Search tips
Search criteria 


Logo of jclinpathJournal of Clinical PathologyVisit this articleSubmit a manuscriptReceive email alertsContact usBMJ
J Clin Pathol. 2007 April; 60(4): 410–414.
Published online 2006 July 5. doi:  10.1136/jcp.2006.038828
PMCID: PMC2001122

Utility of immunohistochemical analysis for cyclo‐oxygenase 2 in the differential diagnosis of osteoblastoma and osteosarcoma



To study the immunoexpression of cyclo‐oxygenase (COX) 2 in osteoblastomas (OBs) and osteosarcomas (OSs), and to assess the utility of immunohistochemical analysis for COX 2 in the differential diagnosis of the two tumour forms.


The immunohistochemical features of COX 2 were studied in 11 OBs and 30 OSs, including 26 high‐grade OSs (16 osteoblastic, 7 chondroblastic, and 3 fibroblastic) and 4 low‐grade OSs.


Tumour cells from all 11 OBs unequivocally showed diffuse, intense and cytoplasmic immunoreactivity for COX 2. Strong cytoplasmic expression of COX 2 was observed in 5 of 26 (19%) high‐grade OSs, all chondroblastic. In one osteoblastic‐type OS, COX 2 was expressed in the chondroblastic component, but this tumour was considered to be COX 2 negative. No COX 2 expression was noted in atypical osteoblastic cells. Staining in the four low‐grade OSs was negative.


The results of immunohistochemical analysis of COX 2 suggest that in addition to the routine histopathological evaluation, COX 2 is a valuable diagnostic marker in the distinction between OB and OS.

Osteoblastoma (OB) is an uncommon benign bone‐forming tumour, most frequently occurring in the vertebral column of patients aged <30 years. OBs have a wide spectrum of clinicoradiological and histopathological features. Besides the classic OB, borderline tumours with radiological and histopathological features between OB and osteosarcoma (OS), such as pseudomalignant OB,1,2 aggressive OB or malignant OB,3,4 exist. Moreover, a fraction of OBs may undergo malignant transformation.5,6 It is often difficult to distinguish an OB from an OS by routine histopathological procedures alone.7 Although clinicoradiological findings are sometimes helpful, appropriate immunohistochemical markers are still not available for the differential diagnosis of the two tumour forms.

Cyclo‐oxygenase (COX) is a key biosynthetic enzyme in prostaglandin synthesis and two forms have been identified: COX 1 and COX 2. COX 1 constitutively occurs in normal tissue, whereas COX 2 may be induced in inflammatory tissue.8 Recent studies showed that the expression of COX 2 is increased in various human tumours9; the enzyme seems to play an important role in carcinogenesis, since it can inhibit apoptosis,10 stimulate angiogenesis11 and increase invasion and metastatic potential.12,13 COX 2 expression has been reported in benign bone tumours such as osteoid osteoma, suggesting that the activation of eicosanoid synthesis by COX 2 has biological importance in such tumours.14,15,16 However, there is little information about COX 2 expression in OB, a tumour form that closely resembles osteoid osteoma histologically. Although the expression of COX 2 in OS has been reported in a small series of tumour samples17,18 and some cell lines,19 the distribution of COX 2 has not been fully elucidated. In this study, we investigated the expression profile of COX 2 in OB and OS, and we assessed the utility of immunohistochemical analysis for COX 2 in the differential diagnosis of the two tumour forms.

Materials and methods

Tumour samples and histological evaluation

A total of 41 primary tumour specimens were retrieved from the pathological files of the National Cancer Centre Hospital, Tokyo, Japan, and Sapporo Medical University Hospital, Sapporo, Japan. Tumours included 11 OBs and 30 OSs. Of the 30 OSs, 26 were high grade (16 osteoblastic, 7 chondroblastic and 3 fibroblastic) and 4 were low grade.

