Main

Giant cell tumor of bone is a locally aggressive neoplasm that represents 5% of all primary bone tumors and has a predilection for long bones and vertebrae of young adults.1 Histologically, giant cell tumor of bone is composed of macrophage-like to spindled mononuclear cells, usually interspersed with numerous large osteoclasts. A cytogenetic hallmark of giant cell tumor of bone is the presence of telomeric associations.2 Importantly, it has been shown that >80% of giant cell tumors harbored H3F3A mutations at the codon Gly34, about 95% of which were p.G34W resulting from a GGG>TGG nucleotide alteration.3, 4, 5, 6 Given their high prevalence in giant cell tumor of bone and near non-existence among potential mimics, H3F3A G34 mutations could serve as a useful diagnostic adjunct.3, 4, 7

Giant cell tumor of soft tissue is histologically similar to giant cell tumor of bone and, as the name indicates, might be regarded as the soft tissue analog of the latter.8, 9, 10, 11 Clinically, similar to its osseous counterpart, giant cell tumor of soft tissue is considered a tumor of low malignant potential, with a tendency to local recurrence while rarely metastasizing. Furthermore, telomeric associations have also been characterized in a single case.12 To determine whether giant cell tumor of soft tissue is similar to its osseous counterpart at the genetic level as well, we analyzed a group of cases for the genotypes of H3F3A and the cognate histone H3 genes containing identical codon 34 sequences (ie, ‘GGG’).

Materials and methods

Tumor Samples

Pathology materials of giant cell tumors of soft tissue, diagnosed between 2010 and 2015, were collected from the consultation files of one of the authors (CDMF). The histology was reviewed and confirmed by all the authors.

DNA Extraction

Tumor tissue, and adjacent normal tissue from selective cases, was dissected from the paraffin tissue slides. DNA was then extracted from the dissected tissue using the BiOstic FFPE Tissue RNA Isolation Kit (Mo Bio Laboratories, Carlsbad, CA, USA) following the manufacturer’s instructions.

PCR and Sanger Sequencing

The extracted DNA was subjected to PCR and bidirectional sequencing, as previously described,13 using specific primers targeting the N-terminal parts of H3F3A, H3F3B, and H3F3C, as well as the whole coding regions of HIST2H3A/C (both genes have identical UTRs and coding sequences and shared the same primers) and HIST2H3D, respectively. Table 1 demonstrates the primer sequences. The electropherograms of H3F3A G34W and H3F3B K36M mutations from a giant cell tumor of bone and a chondroblastoma, respectively, using the same primer sets served as positive controls (Supplementary Figure S1).

Table 1 Sequences of the primers used
Full size table

Results

Clinicopathological Features of the Study Group

A total of 15 giant cell tumors of soft tissue were collected. The patients’ ages ranged from 5 to 89 years (median 63 years). Eleven were female and four were male. The tumor locations included the distal upper extremity (six patients), distal lower extremity (four), thigh (two), arm (one), trunk (one), and face (one). Most tumors were centered on the subcutis. Follow-up data were not collected.

Histologically, the tumors were well circumscribed and showed a striking growth pattern of multiple nodules in 12 cases, separated by sclerotic and often hemosiderotic septa (Figure 1a). The cellular nodules were composed of mononuclear histiocytoid to plump spindle cells and osteoclastic giant cells with rich vasculature (Figure 1b). Both the mononuclear cells and giant cells had similarly round to ovoid nuclei with vesicular chromatin and small distinct nucleoli (Figure 1c). Osseous metaplasia was present in 7 cases (Figures 1a). Mitotic figures ranged from 2 to 18 (median: 5) per 10 high-power fields. Table 2 demonstrates the clinical and pathological information.

Figure 1
figure 1

The histological features of giant cell tumor of soft tissue. (a) The tumor shows a striking multinodular growth pattern, with the tumor nodules separated by sclerotic bands with hemosiderin deposition. Osseous metaplasia is quite often observed (right lower field). (b and c) The major cell types are histiocytoid mononuclear cells admixed with osteoclastic giant cells; both cell types have morphologically similar nuclei that show vesicular chromatin and distinct nucleoli. Rich vasculature is also noted.

Full size image
Table 2 Clinical, pathological, and molecular information of the cases
Full size table

Genotypes of Histone H3 Genes

All cases were shown by direct sequencing to harbor wild-type H3F3A and H3F3B genes. Specifically, no mutation was observed at codon Lys27, Gly34, or Lys36 of either gene (Figures 2a and b).

Figure 2
figure 2

Sanger sequencing demonstrates wild-type H3F3A (a) and H3F3B (b) in all cases. An H3F3C p.H38P single-nucleotide polymorphism (c) and an HIST2H3A/C p.S87S synonymous variant (d) are detected in 3 and 1 case(s), respectively. Highlighted codons are Lys27, Gly34, and Lys36.

