GeneticAnalysisofGiantCellLesionsoftheMaxillofacialandAxial/AppendicularSkeletons PATHOLOGY

(1)

Genetic Analysis of Giant Cell Lesions of the Maxillofacial and Axial/

Appendicular Skeletons

Zachary S. Peacock, DMD, MD,*Joseph H. Schwab, MD,yWilliam C. Faquin, MD, PhD,z Francis J. Hornicek, MD, PhD,xYair Benita, PhD,kDavid H. Ebb, MD,{

and Leonard B. Kaban, DMD, MD#

Purpose: To compare the genetic and protein expression of giant cell lesions (GCLs) of the maxillofacial (MF) and axial/appendicular (AA) skeletons. We hypothesized that when grouped according to biologic behavior and not simply by location, MF and AA GCLs would exhibit common genetic characteristics.

Materials and Methods: This was a prospective and retrospective study of patients with GCLs treated at Massachusetts General Hospital from 1993 to 2008. In a preliminary prospective study, fresh tissue from 6 aggressive tumors each from the MF and AA skeletons (n = 12 tumors) was obtained. RNA was extracted and amplified from giant cells (GCs) and stromal cells first separated by laser capture microdissection.

Genes highly expressed by GCs and stroma at both locations were determined using an Affymetrix Gen- eChip analysis. As confirmation, a tissue microarray (TMA) was created retrospectively from representative tissue of preserved pathologic specimens to assess the protein expression of the commonly expressed genes found in the prospective study. Quantification of immunohistochemical staining of MF and AA lesions was performed using Aperio image analysis to determine whether immunoreactivity was predictive of aggressive or nonaggressive behavior.

Results: Five highly ranked genes were found commonly in GCs and stroma at each location: matrix metalloproteinase-9 (MMP-9), cathepsin K (CTSK), T-cell immune regulator-1 (TCIRG1), C-type lectin domain family-11, and zinc finger protein-836. MF (n = 40; 32 aggressive) and AA (n = 48; 28 aggressive) paraffin-embedded tumors were included in the TMA. The proteins CTSK, MMP-9, and TCIRG1 were confirmed to have abundant expression within both MF and AA lesions. Only the staining levels for TCIRG1 within the GCs predicted the clinical behavior of the MF lesions.

*Assistant Professor, Department of Oral and Maxillofacial Surgery, Massachusetts General Hospital and Harvard School of Dental Medicine, Boston, MA.

yAssistant Professor, Department of Orthopaedics, Massachusetts General Hospital and Harvard Medical School, Boston, MA.

zAssociate Professor, Department of Pathology, Massachusetts General Hospital and Harvard Medical School, Boston, MA.

xAssociate Professor, Department of Orthopaedics, Massachusetts General Hospital and Harvard Medical School, Boston, MA.

kFormer Fellow, Center for Computational and Integrative Biology, Massachusetts General Hospital and Harvard Medical School, Boston, MA.

{Assistant Professor, Department of Pediatrics, Massachusetts General Hospital and Harvard Medical School, Boston, MA.

#Walter Guralnik Distinguished Professor, Department of Oral and Maxillofacial Surgery, Massachusetts General Hospital and Harvard School of Dental Medicine, Boston, MA.

The present study was supported by the Department of Oral and Maxillofacial Surgery Education and Research Fund and the Department of Orthopaedic Surgery, Massachusetts General Hospi- tal and Harvard School of Dental Medicine, Boston, MA.

Z.S.P. and J.H.S. are co-first authors.

Conflict of Interest Disclosures: None of the authors have any relevant financial relationship(s) with a commercial interest.

This study was presented at the Surgical Forum at the Annual Meeting of the American College of Surgeons, October 2014, San Francisco, CA, and published as an abstract in a supplement of the Journal of American College of Surgeons (Peacock ZS, Schwab JH, Faquin WC, et al: Genetic analysis of giant cell tumors of the maxillofacial and axial/appendicular skeleton. J Am Coll Surg 219:S70, 2014).

Address correspondence and reprint requests to Dr Peacock:

Department of Oral and Maxillofacial Surgery, Massachusetts Gen- eral Hospital, Warren 1201, 55 Fruit Street, Boston, MA 02114;

e-mail:zpeacock@partners.org Received May 27 2016 Accepted July 19 2016

Ó 2016 American Association of Oral and Maxillofacial Surgeons 0278-2391/16/30618-8

http://dx.doi.org/10.1016/j.joms.2016.07.014

298

(2)

Conclusions: MMP-9, CTSK, and TCIRG1 are commonly expressed by GCLs of the MF and AA skeletons.

This supports the hypothesis that these lesions are similar but at different locations. TCIRG1 has not been previously associated with GCLs and could be a potential target for molecular diagnosis and/or therapy.

