinophilic material, and the tumor cells were closely apposed. The nuclei tended to have sharp rather than blunt edges (Fig. 2). By immunohistochemistry, pronounced desmin positivity (in more than 1% of tumor cells) was found in five cases and muscle actin positivity of a similar degree in nine cases. A minor component of desmin- and MApositive cells was common (2/29 and 22/29, respectively), but it was often difficult to evaluate whether these cells belonged to the neoplastic population or whether they represented entrapped gastric smooth muscle cells (Fig. 3). This distinction was especially difficult in the case of small tumors. Single scattered S-1 protein positive cells were commonly seen (15/29 cases), and 1 of these 29 tumors also showed proliferating bundles of entrapped nerves; this gave the impression that the minor S-1 protein positive component most likely represent entrapped nerve elements (Fig. 4). Five cellular spindle cell tumors revealed a significant number of S-1 protein positive neoplastic cells. One of these tumors was negative for both markers of muscle FIG. 3 (left). The tumor cells in a gastric spindle cell neoplasm are virtually desmin-negative, whereas entrapped smooth muscle cells are positive. Note that the single desmin-positive cells among the tumor cells probably represent direct continuation from the strand of muscle cells at the margin of the tumor (X2). FIG. 4 (right). A gastric spindle cell tumor shows scattered S-1 protein positive cells which probably represent entrapped Schwann cells or other nerve related cells. Note that also the non-neoplastic muscle (lower right) contains a large number of S-1 protein positive cells and cell processes. The margin between the tumor and the muscle has been marked by arrowheads (X2). Downloaded from on 2 February 218

6 66 MIETTINEN A.J.C.P. May 1988 cells (Fig. 5). Prominent nuclear palisades were found in seven spindle cell tumors, and three of these showed a significant number of S-1 protein-positive cells, two in a geographic pattern (Fig. 6). There was also a high number of desmin and muscle actin positive cells in a spotty pattern suggesting a complex cellular differentiation pattern (Fig. 6), or that either the S-1 protein-positive or the desmin- and MA-positive component represented entrapped proliferating non-neoplastic cells. To allow comparison between palisading and nonpalisading spindle cell tumors, these have been specifically identified in Table 1. By electron microscopy, the spindle cell tumors (eight cases, all S-1 protein negative) typically consisted of primitive-appearing spindle shaped or oval cells without evidence of smooth muscle differentiation; bundles of actin filaments, as would be seen in leiomyoma, were not found. Interposed pools of collagen fibers were common. Group 3. The Round Cell Tumors (n = 7). This group included tumors designated as epithelioid leiomyomas by Appelman and Helwig, 3 but there were also round cell tumors lacking an epithelioid appearance. Vacuolated cells were common. Immunostaining for desmin was weakly positive in two cases showing small cytoplasmic desmin-positive dots. Muscle actin was present in small amounts in a similar pattern in three cases; three cases were negative for both muscle markers. A significant number of S-1 protein positive cells was present in one round cell tumor of the duodenum which was completely negative for desmin and MA; the other round cell tumors were negative. Group 4. Malignant Mesenchymal Tumors (n = 11). These tumors were highly cellular and showed a moderate or high number of mitosis which was used as the main criteria for the malignant diagnosis. Ten of these tumors were of spindle cell, and one of round cell appearance. Desmin positivity was not found in any of the tumors including two cases studied in frozen sections; only the normal entrapped smooth muscle cells were positive. However, muscle actin immunoreactivity was present in five cases in a significant number of tumor cells (Fig. 7). In one case EM showed some cells with cytoplasmic dense bodies resembling those observed in smooth muscle cells. Another, a marker-negative malignant GIST failed to show specific differentiation upon EM. Scattered S-1 protein positive cells were commonly present (5 of 11 cases), but most likely these represented entrapped nerve elements. Discussion The present results show a complex immunohistological spectrum in GISTs. Tumors with a clear cut and simple differentiation pattern seem to be rare, and a large number of cases are difficult to classify