The histopathological diagnosis of each tumour was re‐evaluated by TH. The histological criteria of the diagnosis and the determination of the histological grading of OS were based on textbook descriptions.20 An OB in this study was defined as a bone‐forming neoplasm showing woven bone spicules, which are bordered by prominent osteoblasts without atypia (fig 1A,C1A,C).). Conventional OS is a high‐grade (grade 3, 4) malignant tumour characterised by the presence of osteoids (fig 2A,C2A,C).). This high‐grade OS is subdivided into three major groups: osteoblastic OS (bone and/or osteoids are the predominant matrix), chondroblastic OS (chondroids are the predominant matrix) and fibroblastic OS (mainly composed of spindle cells with only minimal amounts of osseous matrix). Low‐grade (grade 1, 2) OS is classified primarily on the basis of a hypocellular to moderately cellular fibroblastic stroma with osteoids. For light microscopic studies, all specimens were fixed in 10% buffered formalin, decalcified in Plank and Rychlo solution (Wako Pure Chemical Industries, Osaka, Japan) and processed routinely for embedding in paraffin wax. Sections of 4 μm thickness were stained with H&E.

figure cp38828.f1
Figure 1 (A) A classic osteoblastoma (OB) showing partially calcified osteoid and immature bone formation associated with osteoblastic activity in a fibrovascular stroma. (B) The osteoblasts are diffusely positive for cyclo‐oxygenase 2 ...
figure cp38828.f2
Figure 2 (A) A chondroblastic osteosarcoma. (B) Many of the chondroblastic cells are diffusely positive for cyclo‐oxygenase 2 (COX 2). (C) An osteoblastic osteosarcoma showing unmineralised bone matrix. (D) None of the tumour cells ...

Immunohistochemical studies

Immunohistochemical analysis was performed on tissue sections from paraffin wax blocks by the labelled streptavidin–biotin method. The sections were dewaxed, rehydrated and moistened with phosphate‐buffered saline (pH 7.4). They were pretreated in an autoclave at 121°C for 10 min in target retrieval solution (pH 9; Dako, Glostrup, Denmark), before being incubated with a mouse monoclonal anti‐COX 2 antibody (clone CX‐294; 1:100; Dako) in an automated immunostaining system (i6000; BioGenex, San Ramon, California, USA) for 30 min. For positive controls, osteoid osteoma tissues known to be positive for COX 2 were used. For negative controls, sections of normal mouse serum were used. The results of COX 2 expression were evaluated by AH and TH. A described consensus judgement21 was adopted as the proper immunohistochemical score of the tumour based on the number of positive cells: 0, negative staining (0–9%); 1+, weak staining (10–29%); 2+, moderate staining (30–49%); 3+, strong staining (50%–).


We studied 11 OBs and 30 OSs ((tablestables 1 and 22).). In patients with OB, eight tumours were localised in the vertebral column and one tumour each was localised in the femur, scapula and radius. Patients with OB were aged 14–31 years (mean 20 years). Clinical details were available for 8 of 11 patients and these 8 patients were treated with local surgical resection or simple curettage of the tumour. Of the eight patients, seven had no local recurrence or metastatic disease; however, one developed a local recurring tumour 14 years after the initial surgical treatment, which showed malignant transformation pathologically. In patients with OS, 26 tumours were localised in the extremities (18 in the femur, 5 in the tibia and 1 each in the ulna, metatarsal bone and metacarpal bone). Four tumours were localised in other anatomical locations (two in the pelvis and one each in the mandible and rib). The patients with OS were aged 8–67 years (mean 29 years). Histologically, 4 were low‐grade OS and 26 were high‐grade OS, in which 16 were osteoblastic, 7 were chondroblastic and 3 were fibroblastic. Follow‐up information was obtained for 26 of 30 OSs; 6 patients died because of their disease.

Table thumbnail
Table 1 Clinical profiles and immunohistochemical cyclo‐oxygenase 2 expression of 11 osteoblastomas
Table thumbnail
Table 2 Clinical profiles and immunohistochemical cyclo‐oxygenase 2 expression of 30 osteosarcomas

Immunohistochemically, proliferating osteoblasts of all the 11 OBs unequivocally showed diffuse, intense and cytoplasmic immunoreactivity for COX 2, and these were classified as strong staining tumours (3+) (fig 1B,D1B,D).). Peritumoral inflammatory cells were also COX 2‐positive, but the extent of staining was weak to moderate. Expression of COX 2 was observed in 5 of 26 (19%) high‐grade OSs and they were all of the chondroblastic type. Unlike OB, however, staining for COX 2 in OS was equivocal and only observed in chondroblastic cells with a cytoplasmic pattern; the staining in these areas was generally strong (3+; fig 2B2B).). In one OS of the osteoblastic type, <9% of the tumour cells showed COX 2 positivity, but they were all chondroblastic cells. In some cases of OS, osteoclast‐like giant cells and macrophages were also weakly positive. No COX 2 expression was noted in atypical osteoblastic cells (fig 2D2D).). Peritumoral inflammatory cells and inflammatory cells adjacent to necrotic tissue showed weak to moderate COX 2 staining. Staining in all four low‐grade OSs was negative.