Ó 2016 American Association of Oral and Maxillofacial Surgeons J Oral Maxillofac Surg 75:298-308, 2017

Giant cell lesions (GCLs) are intraosseous benign tumors containing multinucleated giant cells (GCs) within a mononuclear stroma. They can affect both the maxillofacial (MF) and the axial/appendicular (AA) skeletons. Lesions at both sites can vary from small, slow-growing tumors recognized as incidental radiographic findings to large destructive lesions, lead- ing to displacement or resorption of adjacent struc- tures or pathologic fracture.^1,2

GCLs of the MF skeleton are more common in females and younger patients (2:1 females to males during the first and second decade) compared with those of the AA skeleton (1.3 to 1.5 females to males during the third to fifth decade).^3-9 MF GCLs are more likely to be asymptomatic, and they are often discovered on routine dental radiographic examinations.^4,10

Controversy remains regarding the biologic relationship between GCLs of the jaws and giant cell tumors (GCTs) in the AA skeleton.^3-5,10-14 This is because subgroups of these tumors (ie, aggressive and nonaggressive) according to biologic behavior have not been differentiated in reported comparisons.

Some investigators have postulated that MF and AA GCLs are distinctly separate lesions because of differences in clinical behavior and histopathologic features.^3,11,15Others have supported the hypothesis that these are similar lesions in different locations representing a continuum of the same disease process.^3-5,16,17 In each location, the pathogenesis has been hypothesized to involve stromal fibroblasts that recruit monocytes, which then transform into multinucleated GCs.^18,19 These GCs have been shown to be phenotypically osteoclasts through immunohistochemistry.²⁰

Clinical and radiographic features can be used to classify GCLs in both locations as aggressive or nonaggressive.^5,10,21Lesions of the MF region are considered aggressive if they are larger than 5 cm, recurrent, or meet 3 of the following 5 criteria: rapid growth, root resorption, tooth displacement, cortical bone thinning, and/or perforation. Nonaggressive lesions grow slowly and are asymptomatic, with a low rate of recurrence after enucleation or curettage. AA lesions are classified according to the classification system of Enneking, also by clinical and radiographic behavior.^1,22Enneking stage 1 (latent) refers to static lesions or those that heal spontaneously. Stage 2

(active) tumors exhibit progressive growth but are limited by natural barriers (ie, cortices). Stage 3 lesions are locally aggressive, with destruction of natural barriers.²² As previously outlined, both the Enneking and the Chuong and Kaban²¹MF classification systems, using clinical and radiographic criteria, can be modified to produce a single, biologically consistent binary classification for GCTs in both locations: aggressive and nonaggressive.^1,12,13

Studies comparing GCLs of the MF and AA skeletons have been limited by inconsistent terminology, and investigators have typically grouped all lesions together, without considering the clinical or biologic behavior. MF GCLs were first referred to as ‘‘giant cell reparative granuloma’’ by Jaffe¹¹in 1953. Reports of spontaneous resolution have been published.^23,24 However, other MF GCLs are destructive and grow rapidly.¹⁰ Additionally, the reluctance to label MF GCLs as ‘‘giant cell tumors’’ can result from the reports of metastases and malignant transformation from AA GCLs.^25-27 However, retrospective analyses of cases of lung metastases from AA GCTs have indicated that these might actually be malignant tumors that happen to contain GCs.²⁸In a recent study, investigators found consistent mutations in histone 3.3 driver variants (H3F3A gene) in the stromal cells of AA GCLs.²⁹ They reported H3.3, H3F3A, and H3F3B in a variety of bone and cartilage tumors.^29,30 A subsequent study by a different group of both aggressive and nonaggressive GCLs of the MF skeleton did not find these specific mutations.³¹ Although this supports that these are different lesions, mutations in histone, the protein that packages DNA into nucleosomes, have been found in a variety of bone and cartilage tumors and in pediatric brain tumors and might not be involved in pathogenesis.^32-34

Our group has previously compared the lesions in each location by biologic behavior using the Chuong and Kaban classification²¹ and a modified version of the Enneking classification¹ (ie, aggressive or nonaggressive) and found the lesions to be similar with regard to the phenotypic, clinical, and radiographic appearance.¹³ Subsequently, it was shown that these lesions are histologically similar and that they could not be differentiated consistently by blinded pathologists.¹⁴These findings sup- port the conclusion that they are similar tumors in different locations.

(3)

The purpose of the present study was to compare the gene and protein expression of GCLs of the MF and AA skeletons. We hypothesized that the gene expression between GCLs at both locations would be consistent and would be confirmed by protein analysis. Second, we hypothesized that the gene expression would have a relationship to the biologic behavior (ie, aggressive vs nonaggressive). Further characterization of the relationship between MF and AA GCLs could provide insight into the etiopatho- genesis of these tumors and help direct specific treatment.