by the marker profile, either because no markers are expressed, or because both muscle and Schwann cell markers are expressed at the same time. Morphologically typical leiomyomas, similar to leiomyomas of peripheral tissues, especially the sharply demarcated nonvascular solitary genital leiomyomas, seem to be relatively rare but are seen in esophagus and notably also in the cardia of the stomach. This kind of tumors were consistently associated with an electron microscopic appearance typical of a leiomyoma and diffuse positivity for two markers of muscle tissue, desmin and MA. Although the previous investigators have stated that it is not possible to differentiate between the diffusely desmin-positive and less positive or desminnegative GIST, 21 the impression gained in this study was that these tumors in fact also look different. Typically blunt-edged nuclei, deeply picrinophilic cytoplasm in van Gieson staining and whorled pattern of growth are the characteristics of "true leiomyomas". Why the esophageal stromal tumors are "true leiomyomas" but gastric and small bowel tumors only rarely so, is not easy to explain because the normal esophagus and stomach have a similar consistently desmin- and MA-positive smooth muscle layer. Uniform desmin positivity of the esophageal leiomyomas has been reported before by three groups of authors. 9,2 ' 21 Most GIST of spindle cell appearance resist classification by the markers. A significant portion of these tumors seem to be negative for both desmin and MA, the two markers for muscle cell differentiation used in this study. Small numbers of desmin and MA-positive cells were often found. In small tumors it is, however, difficult to make a distinction between the entrapped normal smooth muscle cells and tumor cells. Although there would be subtle cytologic difference between these cell types, the strongly positive immunostaining tends to obscure the cytologic details. It is of interest that smooth muscle myosin was not found in gastrointestinal leiomyomas, although leiomyomas of other organs were positive. 7 S-1 protein could not either be found in the desmin- and MA-negative tumors thus giving no support to the idea that the tumors negative for muscle markers would be Schwann cell neoplasms. Electron mi- > FIG. 5. A small bowel spindle cell tumor (A, upper, left) shows S-1 pr< stein positivity in a major portion of tumor cells (B, upper, right). Desmin (C, lower, left) and muscle actin (D, lower, right) immunoreactivity is i seer only in entrapped muscle cells. Notice that the antibody to muscle actins also stains the pericytes of vessels. Hematoxylin and eosin (X2, upper left; others X4). Downloaded from on 2 February 218

7 GASTROINTESTINAL STROMAL TUMORS Downloaded from on 2 February

8 A.J.C.P. May 1988 MIETTINEN 68 * *?«- ;>. * y % - ' *m> * <» ' **%?«* <i v i * sjl^- Downloaded from on 2 February 218 s V - ^

9 Vol. 89 No. 5 GASTROINTESTINAL STROMAL TUMORS 69 FIG. 6. A gastric spindle cell tumor with some nuclear palisading of the tumor cells (A, upper, left) Hematoxylin and eosin (X4) shows a high number of S-1 protein positive cells in a geographic pattern (B, upper, right; X2). Groups of muscle actin-positive cells are present in a spotty pattern (C, lower, left; X2). FIG. 7 (lower, right). A gastric spindle cell sarcoma shows muscle actin immunoreactivity in a high number of tumor cells. (X4). croscopic observations made in this and a previous study 28 do not give any support for smooth muscle or Schwann cell nature in the majority of gastric cellular spindle cell tumors. Some reports 1518 conclude that signs of smooth muscle differentiation can be found in gastric leiomyomas or leiomyosarcomas, sometimes after wide sampling and prolonged search. However, also electron microscopy is subject to the difficulty of mixing up the entrapped normal cells with tumor cells; therefore, the focal presence of smooth muscle cells does not necessarily indicate that the tumor cells have those differentiation properties. Another explanation for the muscle marker negativity in GIST might be that some normal smooth muscle cells may lack desmin; this has been found in aortic wall smooth muscle." 