To differentiate an OB from an OS accurately is of clinical importance because the prognosis and the treatment of the two tumour forms differ. Osteoblastoma has an excellent prognosis and the treatment varies from curettage to local excision. Osteosarcoma, however, is highly malignant with an unfavourable clinical course and the treatment needs a multimodality approach including systemic chemotherapy, radiotherapy and local wide resection to sometimes mutilating en bloc resection of the tumour. Generally, the diagnosis of OB or OS is based on a combination of clinical, radiological and morphological findings. Although OB shows a predilection for the vertebral column and OS frequently affects the metaphysis of the long bone, both forms of tumour can occur in any bone. The age of occurrence and the radiological features of the two forms of tumours often overlap; both tumours occur mainly in the younger age group and cause expansion and destruction of the cortex as well as periosteal bone formation. As described above, histopathologically, OB generally shows an active osteoblastic proliferation with alternating formation of osteoids and woven bone spicules or trabeculae. Scattered foci of osteoclastic bone resorption may appear, but no destructive permeation of pre‐existing bone tissues is noted. In some cases of OB, large and plump osteoblasts with a hyperchromatic nucleus and nucleoli, and occasionally mitoses, may be observed. Owing to the wide spectrum of histopathological findings of OB and the morphological overlapping with OS, some borderline tumours exist, which make the differential diagnosis between OB and OS problematic.1,2,3,4,5,6,7 Unlike OB, OS generally has intense cellular pleomorphism and atypism, extensive areas of necrosis and atypical mitoses, and, in some cases, tumoural cartilage may also be present. The presence of destructive permeation is the most helpful finding in distinguishing OB from OS.7 Nevertheless, the differential diagnosis is sometimes impossible with insufficient material—for example, from a needle biopsy—and may be impossible even with adequate tissue samples.

The absence of reproducible evidence of specific findings minimises the use of immunohistochemistry in the differential diagnosis of the two tumour forms. The literature contains little data regarding COX 2 expression in OBs. Studies on COX 2 expression in osteoid osteoma14,15,16 and chondroblastoma10 suggested that COX 2 expression in these tumours is an important factor for inducing tissue inflammatory reactions. The immunohistochemical feature of COX 2 in the OBs we studied was strikingly similar to that in osteoid osteoma. Whether the relative contribution of COX 2 expression in OB is a factor in inducing inflammatory reaction such as that in osteoid osteoma and chondroblastoma or has another role in tumour development remains to be clarified. Previous authors have reported that a large percentage of OS showed increased COX 2 expression,17,18 but the staining patterns and histological types were not specified. By contrast, the expression of COX 2 in our series was only observed in a limited number of OSs, all of which were of the chondroblastic type. The discrepancy between these results may be due to several reasons, such as different sources of antibodies used, differences in immunohistochemical techniques or different consensus judgement criteria adopted. Moreover, most COX 2‐positive OSs are stained heterogeneously even in the same tissue section, and a chondroblastic component in some conventional OSs are also COX 2 positive. These facts may also lead to a different result if one evaluates the COX 2 staining in one tissue section or on whole tumour sections. COX 2 expression was also detected in a limited number of OS cell lines by using cytogenetic methods,19 but the histological type was not described in detail. It is noteworthy that staining for COX 2 was only observed in chondroblastic cells of these OSs showing COX 2 immunoreactivity. Staining for COX 2 in chondroblastic cells was also shown in chondroblastoma10 and chondrosarcoma.22,23 Although we do not know exactly the incidence of OB that may be impossible to distinguish from OS by classic diagnostic methods, rare borderline tumours between OB and OS do exist.1,2,3,7,24 In this situation, the application of COX 2 immunohistochemistry would be valuable in making a definite assignment.