Materials and Methods

STUDY DESIGN AND POPULATION

This was a prospective and retrospective cohort study of patients with GCLs treated at Massachusetts General Hospital (MGH) from January 1993 to June 2008. The gene expression analysis was prospective and consisted of data from consecutive patients with AA and MF GCLs treated by us. Patients were included if the lesion was identified as aggressive according to the preoperative clinical and radiographic criteria. For the retrospective protein expression analysis, potential subjects were identified from the MGH Giant Cell Patient Registry. The inclusion criteria were histologically confirmed MF or AA GCLs treated by the Department of Oral and Maxillo- facial Surgery or Department of Orthopaedic Surgery and a minimum follow-up period of 6 months.

Patients with hyperparathyroidism, cherubism, Noonan syndrome, inadequate histologic samples, follow-up shorter than 6 months, or inadequate documentation were excluded. The MGH institu- tional review board approved the project (protocol no. 2008P000563).

GENE EXPRESSION ANALYSIS

Study Variables for Protein Expression Analysis The predictor variables were GCL location (AA vs MF) and tissue type (GCs and stromal cells). The primary outcome variable was the level of gene expression. Biologic behavior was classified preoperatively as aggressive or nonaggressive using standard published criteria,^1,12,21 with modification of the Enneking staging systems for AA GCLs into aggressive and nonaggressive lesions, as previously outlined.^13,14MF lesions were considered aggressive if they were larger than 5 cm in size, had recurred, or had at least 3 of the following characteristics: root resorption, tooth displacement, or cortical bone thinning or perforation. For AA lesions, all Enneking stage 3 and any stage 2 lesions with pathologic fracture or recurrence were considered aggressive.

Subjects with aggressive lesions provided informed consent for the use of fresh tissue.

Laser Capture Microdissection

Fresh specimens from 6 consecutive aggressive GCLs from each location (n = 12) were assessed. The specimens were snap frozen in liquid nitrogen imme- diately after curettage in the operating room. RNA was extracted using TRI Reagent (Sigma-Aldrich, St Louis, MO) and then subsequently analyzed for quality using the Agilent Bioanalyzer (Agilent Technologies Inc., Palo Alto, CA) before being subjected to laser capture microdissection (LCM).

Only specimens with acceptable 18S and 28S peaks were used. LCM was performed using RNAse free conditions. The specimens were cut into 8-mm sections and mounted onto uncoated glass slides and kept at 80F. In preparation for LCM, the specimens were thawed and then fixed using 70% ethanol for 45 seconds. The specimens were stained with methylene blue to aid in the identification of cells using 0.1%

methylene blue for 20 seconds. The sections were then gradually dehydrated using ethanol (50, 70, 90, and 100 seconds each for each cycle) before being placed in xylene for 5 to 10 minutes. An Arcturus Pix- Cell II instrument (Applied Biosystems, Life Technolo- gies, Carlsbad, CA) was used for LCM to ensure only tumor cell RNA was used for analysis (Fig 1A,B). In addition, we extracted RNA from stromal cells and multinucleated GCs separately to determine the differences in gene expression.

Extracted RNA was then amplified using Arcturus Picopure RNA isolation kits and amplified using Arcturus RiboAmp RNA amplification kits (Applied Biosystems, Life Technologies). The amplification product was sent to the Harvard Partners Center for Genetics and Genomics facility (Cambridge, MA) to generate biotinylated RNA and subsequent hybridization with GeneChip Human X3P expression arrays (Affymetrix, Santa Clara, CA).

The GeneChip data were compared with 126 primary cell types using the Gene Enrichment Profiler tool, developed by Yair Benita (Xavier Laboratory, Mas- sachusetts General Hospital, Harvard Medical School, Boston, MA). The tool allows the gene expression found in each cell type to be compared with the other cell or tissue types to ‘‘rank’’ the gene expression according to the score found during analysis. The higher the score attributed to a gene within a tissue or cell type, the more specific that particular gene is to that particular tissue.

Primary normal tissue and cells were obtained from the National Center for Biotechnology Information Gene Expression Omnibus³⁵and ArrayExpress.³⁶ All samples obtained were profiled using the Affymetrix U133 Plus 2.0 Array platform (Affymetrix). We used

(4)

version 14 of the custom transcript definition files provided by Brainarray (Microarray Lab, Molecular and Behavioral Neuroscience Institute, University of Mich- igan, Ann Arbor, MI).³⁷These files redefine Affymetrix probes by remapping individual probes to the human genome and adjusting to the most up-to-date annota- tion. The data files were then normalized using the GCRMA module of Bioconductor³⁸ (available at:

https://www.bioconductor.org/), and present and absent cells were calculated for each probe using the

MAS5 module. All probes with no cells present were removed. At least 1 sample was required to have an expression value larger than log₂(100).

The gene expression enrichment analysis was performed as previously described.³⁹The GCL microar- rays were added to the data set of 126 normal tissues and cell types and normalized as a single experiment.

Genes scoring greater than 1,000 on the enrichment scale in any of the GCL samples was considered highly enriched in these tissues (Fig 2).

FIGURE 1. Photomicrograph of a representative aggressive maxillofacial giant cell lesion A, before and B, after laser capture microdissection to extract giant cells (arrows) (hematoxylin and eosin stain, original magnification 200).