19 However, because all normal gastrointestinal smooth muscle seem to be positive, it is dubious whether such a theory would operate in GIST. It is as well possible that many tumors may contain desmin or muscle actins in amounts below the detection limit of the immunohistochemical methods, or that the tumors represent neoplastic cells which have lost the expression of desmin and/or muscle actins. Still another possibility is that desmin and MA immunoreactivity would be lost in formalin-fixed and paraffin-embedded material. This may be possible in selected cases but it is unlikely to account for all marker negativity, because all cases had good positive internal controls in non-neoplastic smooth muscle tissue. Antibody to muscle actins revealed postivity in some cases of GIST in which desmin was not observed, especially in malignant tumors, suggesting that this antibody might detect a less differentiated state of muscle cells than antidesmin does. A similar conclusion was reached by Tsukada and associates 24 in relation to rhabdomyosarcomas, which they found to be more often positive for MA than to desmin. There have been scattered reports of Schwann cell tumors of the gastrointestinal tract, but it was Mazur and Clark 16 who first showed objective documentation on Schwann cell-like differentiation of gastric stromal tumors by the presence of S-1 protein positivity in some of these tumors in analogy with the S-1 protein positivity in peripheral neurofibromas and schwannomas. 27 It is wery logical that S-1 protein positive tumors occur in the gastrointestinal tract, because the Schwann cells and the so called enteric glial cells are positive for S-1 protein. 1 In rat the "enteric glia" displays in addition glial filament immunoreactivity, 14 while in human it does not. 1 The cells associated with the autonomic innervation of the gut probably include several types and are still incompletely characterized. 114 This inevitably limits the understanding of their putative neoplasms. In the present study, seven tumors with a significant component of S-1 positive cells were found, suggesting that tumors related to the enteric nervous system exist among GIST. Two cases with significant S-1 protein positivity also had a significant number of muscle cells between the neural elements which suggests that a complex mixed differentiation pattern is present in some gastrointestinal stromal tumors, or that these tumors are either Schwann cell neoplasms with entrapped muscle cells, or leiomyomas with entrapped and proliferating nerve elements. Which of these alternatives is true is hard to define. The gastrointestinal tumors with S-1 protein immunoreactivity are not easily to compared with the peripheral Schwann cell tumors: they are identical neither with neurofibromas nor Schwannomas, and might be enteric glia tumors. Very recently a small series of peculiar spindle cell type of GIST have been described and named to gastric autonomic nerve tumor (GAN). 25 The foundations for this designation were the presence of neuroendocrine-like granules and absence of smooth muscle cell properties. However, these tumors were S-1 protein negative. 25 The existence of nerve-related tumors among GIST might also thought to be supported by the occurrence of GIST in patients with Recklinghausen's disease; however such tumors may have muscle cell properties by ultrastructure. 22 Gastrointestinal mesenchymal tumors with round cell pattern were previously classified as leiomyoblastomas, now most appropriately as epithelioid leiomyomas or leiomyosarcomas. Also these tumors seemed to be primitive in terms of the markers used in this study: in many tumors, neither muscle cell nor Schwann cell markers could be immunohistologically detected. Knapp and co-workers 15 studied by electron microscopy a large series of leiomyoblastomas and concluded that a small minority of tumor cells resemble normal smooth muscle cells suggesting that these tumors have smooth muscle differentiation properties; in malignant tumors with the epithelioid pattern, features typical of smooth muscle cells were absent. 26 Interestingly, one epithelioid tumor in this study was strongly positive for S-1 protein while being negative for desmin and muscle actin. Such a case may therefore be an example of an epitheli- Downloaded from on 2 February 218

