Osteoclast-like giant cells (GCs) are postulated to derive

Giant Cell Tumors Inquiry Into Immunohistochemical Expression of CD11 (c-kit), Microphthalmia Transcription Factor, Tartrate-Resistant Acid Phosphatase, and HAM-56 Rolando Y. Ramos, MD; Helen M. Haupt, MD; Peter A. Kanetsky, PhD; Rakesh Donthineni-Rao, MD; Carmen Arenas-Elliott, MS; Richard D. Lackman, MD; Anne-Marie Martin, PhD Context. Osteoclast-like giant cells (GCs) in giant cell tumors (GCTs) are thought to derive from a monocytemacrophage lineage. Microphthalmia transcription factor (MITF) is necessary for osteoclast gene expression and tartrate-resistant acid phosphatase (TRAP) activation; c-kit plays a role in regulation of MITF. Objective. To gain insight into the differentiation of GCTs of bone (GCTBs) and GCTs tendon sheath (GCTTSs) by investigating immunohistochemical staining for c-kit, MITF, TRAP, and HAM-56 in the GCs and stroma. Design. Immunoreactivity for CD11 (c-kit), MITF, TRAP, and HAM-56 was studied in 5 GCTBs, 15 GCTTSs, and 5 foreign-body GC controls. Results. Across tumors, MITF and TRAP but not c-kit were generally expressed in GCs; TRAP was variably expressed in stromal cells. The MITF was expressed more consistently in stromal cells of GCTTSs than GCTBs (P.1). The GCTBs showed more intense MITF stromal (P.1) and TRAP GC staining (P.4) than GCTTSs. HAM-56 staining by stromal cells was associated with MITF stromal staining (r 2.6, P.1). Conclusions. Results suggest that MITF and TRAP are expressed during osteoclast differentiation and that a proportion of mononuclear cells in GCTs express the macrophage marker HAM-56. Both GCTBs and GCTTSs show similar patterns of immunohistochemical expression. (Arch Pathol Lab Med. 25;129:6 65) Osteoclast-like giant cells (GCs) are postulated to derive from the monocyte-macrophage lineage, 1 6 and the GCs are believed to differentiate from mononuclear stromal cells. 1 Molecules that are necessary for osteoclastogenesis are macrophage colony-stimulating factor 4,5, and the receptor for activation of nuclear factor B ligand. 4,6,8 Macrophage colony-forming units, under the stimulation of macrophage colony-stimulating factor, proliferate to form monocyte-macrophage cells, which under continued stimulation by macrophage colony-stimulating factor and receptor for activation of nuclear factor B ligand differentiate into osteoclast precursor mononuclear cells, which are tartrate-resistant acid phosphatase (TRAP) positive. 4 These mononuclear cells, under further stimulation by macrophage colony-stimulating factor, receptor for activation of nuclear factor B ligand, and interleukin 1, differentiate to multinucleated osteoclasts. 4 Accepted for publication October 29, 24. From the Departments of Pathology (Drs Ramos, Haupt, Arenas-Elliott, and Martin), and Orthopedic Surgery (Dr Lackman), Pennsylvania Hospital, Philadelphia; the Department of Biostatistics and Epidemiology and Center for Clinical Epidemiology and Biostatistics (Dr Kanetsky), Department of Orthopedic Surgery (Dr Lackman), and Department of Medicine (Dr Martin), University of Pennsylvania, Philadelphia; and Department of Orthopedic Surgery, University of California, Davis (Dr Donthineni-Rao). The authors have no relevant financial interest in the products or companies described in this article. Corresponding author: Helen M. Haupt, MD, Pennsylvania Hospital, 8 Spruce St, Philadelphia, PA 191 (e-mail: haupt@pahosp.com). Reprints not available from the authors. Other important molecules in osteoclastogenesis are the microphthalmia transcription factor (MITF), a basic helixloop-helix zipper factor, and the Ets family factor PU.1, expressed by hematopoietic cells, which act synergistically to activate the target gene TRAP. 9 11 The MITF is expressed in osteoclast progenitor cells, 2 and its expression during melanocyte differentiation is modified by c-kit signaling. 