Peacock et al. Genetic Analysis of GCLs of MF and AA Skeletons. J Oral Maxillofac Surg 2017.

(5)

PROTEIN EXPRESSION ANALYSIS

Study Variables for Protein Expression Analysis The predictor variables were GCL location (AA vs MF), behavior (aggressive vs nonaggressive), and tissue type (GCs and stromal cells). The outcome variable was the average optical density of staining found on Aperio analysis. Comparisons in optical density were made between locations and between behavior type.

Construction of Tissue Microarray

Paraffin blocks containing GCLs from subjects who had met the inclusion criteria were sectioned into slides (5mm) and used to mark 3 separate areas containing representative tumor on the paraffin block.

The 3 areas of tumor marked were then cored and used to create a tissue microarray (TMA) containing all GCTs from both locations on 1 paraffin block.

Commercially available antibodies to the proteins corresponding to the commonly over-expressed genes were then optimized using positive and negative controls under the direction of a board-certified pathologist. Antibodies to the following proteins were analyzed: matrix metalloproteinase-9 (MMP-9; mouse monoclonal; catalog no. MAB3309; clone 56-2A4; Milli- pore, Billerica, MA), cathepsin K (CTSK; mouse monoclonal; catalog no. ab66237; Abcam, Cambridge, MA), T-cell immune regulator-1 (TCIRG1; rabbit polyclonal;

catalog no. NBP1-88110; Novus Biologicals, Littleton,

CO), stem cell growth factor (SCGF; zinc-finger 836, mouse monoclonal; catalog no. ab90238; Abcam), and C-type lectin domain family 11 (CLEC11A; rabbit polyclonal; Abgent Inc, San Diego, CA).

Immunohistochemistry

Sections (5 mm) of the TMA were stained using the following immunohistochemical techniques. For MMP-9, immunohistochemistry was performed manu- ally using Decloaking Chambers (Biocare Medical, Concord, CA) for antigen retrieval using citrate (pH 6.0) solution. The primary antibody was incu- bated at room temperature for 1 hour at a 1:1,500 dilution. Mouse Envision-HRP reagent (Dako, Carpin- teria, CA) was used to detect antigen, followed by development in diaminobenzene for 5 minutes. For cathepsin-K (1:100) and TCIRG1 (1:400), CLEC11A (1:50), and SCGF antibodies (1:50), Leica’s BOND-III staining platform (Leica Biosystems Inc, Buffalo Grove, IL) was used with the Bond Polymer Refine Detection Kit (Leica Biosystems Inc). Antigen retrieval was performed online using Bond Epitope Retrieval Solution 1 (cathepsin-K and SCGF) or 2 (TCIRG1, CLEC11A;

Leica Biosystems Inc) for 30 minutes.

Quantification of Immunohistochemical Staining The level of staining was quantified in each lesion using Aperio (Leica Biosystems Inc) digital pathology scanning and analysis using a color deconvolution

FIGURE 2. Genomic heat map allowing visualization of microarray values for giant cells and stroma from maxillofacial and axial/appendicular skeletons and 126 normal cell types. Those with greater gene enrichment in that cell type are depicted in brighter shades of red.

(6)

protocol.^40,41The average optical density of the GCs and stroma were calculated using 3 different areas within each specimen. Positive controls for each antibody were included in the TMA. Omitting the primary antibody served as the negative control.

STATISTICAL ANALYSIS

Data were entered into a statistical database (SPSS, version 22.0; SPSS Inc, Chicago, IL) for analysis.

Descriptive statistics were calculated. For comparisons between optical density by location and behavior, an independent t test was used. The Bonferroni correction was used given the multiple comparisons (P#.00625 was considered significant).

Results

GENE EXPRESSION ANALYSIS

The 4 tissue types studied (GCs and stroma from each location) were clustered together on an unsuper- vised analysis of gene expression. The genes with the highest enrichment scores for each of the 4 tissue types are provided (Table 1). Five genes were shared by all 4 mentioned tissue types: MMP-9, CTSK, TCIRG1, CLEC11A, and ZNF836. Of these, 3 (MMP- 9, CTSK, and TCIRG1) were in the top 8 expressed genes of each tissue type (Table 1). Genes scoring above 1,000 for any of the GCT samples were selected, and a heat map was generated for the entire data set (Fig 3). Genes enriched in stroma and GCs were clearly distinct from those in other tissues.

ANALYSIS OF PROTEIN EXPRESSION Demographic Data

The study sample included 88 subjects (40 MF and 48 AA) who met the inclusion criteria. A modified classification system identified 32 and 28 aggressive lesions of the MF and AA skeletons, respectively; 8 MF and 20 AA were classified as nonaggressive.

Immunohistochemistry

Staining using the antibodies for MMP-9 and CTSK was positive in both GCs and stroma (Table 2, Figs 3, 4). TCIRG1 showed positive staining in GCs but less so in the stroma (Table 2,Fig 5). CLEC11A and ZNF836 did not show positive staining in either tissue type.

