Introduction

Giant cell tumour of bone (GCTOB) is the prototype of giant cell rich neoplasms of the skeleton. The term giant cell tumour was coined by Bloodgood in 1912 [1] and it was not until 1940 that Jaffe distinguished giant cell tumour of bone from other bone tumours containing many osteoclast-like giant cells [2]. This lesion represents 4% to 5% of all primary bone tumors and mainly occurs in skeletally mature patients (peak incidence between ages 20 and 45 years) with a slight female predominance [3–5]. It most commonly arises at the epiphyses of long bones like the distal femur, proximal tibia, distal radius and proximal humerus [6]. This tumor can be locally aggressive with a tendency for recurrence. Lung metastases occur infrequently; more rarely, this tumor behaves as a sarcoma [4, 7]. Because of its different evolution and prognosis, GCTOB must be distinguished from other multinucleated giant cell-containing tumors and pseudotumors. Differential diagnosis can be challenging, particularly in instances of limited sampling such as with needle-core biopsies. It is based not only on histology, but also on clinical and radiological data. There is currently no well-accepted diagnosis marker available for GCTOB, but recent studies using immunohistochemistry and molecular methods have demonstrated overexpression of p63 in the stromal cells of most giant cell tumors of bone and advocate its use as a diagnostic marker [3, 4, 6]. P63 was identified in 1998 [8]. It belongs to the family of transcription factors that also includes p53 and p73 [9]. It is mostly used as a diagnostic aid in breast, prostate, and salivary gland cancer because of its high sensitivity and specificity for mammary and salivary myoepithelial cells and prostatic basal cells [3, 10–12]. It can be a useful tool in distinguishing urothélial carcinoma from prostatic carcinoma [13] and it can also be used as a prognosis factor as in adenoid cystic carcinoma [14].

The purpose of this study is to determine whether GCTOB expresses p63, and whether p63 can be used as a biomarker to discriminate GCTOB from other giant cell-rich tumors.

Methods

This study concerns 48 giant cell-containing tumors and pseudotumors of bone that were retrieved from department of pathology of Hassan II University Hospital in Fez, from January 2009 to February 2012. They include 12 osteosarcomas, 8 osteoblastomas, 5 GCTOB (Figure 1), 5 aneurysmal bone cysts (ABCs) (Figure 2), 4 osteoid osteomas (OO), 4 central giant cell granulomas (CGCGs) (Figure 3), 4 non ossifiant fibromas (NOFs), 3 chondromyxoid fibromas (CMFs), 1 fibrous dysplasia (FD), 1 chondroblastoma and 1 Langerhans cell histiocytosis (LCH). The data were collected prospectively from pathology reports, from forms filled by trauma surgeons, pediatric surgeons and otorhinolaryngologists, and from radiographs. A form was filled for each patient, including the following informations: patient’s name, age, sex, tumor location, histological type and P63 expression. The demographic data and location of these cases are shown in Table 1.

Figure 1
figure 1

Histological findings of giant cell tumor of bone: the tumor is composed of round mononuclear stromal cells and uniformly scattered multinucleated giant cells, many of which contain a large number of nuclei. Characteristically, the nuclei of both stromal and giant cells are very similar. (hematoxylin-eosin stain, original magnification × 200).

Full size image
Figure 2
figure 2

Histological findings of aneurysmal bone cyst: the tumor is composed of blood-filled cystic spaces lined by fibrous septa that are composed of uniform fibroblasts and multinucleated giant cells (hematoxylin-eosin stain, original magnification × 200).

Full size image
Figure 3
figure 3

Histological findings of central giant cell granuloma: the tumor consists of spindled fibroblasts admixed with numerous multinucleated giant cells that tend to be arranged in small clusters. They contain fewer nuclei than seen in giant cell tumour of bone. Scattered lymphocytes are present (hematoxylin-eosin stain, original magnification × 200).

Full size image
Table 1 Demographic data and location of tumours
Full size table

All specimens were fixed in 10% buffered formalin, embedded in paraffin and 4 micron-thick sections were stained with hematoxylin and eosin for routine histological examination.

Immunohistochemical staining

P63 expression was evaluated by immunohistochemistry. All immunohistochemical stains were performed on a Ventana Benchmark LT automated immunostainer, on 3 micron-thick sections that were incubated with a mouse monoclonal antibody against p63 (clones 463M-17, prediluated, ready to use, Cell Marque Datasheet).

The stained slides were examined without knowing the original histologic diagnosis. As there is no consensual scoring, we evaluated intensity of staining as weak (1+), moderate (2+), and strong (3+), and percentage of staining cells. A case was considered positive when nuclear staining of a single lesional cell or more was found.

Statistical analysis

The calculation of average age, median age, sex ratio and rate of P63 expression was done using Epi-info software. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated in the GCTOB vs. not GCTOB and P63 positive vs. P63 negative groups using the SPSS software package (version 17).

Results

The patients’ age ranged between 7 and 59 years with an average of 23.8 years and a median of 21 years. A discrete male predominance was noted (sex ratio = 1.2).

Immunohistochemical analysis showed a P63 nuclear expression in all GCTOB (Figure 4), 2 of 4 osteoid osteomas (50%), 2 of 5 ABCs (40%) (Figure 5), 3 of 8 osteoblastomas (37.5%), 1 of 3 CMFs (33.3%), 1 of 4 NOFs (25%), 1 of 12 osteosarcomas (8.3%) and in the single case of chondroblastoma included in this series. The staining was observed only in the nucleus of the mononuclear cells and no staining was present in the multinucleated giant cells. No P63 immunoreactivity was detected in any of the cases of CGCG (Figure 6), LCH, and FD. Strong staining was seen in 40% of GCTOB (2 cases) and in one case of osteoblastomas (33.3% of P63 positive osteoblastomas). Moderate staining was seen in 2 cases of GCTOBs (40%) and in one case of ABCs. In other tumors expressing P63, staining intensity was weak. Staining was seen in 30%-60% of tumor cells in GCTOB and in 20% and 50% of tumor cells in ABCs. In other tumors, percentage of reactive cells was lower (5%-30% in osteoblastomas, 10% in osteoid osteomas, osteosarcomas and CMFs, and 5% in chondroblastoma and NOFs).