12 Thus, c-kit activation of MITF may play a role in promotion of TRAP-positive mononuclear cell and multinucleated osteoclast formation. It has been shown that, although osteoclast precursors originate from mononuclear cells in both c-kit dependent and c-kit independent modes, 8,1 nonetheless, 9% of the precursors required c- Kit signaling for differentiation. 8 Although giant cell tumors of bone (GCTBs) and giant cell tumors of tendon sheath (GCTTSs) are recognized as separate entities, both contain cells that appear to be multinucleated osteoclast-like cells and mononuclear cells. It has been postulated that giant cell tumors (GCTs) consist of basic cell types, namely, osteoclast-like GCs, mononuclear cells, and a group of as yet uncharacterized tumorspecific cells. 14 We studied immunohistochemical expression of CD11 (c-kit), MITF, TRAP, and HAM-56 in GCTBs and GCTTSs to further define the sequence of differentiation in these lesions. MATERIALS AND METHODS Case Selection and Review of Pathologic Findings All cases were selected from the archives of the Department of Pathology, Pennsylvania Hospital, and cases that were initially diagnosed as GCTBs and GCTTSs from 1999 to 2 were re- 6 Arch Pathol Lab Med Vol 129, March 25 Giant Cell Tumors Ramos et al

Table 1. Reaction Conditions for Antibodies* Antibody Clone Staining Antigen Retrieval Dilution Manufacturer Macrophage HAM-56 Cytoplasmic Citrate ph 6. 1:5 Dako Corporation, Carpinteria, Calif CD11 (c-kit) Polyclonal Membranous NT 1:2 Dako Corporation MITF 4CA5 Nuclear EDTA ph.6 1:2 Novocastra Laboratories Ltd, Newcastle upon Tyne, United Kingdom TRAP 26E5 Cytoplasmic EDTA ph.6 1:1 Novocastra * NT indicates no treatment; MITF, microphthalmia transcription factor; TRAP, tartrate-resistant acid phosphatase. Figure 1. Giant cell tumor of bone (GCTB). A, Multinucleated giant cells (GCs) and mononuclear cells (hematoxylin-eosin, original magnification ). B, Antigen expression by mononuclear cells (immunoperoxidase, HAM-56, original magnification ). C, Cytoplasmic and nuclear staining of GCs and mononuclear cells (immunoperoxidase, microphthalmia transcription factor, original magnification ). D, Cytoplasmic staining of GCs and mononuclear cells (immunoperoxidase, tartrate-resistant acid phosphatase, original magnification ). viewed in accordance with the institutional review board approved protocol 816. The selected cases included 5 GCTBs, 15 GCTTSs (14 localized, 1 diffuse), and 5 foreign-body GC reactions. Hematoxylin-eosin stained slides were reviewed for histologic evidence of fibrosis, osteoid formation, and vascular invasion. Giemsa stain was performed on all cases to identify and control for mast cells, which express c-kit. 15 If mast cells were present, these foci were not evaluated when determining c-kit immunoreactivity. Immunohistochemical Staining Formalin-fixed, paraffin-embedded tissue sections were cut at M, heated at 58 C, deparaffinized in xylene, and hydrated in a graded series of alcohols. Antigen retrieval was performed by microwaving in citrate (ph 6. buffer) or directly on the Ventana Discovery Immunostainer (Table 1). Immunohistochemical analysis was performed on nondecalcified blocks using an indirect avidin-biotin complex system and diaminobenzidine as the chromogen. The slides were counterstained with hematoxylin and rehydrated using xylene. In the case of c-kit, appropriate positive controls, including gastrointestinal stromal tumors, were used. Furthermore, to determine a working concentration, a dilution gradient (1:25, 1:5, 1:1, and 1:2) was performed using the CD11 (c-kit) antibody. Appropriate staining of control cases was achieved at 1:2; this concentration was used in the study. Statistical Analysis Immunohistochemical staining was evaluated independently by 2 pathologists. Staining was evaluated in GCs and stromal cells for both intensity of staining and percentage of cells stained. Intensity of expression was graded as strong ( ), moderate (2 ), weak (1 ), or absent (), and percentage of expression was categorized as less than 1%, 1% to %, more than % to 6%, more than 6% to 9%, and more than 9%. For each antibody, the percentage of staining and staining intensity categories among tumor groups were compared nonparametrically using the Kruskal-Wallis test. For antibodies that demonstrated a significant difference in rank values, the per- Arch Pathol Lab Med Vol 129, March 25 Giant Cell Tumors Ramos et al 61