The difference in the optical density of staining according to behavior and location is summarized in Table 2. No statistically significant difference was found for MMP-9 or CTSK with regard to behavior.

TCIRG1 had greater staining intensity of the GCs of nonaggressive MF lesions compared with aggressive MF lesions (P < .001).

Discussion

In this set of experiments, we compared the genetic and protein expression of GCLs of the MF and AA skeletons. We hypothesized that a common genetic expression would be present between GCLs at both locations that could be confirmed by protein expression analysis. The intensity of antibody staining was hypothesized to be related to the biologic behavior.

GCLs of the AA and MF skeletons share common genetic expression of MMP-9, CTSK, and TC1RG1, supporting the hypothesis that they are likely similar lesions at different anatomic locations. The common genetic expression was confirmed by assessment of protein expression using a TMA to quantify immunoreactivity.

Both MMP-9 and CTSK have been shown to be expressed in GCLs in previous studies.^42-46 They are thought to be involved in collagen degradation and hence osteolysis.⁴⁷ Mutations in TCIRG1 have been associated with osteopetrosis, which is thought to

Table 1. TOP 10 MOST ENRICHED GENES FOR EACH OF THE 4 TISSUE TYPES STUDIED

Maxillofacial Axial/Appendicular Giant Cells Stroma Giant Cells Stroma

MMP-9* MMP-9* TCIRG1* MMP-9*

ACP5 OMD ZNF836 HOXC9

CTSK* FNDC1 CTSK* CTHRC1

TCIRG1* CTSK* MMP-9* CTSK*

HYAL1 MSX1 TNFRSF11A MXRA5

CCR1 CTHRC1 CLEC11A CLEC11A

TNFRSF11A TCIRG1* ZNF383 S100A2

SF3A2 COL24A1 CCR1 TCIRG1*

CLEC11A ITGA10 ARRB1 ASPN

SLCO4A1 SLC2A10 LOC653390 ZNF836 Abbreviations: ACP5, acid phosphatase 5; ARRB1, arrestin beta 1; ASPN, asporin; CCR1, C-C motif chemokine receptor 1; CLEC11A, C-type lectin domain family 11; COL24A1, collagen type XXIV alpha 1; CTHRC1, collagen triple helix repeat containing 1); CTSK, cathepsin K; FNDC1, fibro- nectin type III domain containing 1; HOXC9, homeobox C9; HYAL1, hyaluronoglucosaminidase 1; ITGA10, integrin subunit alpha 10; LOC653390, hypothetical protein LOC100292228 or RRN3 RNA polymerase I transcription factor homolog (S. cerevisiae) pseudogene; MMP-9, metalloproteinase-9; MSX1, Msh homeobox 1; MXRA5, matrix- remodeling associated 5; OMD, osteomodulin; S100A2, S100 calcium binding protein A2; SF3A2, splicing factor 3a subunit 2; SLC2A10, solute carrier family 2 member 10;

SLCO4A1, solute carrier organic anion transporter family member 4A1; TCIRG1, T-cell immune regulator-1;

TNFRSF11A, tumor necrosis factor receptor superfamily member 11a; ZNF, zinc finger protein.

* Found in all 4 tissue types.

(7)

FIGURE 3. Tissue microarray staining of matrix metalloproteinase-9 (200) for A, maxillofacial and B, axial/appendicular giant cell tumors.

Staining is intense in both the giant cells (arrows) and the cytoplasm of mononuclear stromal cells (arrowheads).

Table 2. COMPARISON OF ANTIBODY STAINING BY LOCATION AND BEHAVIOR

Protein

Maxillofacial Axial/Appendicular

Aggressive Nonaggressive P Value Aggressive Nonaggressive P Value GCs Stroma GCs Stroma GCs* Stroma* GCs Stroma GCs Stroma GCs* Stroma*

CTSK 0.1476 0.1261 0.1584 0.1541 .35 .20 0.1394 0.1359 0.1436 0.1262 .19 .33 MMP-9 0.1058 0.1042 0.1063 0.1004 .87 .49 0.1026 0.1036 0.1047 0.1057 .35 .59 TCIRG1 0.1361 0.1075 0.2037 0.1569 < .0001^y .011 0.1305 0.1093 0.1308 0.1098 .96 .94 Data presented as optical density.

Abbreviations: CTSK, cathepsin K; GC, giant cell; MMP-9, matrix metalloproteinase-9; TCIRG1, T-cell immune regulator-1.

* Comparing aggressive and nonaggressive lesions.

y Statistically significant.

(8)

result from defective osteoclast function.⁴⁸ This is relevant to GCLs because the multinucleated GC have been shown to be phenotypically osteoclasts.²⁰ Further study of TCIRG1 in GCLs might shed light on the still poorly understood pathogenesis of GCTs.