Figure 4
figure 4

Immunohistochemical findings of GCTOB: strong nuclear staining with P63 in mononuclear cells (original magnification × 100).

Full size image
Figure 5
figure 5

Immunohistochemical findings of ABC: moderate and focal nuclear staining with P63 in mononuclear cells (original magnification × 100).

Full size image
Figure 6
figure 6

Immunohistochemical findings of CGCG: negative nuclear staining with P63 in mononuclear cells (original magnification × 100).

Full size image

The sensitivity and negative predictive value (NPV) of P63 immunohistochemistry for the diagnosis of GCTOB were 100%. The specificity and positive predictive value (PPV) were 74.42% and 59.26% respectively.

Discussion

In this study, we showed that all GCTOBs express P63. Dickson [4] and Linden [15] found similar results by immunohistochemistry. They reported P63 overexpression in all GCTOB. In De la Rosa’s study [3], P63 immunoreactivity was seen in 20 of 23 GCTOBs (86.9%). Similar results were reported by Lee [6] who showed a P63 overexpression by immunohistochemistry in 81% of cases (n=26) with a strong staining in 69% (Table 2). The immunostaining was mostly confined to the mononuclear component [3, 4, 6]. This strong expression of P63 suggests that this protein may be implicated in the pathogenesis of GCTOB but determining its exact role requires further investigation.

Table 2 P63 expression of in current series and in other published series
Full size table

The relationship of GCTOB and central giant cell granuloma has long been controversial. The absence of p63 expression in CGCG suggests that these tumors may have a pathogenesis that differs from that of GCTOB. P63 negativity found in all cases of CGCG in our study is consistent with results obtained by Dickson [4] and Lee [6] who found negativity in all cases (n=12 in each series). De la Rosa [3] showed different results with p63 positivity in all cases (n = 4) (Table 2).

Only one case (8.3%) of osteosarcomas included in our study showed overexpression of P63. Proportion of immunoreactive cells was less than 10% and staining was 1+ in intensity. The rate of P63 expression in other series remains low (2 cases/13 in Lee’s study, with low intensity [6], and 2 cases/4 in De la Rosa’s study [3]) (Table 2).

In this work, we recorded a single case of chondroblastoma. The immunohistochemical study showed P63 expression by less than 10% of tumor cells with low intensity. The rate of expression in other studies is variable. In Dickson’s study, 3 of 10 chondroblastomas expressed p63 (30.0%); this ranged from 7–75% of cells, and staining was predominantly mild-moderate in intensity [4]. De Larosa found a higher expression (83.3%, 10 of 12 chondroblastomas) with moderate staining in 6 cases, weak staining in 3 cases and strong staining in only one case [3]. Lee showed P63 staining in 40% of cases (6 of 15). To differentiate between chondroblastoma expressing P63 and GCTOB, he used PS100: chondroblastoma shows positive S-100 immunostaining whereas only occasional weak S-100 immunostaining is seen in GCTOB [6]. In the same study, no P63 staining was seen in chondromyxoïd fibromas (n=12) (Table 2).

The rate of P63 expression in ABC in Dickson’s [4] and Lee’s [6] studies is lower than that obtained in our study: 28.6% (2 cases/7) and 20% (5cases/25) respectively. De la Rosa [3] and Linden [15] found higher results: 62.5% and 100% respectively (Table 2). If some cases of ABC are P63 +, they could be a component of a GCTOB.

In fibrous dysplasia, our results are concordant with those found by De La Rosa [3] (two cases all negative) and Lee (4 cases all negative) [6]. Non ossifiant fibroma showed P63 expression in one case with weak and focal staining. De la Rosa found similar results with P63 expression in 1 of 6 cases (16.6%) (Table 2). Proportion of positive cells was less than 10% and staining intensity was weak [3].

In current study, 50% of osteoid osteomas and 37.5% of osteoblastomas expressed P63. LCH showed no P63 immunostaining. These tumors were not included in the other studies.

P63 contribution in the differential diagnosis between GCTOB and other multinucleated giant cell-containing lesions of bone is variable. Dickson [4] considers that P63 can be useful as a biomarker for the differential diagnosis between GCTOB and other lesions particularly central giant cell granuloma, since these do not express P63. De La Rosa [3] found a high P63 negative predictive value (91.17%) but a low specificity (53.36%) which limits the use of this protein as an immunohistochemical marker for differential diagnosis. Lee [6] considers that the use of P63 can help in histological diagnosis of GCTOB. In current study, the P63 negative predictive value is 100%, this means that in difficult cases, P63 negativity can eliminate a GCTOB. The positive predictive value is low (59.26%). However, except a case of osteoblastoma, a strong staining was found only in GCTOB. Therefore, it is strongly suggestive of this tumor.

Conclusion

This study shows that P63 may serve as a biomarker for the differential diagnosis between GCTOB and other morphologically similar lesions, particularly CGCG since the latter does not express P63. Other giant cell-containing lesions express P63, decreasing its specificity as a diagnostic marker, but a strong staining was seen, except a case of chondroblastoma, only in GCTOB.