Figure 2. Localized giant cell tumor of tendon sheath (GCTTS). A, Multinucleated giant cells (GCs) and mononuclear cells (hematoxylin-eosin, original magnification ). B, Mononuclear cells of macrophage lineage demonstrate cytoplasmic staining; multinucleated GCs are negative (immunoperoxidase, HAM-56, original magnification ). C, Both GCs and some mononuclear cells show cytoplasmic staining (immunoperoxidase, tartrate-resistant acid phosphatase, original magnification ). Figure. Osteoclastic differentiation sequence of giant cell tumor. CFU-M indicates macrophage colony-forming units; c-kit, CD11; MITF, microphthalmia transcription factor; TRAP, tartrate-resistant acid phosphatase; RANK-L, nuclear factor B ligand; and IL-1, interleukin 1. centage of staining and staining intensity categories were dichotomized according to the median value (more than median value vs median value or less) in the overall sample. Correlations between antibody percentage of staining and staining intensity categories were determined using Spearman correlation coefficients. RESULTS Patient Population The 5 GCTBs were from 29 patients, with 6 being recurrences. Of the 5 GCTBs studied, 1 (plus the 6 recurrences) were from female patients, who ranged in age from 15 to 4 years (average age, 24.5 years), and 16 were from male patients, who ranged in age from 15 to 4 years (average age, 2.2 years). The 15 GCTTSs were from 14 patients, with 1 recurrence of a localized GCTTS, and included 1 female patients, who ranged in age from 14 to 6 years (average age, 48.2 years), and 4 male patients, who ranged in age from 19 to 58 years (average age, 6.5 years). Of the recurrent GCTBs, 5 were in female patients (P.). The recurrent GCTTS was in a female patient. No tumors metastasized. The sites of GCTBs were as follows: tibia, 11; femur, 9; humerus, ; hand, 2; wrist, 1; sacrum, 1; fibula, 1; and pelvis, 1. The sites of localized GCTTSs were as follows: hand, 5; knee, 4; ankle, ; and foot, 1; the diffuse GCTTSs were from the hip. Pathologic Findings The GCTBs demonstrated variable populations of mononuclear cells and multinucleated osteoclast-like GCs. Significant fibrosis ( 4% of tumor) was present in 5 cases (14%); osteoid with benign-appearing osteoblastic rimming was present in 8 cases (22%). Vascular invasion was identified in 1 case. Foci of mast cells were identified by Giemsa stain in 15 cases (4%). The GCTTSs were characterized by a mixed population of mononuclear cells and multinucleated GCs, with varying amounts of xanthomatous macrophages, fibrosis, and hemosiderin deposition. Mast cells were identified in 8 cases (5%). All foreignbody GC controls contained varying numbers of GCs, histiocytes, and mast cells. Immunohistochemical Analysis In both GCTBs and GCTTSs, HAM-56 staining was present in mononuclear cells but not in GCs (Figures 1 and 2); positive staining for HAM-56 was present in the multinucleated cells of all foreign-body GC controls. Generally, MITF was expressed in both mononuclear cells and GCs of GCTBs and GCTTSs, as well as in multinucleated foreign-body GCs. Staining was nuclear (Fig- 62 Arch Pathol Lab Med Vol 129, March 25 Giant Cell Tumors Ramos et al