Both CLEC11A, a protein associated with monocytes, and ZNF836, a transcription factor not studied in any tissue type, had common genetic expression between GCLs at each location.⁴⁹ The commercial antibodies to these proteins were difficult to optimize as a first step and resulted in negative staining of the TMA and its positive controls. The relative obscurity of these proteins limits commercial availability, and confirmation studies do not exist.

Only TCIRG1 had a statistically significant association with biologic behavior. TCIRG1 staining intensity was greater in nonaggressive MF lesions. Not enough

is known about the role of TCIRG1 in these lesions to make conclusions. One would expect MMP-9 and CTSK to have greater expression in aggressive lesions;

however, this was not seen in the present study.

A recent study that performed whole genome sequencing of GCLs of the AA skeleton found that 92% (49 of 53) of GCTs of bone harbored histone 3.3 variants exclusively in H3F3A (p.Gly34 Trp or p.Gly34 Leu mutations).²⁹They also found consistent mutations in this histone gene at a different location for 95% (73 of 77) of chondroblastomas and to a much lesser extent in osteosarcoma and 2 variants of chondrosarcoma. The mutations were restricted to the stromal cell population and not detected in osteoclasts or their precursors. Gomes et al³¹ in a subsequent study did not find identical mutations in the H3F3A gene of 9 GCLs (6 aggressive) of the MF

FIGURE 4. Tissue microarray staining of cathepsin K (200) for A, maxillofacial and B, axial/appendicular lesions. Staining is intense in both the giant cells (arrows) and mononuclear stromal cells (arrowheads).

(9)

skeleton using direct sequencing of an amplified polymerase chain reaction product.³¹ This discrepancy supports the theory that GCLs of the AA skeleton are different from MF lesions, but it is possible that other specific histone mutations would be uncovered with whole genome sequencing or perhaps by separating stromal cells from GCs as the first group did. The relationship of mutations in genes coding for histones to the pathogenesis of GCLs is unclear, and these studies did not confirm the significance of the mutations by assessing the corresponding protein expression. It has been thought that dysfunctional H.3 histone disrupts epigenetic post-translational modifi- cations in other genes involved in tumorigenesis.⁵⁰ Other studies that compared AA and MF GCLs did not compare them according to biologic behavior (ie, aggressive vs nonaggressive).^4,15,17,51 We believe that similarities exist between aggressive GCLs of the

MF skeleton and the classic GCTs of the AA skeleton.

The nonaggressive variant seems less related to AA GCLs. Our studies have consistently shown that when separating the lesions by behavior, clinically, radiographically, histologically, and now genetically, the lesions exist on a continuous spectrum.

The present study had several limitations. First, because our center is a tertiary referral center, most of the MF GCLs in the present study were of the aggressive subtype. This is in contrast to previous published studies, in which most GCLs would be classified as nonaggressive.²³Given the relative paucity of nonaggressive MF lesions, conclusions regarding histologic comparisons with the other groups might not be significant because of inadequate power to detect differences.

It is unclear how the ability of proteins in formalin- fixed paraffin-embedded specimens to bind antibody

FIGURE 5. Tissue microarray staining of T-cell immune regulator-1 (200) for A, maxillofacial and B, axial/appendicular lesion. Staining is more intense with the giant cells (arrows) than for the mononuclear stromal cells (arrowheads).

(10)

changes or diminishes over time. Many specimens included in the present study were more than 10 years old. In addition, quantification of immunohistochemistry is often unreliable in decalcified specimens.⁵² We minimized these limitations with the use of a TMA.

A TMA has many advantages over standard immunohistochemical staining of individual specimen slides.^53,54 All included subjects’ lesions were added to a single paraffin block, from which sections provide an annotated cohort of lesions to be stained together under the same conditions. The cores used to create the TMA were taken only from areas identified by the pathologist as being tumor, simplifying scoring.

In conclusion, GCLs of the AA and MF skeletons share expression of MMP-9, CTSK, and TCIRG1. This supports our hypothesis that these tumors represent similar lesions in different locations. TCIRG1 has not been described in GCLs and represents a potential target for molecular diagnosis or therapy.

Acknowledgment

The authors acknowledge Brad Tricomi for his assistance with scoring the TMA and performing the statistical analyses.

References

1. Enneking WF: A system of staging musculoskeletal neoplasms.

Clin Orthop Relat Res 204:9, 1986

2. Kruse-L€osler B, Diallo R, Gaertner C, et al: Central giant cell granuloma of the jaws: A clinical, radiologic, and histopathologic study of 26 cases. Oral Surg Oral Med Oral Pathol Oral Radiol En- dod 101:346, 2006

3. Austin LT Jr, Dahlin DC, Royer RQ: Giant-cell reparative granuloma and related conditions affecting the jawbones. Oral Surg Oral Med Oral Pathol 12:1285, 1959

4. Auclair PL, Cuenin P, Kratochvil FJ, et al: A clinical and histomor- phologic comparison of the central giant cell granuloma and the giant cell tumor. Oral Surg Oral Med Oral Pathol 66:197, 1988 5. Waldron CA, Shafer WG: The central giant cell reparative granu-

loma of the jaws: An analysis of 38 cases. Am J Clin Pathol 45:

437, 1966

6. Eisenbud L, Stern M, Rothberg M, et al: Central giant cell granuloma of the jaws: Experiences in the management of thirty-seven cases. J Oral Maxillofac Surg 46:376, 1984

7. Sidhu MS, Parkash H, Sidhu SS: Central giant cell granuloma of jaws—Review of 19 cases. Br J Oral Maxillofac Surg 33:43, 1995 8. Nielsen GP, Rosenberg AE: Diagnostic Pathology: Bone. Salt Lake

City, UT, Amirsys, 2013, pp 9-2–9-25

9. Frassica FJ, Sanjay BK, Unni KK, et al: Benign giant cell tumor.

Orthopedics 16:1179, 1993

10. Ficarra G, Kaban LB, Hansen LS: Central giant cell lesions of the mandible and maxilla: A clinicopathologic and cytometric study.

Oral Surg Oral Med Oral Pathol 64:44, 1987

11. Jaffe HL: Giant-cell reparative granuloma, traumatic bone cyst, and fibrous (fibro-osseous) dysplasia of the jawbones. Oral Surg Oral Med Oral Pathol 6:159, 1953

12. Chuong R, Kaban LB, Kozakewich H, et al: Central giant cell lesions of the jaws: A clinicopathologic study. J Oral Maxillofac Surg 44:708, 1986

13. Resnick CM, Margolis J, Susarla SM, et al: Maxillofacial and axial/

appendicular giant cell lesions: Unique tumors or variants of the same disease?—A comparison of phenotypic, clinical, and radiographic characteristics. J Oral Maxillofac Surg 68:130, 2010

14.Peacock ZS, Resnick CM, Susarla SM, et al: Do histologic criteria predict biologic behavior of giant cell lesions? J Oral Maxillofac Surg 70:2573, 2012

15.Murphey MD, Nomikos GC, Flemming DJ, et al: From the archives of AFIP: Imaging of giant cell tumor and giant cell reparative granuloma of bone: Radiologic-pathologic correlation.

Radiographics 21:1283, 2001

16.Kauzman A, Li SQ, Bradley G, et al: Central giant cell granuloma of the jaws: Assessment of cell cycle proteins. J Oral Pathol Med 33:170, 2004

17.Wang C, Song Y, Peng B, et al: Expression of c-Src and comparison of cytologic features in cherubism, central giant cell granuloma and giant cell tumors. Oncol Rep 15:589, 2006

18.Itonaga I, Hussein I, Kudo O, et al: Cellular mechanisms of osteoclast formation and lacunar resorption in giant cell granuloma of the jaw. J Oral Pathol Med 32:224, 2003

19.Cowan RW, Singh G: Giant cell tumor of bone: A basic science perspective. Bone 52:238, 2013

20.Flanagan AM, Nui B, Tinkler SM, et al: The multinucleate cells in giant cell granulomas of the jaw are osteoclasts. Cancer 62:1139, 1988

21.Chuong R, Kaban LB: Diagnosis and treatment of jaw tumors in children. J Oral Maxillofac Surg 43:323, 1985

22.Enneking WF: Staging of musculoskeletal tumors, in Enneking WF (ed): Musculoskeletal Tumor Surgery. New York, Churchill Livingstone, 1983, pp 87–88

23.Bernier JL, Cahn LR: The peripheral giant cell reparative granuloma. J Am Dent Assoc 49:141, 1954

24.Worth HM: Principles and Practice or Oral Radiology Interpreta- tion. Chicago, Chicago Year Book Medical Publishers, 1963, pp 498–505

25.Bertoni F, Present D, Sudanese A, et al: Giant-cell tumor of bone with pulmonary metastases: Six case reports and a review of the literature. Clin Orthop Relat Res 237:275, 1988

26.Siebenrock KA, Unni KK, Rock MG: Giant-cell tumour of bone metastasising to the lungs: A long-term follow-up. J Bone Joint Surg Br 80:43, 1998

27.Mintz GA, Abrams AM, Carlsen GD, et al: Primary malignant giant cell tumor of the mandible: Report of a case and review of the literature. Oral Surg Oral Med Oral Pathol 51:164, 1981 28.Rock MG, Pritchard DJ, Unni KK: Metastases from histologically

benign giant-cell tumor of bone. J Bone Joint Surg 66:269, 1984 29.Behjati S, Tarpey PS, Presneau N, et al: Distinct H3F3A and H3F3B driver mutations define chondroblastoma and giant cell tumor of bone. Nat Genet 45:1479, 2013

30.Presneau N, Baumhoer D, Behjati S, et al: Diagnostic value of H3F3A mutations in giant cell tumour of bone compared to osteoclast-rich mimics. J Pathol Clin Res 1:113, 2015