10 61 MIETTINEN A.J.C.P.«May 1988 oid Schwann cell or autonomic nerve tumor. Larger materials with integrated light microscopic, ultrastructural and immunohistochemical approach should be studied to clarify the taxonomy of Schwann cell or autonomic nervous types of GIST. In conclusion, only a small fraction of gastrointestinal stromal neoplasms are readily verified as smooth muscle cell or Schwann cell tumors by the use of differentiation markers, and a large fraction of these neoplasms seem to be poorly differentiated. Acknowledgments. The skillful technical assistance of Marjatta Miettinen and Kirsti Tuominen is gratefully acknowledged. References 1. Achtstatter T, Moll R, Anderson A, Kuhn C, Pitz S, Schwechheimer K, Franke WW: Expression of glial filament protein (GFP) in nerve sheaths and non-neural cells re-examined using monoclonal antibodies, with special emphasis on the co-expression of GFP and cytokeratins in epithelial cells of human salivary gland and pleomorphic adenomas. Differentiation 1986;25: Appelman H, Helwig EB: Cellular leiomyomas of the stomach in 49 patients. Arch Pathol Lab Med 1977;11: Appelman HD, Helwig EB: Gastric epithelioid leiomyoma and leiomyosarcoma (leiomyoblastoma) Cancer 1976:38: Brown EF, Banner BF, Gould VE: Differential diagnosis of gastrointestinal schwannomas and leiomyomas: a detailed histologic study with electron microscopic correlation. Lab Invest 1984;5:7A. 5. Debus E, Weber K, Osborn M: Monoclonal antibodies to desmin, the muscle-specific intermediate filament protein. EMBOJ 1983;2: Denk H, Krepler R, Artlieb U, Gabbiani G, Rungger-Brandle E, Leoncini P, Franke WW: Proteins of intermediate filaments: An immunohistochemical and biochemical approach to the classification of soft tissue tumors. Am J Pathol 1983;11: Donner L, de Lanerolle P, Costa J: Immunoreactivity of paraffinembedded normal tissues and mesenchymal tumors for smooth muscle myosin. Am J Clin Pathol 1983;8: Enzinger FM, Weiss SW: Soft tissue tumors. St Louis. CV Mosby, 1983, pp Evans DJ, Lampert IA, Jacobs M: Intermediate filaments in smooth muscle tumours. J Clin Pathol 1983;36: Ferri G-L, Probert L, Cocchia D, Michetti F, Marangos PJ, Polak JM: Evidence for the presence of S-1 protein in the glial component of the human enteric nervous system. Nature 1982;297: Gabbiani G, Schmid E, Winter S. Chaponnier C. de Chastonay C, Vandekerckhove J, Weber K, Franke WW: Vascular smooth muscle cells differ from other smooth muscle cells: predominance of vimentin filaments and a specific a-type actin. Proc Nat Acad Sci USA 1981;78: Hjermstad BM, Sobin LH, Helwig EB: Stromal tumors of the gastrointestinal tract: myogenic or neurogenic? Am J Surg Pathol 1987;11: Hsu S-M, Raine L, Fanger H: Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: A comparison between ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 1981;29: Jessen KR, Mirsky R: Glial cells in the enteric nervous system contain glial fibrillary acidic protein. Nature 198,286: Knapp RH, Wick MR, Goellner JR: Leiomyoblastomas and their relationship to other smooth-muscle tumors of the gastrointestinal tract: An electron microscopic study. Am J Surg Pathol 1984;8: Mazur MT, Clark HB: Gastric stromal tumors. Reappraisal of histogenesis. Am J Surg Pathol 1983;7: Ming S-C: Tumors of the esophagus and stomach. Second series. Fascicle 7 Washington, D.C., Armed Forces Instute of Pathology, 1973, pp Nevalainen TJ, Linna MI: Ultrastructure of gastric leiomyosarcoma. Virchows Arch (Path Anat Histol) 1978;379: Osborn M, Caselitz J, Weber K: Heterogeneity of intermediate filament expression in vascular smooth muscle: A gradient in desmin positive cells from the rat aortic arch to the level of the arteria iliaca communis. Differentiation 1981;2: Pike A, Appelman H, Lloyd RV: Differentiation of gut stromal tumors: an immunohistochemical study. Lab Invest 1986;54:5A. 21. Saul SH, Rast ML, Brooks JJ: The immunohistochemistry of gastrointestinal stromal tumors. Evidence supporting an origin from smooth muscle. Am J Surg Pathol 1987; 11: Schaldenbrand JD, Appelman HD: Solitary solid stromal gastrointestinal tumors in von Recklinghausen's disease with minimal smooth muscle differentiation. Hum Pathol 1984; 15: Tsukada T, Tippens D, Gordon D, Ross R, Gown AM: HHF35. a muscle-actin-specific monoclonal antibody: I. Immunocytochemical and biochemical characterization. Am J Pathol 1987;126: Tsukada T, McNutt MA, Ross R, Gown AM: HHF35, a muscleactin-specific monoclonal antibody: II. Reactivity in normal, reactive, and neoplastic human tissues. Am J Pathol 1987;127: Walker P, Dvorak AM: Gastrointestinal autonomic nerve (GAN) tumor: Ultrastructural evidence for a newly recognized entity. Arch Pathol Lab Med 1986;11: Weiss RA, Mackay B: Malignant smooth muscle tumors of the gastrointestinal tract: An ultrastructural study of 2 cases. Ultrastruct Pathol 1981:2: Weiss SW, Langloss JM, Enzinger FM: Value of S-1 protein in the diagnosis of soft tissue tumors with particular reference to benign and malignant Schwann cell tumors. Lab Invest 1983;49: Welsh RA, Meyer AT: Ultrastructure of gastric leiomyoma. Arch Pathol 1969;87: Downloaded from on 2 February 218