Table 2. Percentage of Antibody Staining in Giant Cell Tumors of Bone (GCTBs) Compared With Giant Cell Tumors of Tendon Sheath (GCTTSs)* Antibody Staining, % MITF GCs MITF MCs c-kit GCs c-kit MCs TRAP GCs TRAP MCs HAM-56 GCs HAM-56 MCs GCTBs, % (n 5) 6 6 1 5 29 62 9 1 1 1 91 9 1 2 54 11 9 GCTTSs, % (n 15) P Value 9.2 2 1 54 1.1 1.99 1.99 1.99 8 1.24 1.99 2.1 * Immunohistochemical staining was evaluated independently by 2 pathologists. Staining was evaluated in giant cells and stromal cells for percentage of cells stained. Percentage of expression was categorized as less than 1%, 1% to %, more than % to 6%, more than 6% to 9%, and more than 9%. MITF indicates microphthalmia transcription factor; GCs, giant cells; MCs, mononuclear cells; and TRAP, tartrateresistant acid phosphatase. Calculated with the Kruskal-Wallace test. ures 1 and 2). The MITF strongly stained the nuclei of all xanthoma cells in GCTTSs. The MITF was more often expressed in mononuclear cells of GCTTSs than GCTBs (P.1, Table 2). The GCTBs showed more intense MITF mononuclear cell staining than the GCTTSs (P.1, Table ). Membranous staining with c-kit was identified as an antibody concentration of 1:25 and 1:5 in stromal cells but not GCs in some GCTBs and GCTTSs. However, staining was lost at dilutions of 1:1 and 1:2. The recommended dilution is 1:2, which appropriately stained the gastrointestinal stromal tumor controls. Results at 1:2 were used in the statistical analysis. TRAP was generally expressed in GCs of both GCTBs and GCTTSs and GC reaction controls. TRAP was variably expressed in mononuclear cells of both tumor types; staining was cytoplasmic (Figures 1 and 2). The GCTBs showed more TRAP GC staining (P.4, Table ) than GCTTSs. HAM-56 staining in mononuclear cells correlated with MITF expression by mononuclear cells (r 2.6, P.1). COMMENT Both GCTBs and GCTTSs express antigens of the monocyte-macrophage (HAM-56) and osteoclastic GC (MITF and TRAP) lineage, as well as c-kit in patterns of expression, which help to elucidate the sequential differentiation of these tumors. The HAM-56 staining in mononuclear cells in GCTBs and GCTTSs supports a monocyte-macrophage origin, since HAM-56 is known to stain macrophages, 16 although not exclusively. We observed that HAM-56 failed to stain the GCs of any tumor but did stain the multinuclear GC controls, suggesting a difference in antigen expression by GC tumors. Thus, as mononuclear cells differentiate into multinucleated GCs, HAM-56 staining is lost. In our study, c-kit was not expressed in mononuclear Arch Pathol Lab Med Vol 129, March 25 Giant Cell Tumors Ramos et al 6

Table. Antibody MITF GCs MITF MCs c-kit GCs c-kit MCs TRAP GCs TRAP MCs HAM-56 GCs HAM-56 MCs Antibody Staining Intensity in Giant Cell Tumors of Bone (GCTBs) Compared With Giant Cell Tumors of Tendon Sheath (GCTTSs)* Staining Intensity No Staining Cells Stained in GCTBs, % (n 5) 1 2 5 2 2 54 1 1 24 6 6 1 1 Cells Stained in GCTTSs, % (n 15) P Value 4 5.5 8 6 1 1.1.99.99 2 2 5.4 2 1 6.98 1.99 1.99 * Staining was recorded if it was seen in more than 1% of cells. Immunohistochemical staining was evaluated independently by 2 pathologists. Staining was evaluated in giant cells and stromal cells for intensity of staining. Intensity of expression was graded as strong ( ), moderate (2 ), or weak (1 ) in at least 1% of cells, or absent (). MITF indicates microphthalmia transcription factor; GCs, giant cells; MCs, mononuclear cells; and TRAP, tartrate-resistant acid phosphatase. Calculated with the Kruskal-Wallace test. cells or GCs of either GCTBs or GCTTSs at an appropriate dilution of 1:2. c-kit is present in pluripotent hematopoietic stem cells 1 ; however, studies have shown that its expression is low. 18 A previous study by Gattei et al, 19 using indirect immunofluorescence, found detectable c-kit expression by bone marrow derived human preosteoclastic mononuclear cells and by multinucleated osteoclasts of bone and of GCTBs. In addition, an in vitro study by Yamazaki et al 8 demonstrated that osteoclast differentiation from precursor cells can occur in both c-kit dependent and independent modes and that most cells required c- Kit. A further study by Haydon et al 14 investigated c-kit