Full size image

Sufficient DNA remained for subsequent analyses in 14 cases. Of these, 11 cases had a wild-type H3F3C gene. The other three cases harbored a p.H38P single-nucleotide polymorphism (rs3759295), which was also observed in the respective normal DNA (Figure 2c). Of the 11 cases where the PCR was successful, the HIST2H3A/C and HIST2H3D genes were wild type in all, except for 1 case that had a homozygous p.S87S synonymous variant in the HIST2H3A/C genes (Figure 2d). Specifically, no mutation was found at codon Gly34 in any of these cognate histone H3 genes. See Table 2 for details of the sequencing results.

Discussion

Histones are nuclear proteins that function to package the genome into chromatin units known as nucleosomes. Modifications of histones are central to chromatin remodeling and gene expression regulation, thus being critical to a number of cellular processes.14, 15 Dysregulation of histone modifications has prominent implications in cancer formation.15, 16 Recently, recurrent mutations involving the codons Lys27, Lys36, and Gly34 of the histone variant H3.3 (encoded by the H3F3A and H3F3B genes) have been identified in pediatric glial tumors, chondroblastoma, and giant cell tumor of bone, respectively.3, 17, 18 These mutations probably alter the global landscape of epigenetic modifications through interfering with histone methyltransferases, such as EZH2.19, 20, 21

Giant cell tumors of bone and of soft tissue are similar in multiple aspects. The histological resemblance probably accounts for the shared terminology, as both tumors are characterized by striking number of large osteoclastic giant cells and histiocytoid mononuclear cells. The mononuclear cells in both tumors have been shown to exhibit an osteoblastic phenotype.22 Both often have rich vasculature and can show prominent osseous metaplasia. Clinically, both tumors are considered to be locally aggressive. Cytogenetically, giant cell tumor of bone typically showed clonal and non-clonal telomeric associations, which has also been reported in one giant cell tumor of soft tissue.2, 12 On the other hand, however, there is no ignoring some differences between these two entities. The oftentimes prominent multinodular pattern of soft tissue giant cell tumor is not a feature of the bone counterpart. In contrast to giant cell tumor of bone, which mostly afflicts young to middle-aged adults, giant cell tumor of soft tissue tends to occur in patients with a wider range of age.9, 10, 11, 23 It is also noteworthy that the phenomenon of telomeric associations is not specific and can be found in a variety of neoplasms.24, 25, 26, 27, 28 Therefore, the identification of telomeric associations in just one case of soft tissue giant cell tumor can hardly be regarded as solid evidence that both tumors belong to one unified entity.

In view of the high prevalence of H3F3A Gly34 mutations in giant cell tumor of bone, we carried out genotyping of the H3F3A gene on the soft tissue counterpart to delineate the genetic linkage between these two tumors. The current data revealed no mutation of the H3F3A gene in any of the 15 giant cell tumors of soft tissue, demonstrating a disparity between these tumor types at the genetic level. However, it did not exclude the possible involvement of alternative histone proteins in the pathogenesis of giant cell tumor of soft tissue. As G34W accounts for a vast majority of the H3F3A mutations in giant cell tumor of bone, it is reasonable to postulate that a mutant tryptophan at codon 34 might have virtually irreplaceable biological significance that is specific to this tumor type. Among the codons Gly34 of all histone H3 proteins (coded by GGG, GGC, or GGT), only those with the GGG sequence can be conveniently converted to tryptophan (coded by TGG) with a single nucleotide substitution. This constituted the rationale to scrutinize all histone H3 genes with a codon 34 composed of GGG sequence other than H3F3A, ie, H3F3B, H3F3C, HIST2H3A/C, and HIST2H3D. No somatic mutation was disclosed in these genes. These results further argue for the separation of giant cell tumors of bone and of soft tissue into two distinct entities.

The current study has limitations. First, despite the meticulous dissection to enrich the tumor tissue, the presence of numerous admixed non-neoplastic cells (such as the osteoclastic giant cells) was inevitable and could decrease the sensitivity of mutation detection by Sanger sequencing. Second, as the HIST2H3A and HIST2H3C genes have identical coding and flanking sequences and therefore share the same primer set, the sensitivity of detecting mutations in either gene is further lowered. Finally, the possibility remains that giant cell tumor of soft tissue might harbor aberrations of other histone proteins or associated factors involved in the epigenetic pathways, thus imparting a link to its osseous counterpart at the molecular level. More comprehensive or sensitive techniques, such as genome-wide or deep next-generation sequencing, may be required to dissect these issues.

In conclusion, we have analyzed the genotypes of H3F3A and other cognate histone H3 genes in a series of 15 giant cell tumors of soft tissue. The results suggest that giant cell tumor of soft tissue might be genetically different from its osseous counterpart, and these two tumor types might therefore be better regarded as two distinct entities. More studies are warranted to shed light on the pathogenesis of giant cell tumor of soft tissue.