31.Gomes CC, Diniz MG, Amaral FR, et al: The highly prevalent H3F3A mutation in giant cell tumours of bone is not shared by sporadic central giant cell lesion of the jaws. Oral Surg Oral Med Oral Pathol Oral Radiol 118:583, 2014

32.Lewis PW, M€uller MM, Koletsky MS, et al: Inhibition of PRC2 activity by a gain-of-function H3 mutation found in pediatric glioblastoma. Science 340:857, 2013

33.Chan KM, Fang D, Gan H, et al: The histone H3.3K27M mutation in pediatric glioma reprograms H3K27 methylation and gene expression. Genes Dev 27:985, 2013

34.Bjerke L, Mackay A, Nandhabalan M, et al: Histone H3.3 mutations drive pediatric glioblastoma through upregulation of MYCN. Cancer Discov 3:512, 2013

35.Barrett T, Suzek TO, Troup DB, et al: NCBI GEO: Mining millions of expression profiles—Database and tools. Nucleic Acids Res 33:D562, 2005

36.Parkinson H, Kapushesky M, Shojatalab M, et al: ArrayExpress—

A public database of microarray experiments and gene expression profiles. Nucleic Acids Res 35:D747, 2007

37.Dai M, Wang P, Boyd AD, et al: Evolving gene/transcript defini- tions significantly alter the interpretation of GeneChip data.

Nucleic Acids Res 33:e175, 2005

38.Bairoch A, Apweiler R, Wu CH, et al: The Universal Protein Resource (UniProt). Nucleic Acids Res 33:D154, 2005 39.Benita Y, Cao Z, Giallourakis C, et al: Gene enrichment profiles

reveal T-cell development, differentiation, and lineage-specific

(11)

transcription factors including ZBTB25 as a novel NF-AT repressor. Blood 115:5376, 2010

40. Aperio Digital Pathology. Available at: http://www.

leicabiosystems.com/pathology-imaging/aperio-digital-pathology/.

Accessed March 20, 2016

41. Prasad K, Prabhu GK: Image analysis tools for evaluation of microscopic views of immunohistochemically stained specimen in medical research—A review. J Med Syst 36:2621, 2012 42. Matos FR, Nonaka CF, Miguel MC, et al: Immunoexpression of

MMP-9, VEGF, and vWF in central and peripheral giant cell lesions of the jaws. J Oral Pathol Med 40:338, 2011

43. Tobon-Arroyave SI, Mideros-Simarra SM, Casta~no-Ramırez LM, et al: Overexpression of matrix metalloproteinase (MMP)-1 and -9 in central giant cell lesions of the jaws: Implications for clinical behavior. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 110:755, 2010

44. Lindeman JH, Hanemaaijer R, Mulder A, et al: Cathepsin K is the principal protease in giant cell tumor of bone. Am J Pathol 165:

593, 2004

45. Schoedel KE, Greco MA, Stetler-Stevenson WG, et al: Expression of metalloproteinases and tissue inhibitors of metalloproteinases in giant cell tumor of bone: An immunohistochemical study with clinical correlation. Hum Pathol 127:1144, 1996 46. Ueda Y, Imai K, Tsuchiya H, et al: Matrix metalloproteinase 9

(gelatinase B) is expressed in multinucleated giant cells of human giant cell tumor of bone and is associated with vascular invasion. Am J Pathol 148:611, 1996

47.Dodds RA, James IE, Rieman D, et al: Human osteoclast cathepsin K is processed intracellularly prior to attachment and bone resorption. J Bone Miner Res 16:478, 2001

48.Del Fattore A, Cappariello A, Teti A: Genetics, pathogenesis and complications of osteopetrosis. Bone 42:19, 2008

49.Hiraoka A, Yano Ki K, Kagami N, et al: Stem cell growth factor: in situ hybridization analysis on the gene expression, molecular characterization and in vitro proliferative activity of a recombi- nant preparation on primitive hematopoietic progenitor cells.

Hematol J 2:307, 2001

50.Yuen BT, Knoepfler PS: Histone H3.3 mutations: A variant path to cancer. Cancer Cell 24:567, 2013

51.Agaram NP, LeLoarer FV, Zhang L, et al: USP6 gene rearrange- ments occur preferentially in giant cell reparative granulomas of the hands and feet but not in gnathic location. Hum Pathol 45:1147, 2014

52.Gertych A, Mohan S, Maclary S, et al: Effects of tissue decalcifica- tion on the quantification of breast cancer biomarkers by digital image analysis. Diagn Pathol 25:213, 2014

53.de Jong D, Rosenwald A, Chhanabhai M, et al: Immunohisto- chemical prognostic markers in diffuse large B-cell lymphoma:

Validation of tissue microarray as a prerequisite for broad clinical applications—A study from the Lunenburg Lymphoma Biomarker Consortium. J Clin Oncol 25:805, 2007

54.Camp RL, Neumeister V, Rimm DL: A decade of tissue microar- rays: Progress in the discovery and validation of cancer biomarkers. J Clin Oncol 26:5630, 2008