expression by immunohistochemical analysis in 1 paraffin-embedded archival GCTB samples. They found that all 1 tumors stained strongly with c-kit. Sequence analysis of 2 of the 1 samples for activating mutations failed to reveal alterations in the c-kit gene. Cell lines were generated from 4 of these tumors using fresh tissue. When cultured cells were exposed to ST151, growth was inhibited at 1 M and cell death occurred at 2 M. Fibroblast controls demonstrated no growth inhibition or cell death. These investigations suggest that c-kit is present during osteoclast differentiation. c-kit expression is not equivalent to functional activation. 2 Furthermore, we and others have shown that the ability to stain for c-kit may be concentration dependent. 15,2 In a study of desmoid tumors, c-kit positivity was found to be present at low dilutions and lost with higher dilutions. 15,2 In this study, a dilution gradient for c-kit of 1:25, 1:5, 1:1, and 1:2 was performed, and results were reported with the 1:2 dilution, which is the manufacturer s suggested dilution for c-kit (Dako Corporation, Carpinteria, Calif). Although some cells stained at dilutions of 1:25 and 1:5, nonetheless, staining for c- Kit was lost at the 1:1 and 1:2 dilutions. However, based on the studies of Gattei et al 19 and Yamazaki et al, 8 the question of c-kit relevance in osteoclastogenesis still remains. Additional assays, such as c-kit gene expression analysis, immunoblotting, and a more extensive mutation screening, would be useful in future studies to investigate the presence of a possible c-kit alteration in GCTs. The MITF is expressed in melanocytes, 21 and others have shown that activation of MITF is c-kit dependent. 12,22 Osteoclast promotion from mononuclear cells is also at least partially c-kit dependent. 8 The MITF is known to act with PU.1 to activate TRAP during osteoclastogenesis. 9,1 Seethala et al 2 studied MITF expression in 2 GC lesions. They found that MITF was widely expressed in GCs and adjacent mononuclear cells in all GC lesions but was not expressed in mononuclear cells in lesions that did not con- 64 Arch Pathol Lab Med Vol 129, March 25 Giant Cell Tumors Ramos et al

tain multinucleated GCs. They concluded that MITF might play a role in the differentiation of mononuclear cells to form GCs in GCTs. In our study, MITF was generally expressed in both mononuclear cells and GCs of GCTBs and GCTTSs, as well as in multinucleated foreign-body GCs. TRAP expression has been shown in vitro to be acquired during osteoclastogenesis. 4 In our study, TRAP was expressed by both GCs and mononuclear cells, suggesting that the antigen is acquired in vivo during the monocytic osteoclast precursor phase of differentiation of GCT. TRAP expression in the GCTs of our study persisted in the fully differentiated multinuclear GCs. Suda et al, 4 who reviewed previous studies of osteoclast differentiation and TRAP expression, found similar conclusions. Similarly, MITF, which is important for osteoclastic differentiation and activation of TRAP, 9,1 was expressed in both mononuclear cells and GCs of both GCTBs and GCTTSs. In summary, the results of this study suggest that MITF and TRAP are expressed during osteoclast differentiation and that a proportion of mononuclear cells in GCTs expresses the macrophage marker HAM-56. Based on the observations in our study and the model previously published by Suda et al, 4 we propose a simplified schematic model of osteoclastic differentiation in GCTTSs (Figure ). We propose that these lesions differentiate from macrophage-monocyte precursors and that MITF activation drives the genetic expression of TRAP in osteoclastic mononuclear precursors and multinucleated GCs. Any role of c-kit remains unclear. Similar patterns of immunohistochemical expression in GCTBs and GCTTSs suggest a similar differentiation sequence. We thank Anna Marie McClain for her meticulous assistance with the manuscript. References 1. Aubin JE, Bonneleye E. Osteoprotegerin and its ligand: a new paradigm for regulation of osteoclastogenesis and bone resorption. Osteoporos Int. 2;11: 95 91. 2. Kawaguchi N, Noda M. MITF is expressed in osteoclast progenitors in vitro. Exp Cell Res. 2;26:284 291.. Kong YY, Penninger JM. Molecular control of bone remodeling and osteoporosis. Exp Gerontol. 2;5:94 956. 4. Suda T, Takahashi N, Udagawa N, et al. Modulation of osteoclast differentiation and function by the new members of the tumor necrosis factor receptor and ligand families. Endocr Rev. 1999;2:45 5. 5. Takahashi N, Udagawa N, Suda T. A new member of tumor necrosis factor ligand family, ODF/OPGL/TRANCE/RANKL, regulates osteoclast differentiation and function. Biochem Biophys Res Commun. 1999;256:449 455. 6. Teitelbaum SL. Bone resorption by osteoclasts. Science. 2;289:154 158.. Weilbacher KN, Motyckova G, Huber WE, et al. Linkage of M-CSF signaling to MITF, TFE-, and the osteoclast defect in MITF (mi/mi) mice. Mol Cell. 21; 8:49 58. 8. Yamazaki H, Kunisada T, Yamane T, et al. Presence of osteoclast precursors in colonies cloned in the presence of hematopoietic colony-stimulating factors. Exp Hematol. 21;29:68 6. 9. Luchin A, Suchting S, Merson T, et al. Genetic and physical interactions between microphthalmia transcription factor and PU.1 are necessary for osteoclast gene expression and differentiation. J Biol Chem. 21;26:6 61. 1. Luchin A, Purdom G, Murphy K, et al. The microphthalmia transcription factor regulates expression of the tartrate-resistant acid phosphatase gene during terminal differentiation of osteoclasts. J Bone Min Res. 2;15:451 46. 11. Mansky KC, Sulzbacher S, Purdom G, et al. The microphthalmia transcription factor and the related helix-loop-helix zipper factors TFE- and TFE-C collaborate to activate the tartrate-resistant acid phosphatase promoter. J Leukoc Biol. 22;1:4 1. 12. Hemesath TJ, Price ER, Takemoto C, et al. MAP kinase links the transcription factor microphthalmia to c-kit signaling in melanocytes. Nature. 1998;91: 298 1. 1. Muguruma Y, Lee MY. Isolation and characterization of murine clonogenic osteoclast progenitors by cell surface phenotype analysis. Blood. 1998;91:122 129. 14. Haydon RC, Deyrup A, Ishikawa AR, et al. Upregulation of c-kit in giant cell tumor of the bone: possible therapeutic target using STI-51. Paper presented at Orthopedic Research Society and Musculoskeletal Tumor Society Meeting; 2. 15. Lucas DR, Al-Abbadi M, Tabaczka P, et al. c-kit expression in desmoid fibromatosis: comparative immunohistochemical evaluation of two commercial antibodies. Am J Clin Pathol. 2;119:9 45. 16. Helm KF. Immunohistochemistry of skin tumors. In: Dabbs DJ. Diagnostic Immunohistochemistry. Philadelphia, Pa: Churchill Livingstone; 22:2. 1. Orkin SH. Diversification of haematopoietic stem cells to specific lineages. Nat Rev Genet. 2;1:5 64. 18. Doi H, Inaba M, Yamamoto Y, et al. Pluripotent hematopoietic stem cells are c-kit low. Proc Natl Acad Sci U S A. 199;94:251 251. 19. Gattei V, Aldinucci D, Quinn JM. et al. Human osteoclasts and preosteoclast cells (FLG 29.1) express functional c-kit receptors and interact with osteoblast and stromal cells via membrane-bound stem cell factor. Cell Growth Differ. 1996;:5 6. 2. Hornick JL, Fletcher CDM. Validating immunohistochemical staining for KIT (CD 11). Am J Clin Pathol. 2;119:25 2. 21. King R, Googe PB, Weilbaecher KN, et al. Microphthalmia transcription factor expression in cutaneous benign, malignant melanocytic and non-melanocytic tumors. Am J Surg Pathol. 21;25:51 5. 22. Price ER, Ding HF, Badalian T, et al. Lineage-specific signaling in melanocytes: c-kit stimulation recruits p/cbp to microphthalmia. J Biol Chem. 1998;2:198 1986. 2. Seethala RR, Goldblum JR, Lehman M, et al. Immunohistochemical evaluation of MITF expression in giant cell lesions. Mod Pathol. 2;16:2A. Arch Pathol Lab Med Vol 129, March 25 Giant Cell Tumors Ramos et al 65