COX 2 expression in many tumours, in particular those of the gastrointestinal tract, has been strongly implicated in carcinogenesis.25 Pharmacodynamic studies showed that selective inhibitors of COX 2 have the effects of antiangiogenesis and proapoptosis, and therefore suppress tumour growth.26 COX 2‐based treatment is of growing interest and has emerged for clinical use.27 Osteosarcoma is a highly aggressive bone tumour and has a high mortality even when systemic chemotherapy is given. A study of COX 2 inhibitors in OS cell lines showed a possible therapeutic role in counteracting the tumorigenicity of this tumour.19 Since COX 2 inhibitors inhibit tumour growth through both COX 2‐dependent and independent pathways,19,28 further in vitro and in vivo studies are warranted to reveal the roles of COX 2 inhibitors in these tumours without regarding whether they have COX 2 expression. Our immunohistochemical findings, however, suggest that COX 2 would be beneficial in distinguishing between OB and OS in a clinical setting.

In summary, we studied the expression profile of COX 2 in OB and OS and found that there was strong and diffused expression of COX 2 in OB, but it was only observed in the chondroblastic cells of OS. Our findings suggested that in addition to histopathological evaluation, COX 2 is a valuable immunohistochemical marker in the differential diagnosis between OB and OS.

Take‐home messages

  • All the osteoblastomas (OB) showed unequivocally strong immunoreactivity for cyclo‐oxygenase (COX) 2. Expression of COX 2 was observed in 5 of 26 (19%) high‐grade osteosarcomas (OS), all chondroblastic. In one osteoblastic‐type OS, COX 2 was only expressed in the chondroblastic component.
  • The application of COX 2 immunohistochemistry would be helpful in making a definite assignment in case it is difficult to make an accurate diagnosis between OB and OS.


This work was supported by Grants‐in‐Aid for the Scientific Research (Clinical Research for Evidence Based Medicine (H17‐Shoni‐001)) and for the Cancer Research (16‐6) from the Ministry of Health, Labour and Welfare, Japan.


COX - cyclo‐oxygenase

OB - osteoblastomas

OS - osteosarcomas


Competing interests: None declared.


1. Cheung F M, Wu W C, Lam C K. et al Diagnostic criteria for pseudomalignant osteoblastoma. Histopathology 1997. 31196–200.200 [PubMed]
2. Mirra J M, Kendrick R A, Kendrick R E. Pseudomalignant osteoblastoma versus arrested osteosarcoma: a case report. Cancer 1976. 372005–2014.2014 [PubMed]
3. Dorfman H D, Weiss S W. Borderline osteoblastic tumors: problems in the differential diagnosis of aggressive osteoblastoma and low‐grade osteosarcoma. Semin Diagn Pathol 1984. 1215–234.234 [PubMed]
4. Schajowicz F, Lemos C. Malignant osteoblastoma. J Bone Joint Surg Br 1976. 58202–211.211 [PubMed]
5. Mayer L. Malignant degeneration of so‐called benign osteoblastoma. Bull Hosp Jt Dis 1967. 284–13.13
6. Seki T, Fukuda H, Ishii Y. et al Malignant transformation of benign osteoblastoma. A case report. J Bone Joint Surg Am 1975. 57424–426.426 [PubMed]
7. Lucas D R, Unni K K, McLeod R A. et al Osteoblastoma: clinicopathologic study of 306 cases. Hum Pathol 1994. 25117–134.134 [PubMed]
8. Dubois R N, Abramson S B, Crofford L. et al Cyclooxygenase in biology and disease. FASEB J 1998. 121063–1073.1073 [PubMed]
9. Cao Y, Prescott S M. Many actions of cyclooxygenase‐2 in cellular dynamics and in cancer. J Cell Physiol 2002. 190279–286.286 [PubMed]
10. Tsujii M, DuBois R N. Alterations in cellular adhesion and apoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase 2. Cell 1995. 83493–501.501 [PubMed]
11. Tsujii M, Kawano S, Tsuji S. et al Cyclooxygenase regulates angiogenesis induced by colon cancer cells. Cell 1998. 93705–716.716 [PubMed]
12. Kakiuchi Y, Tsuji S, Tsujii M. et al Cyclooxygenase‐2 activity altered the cell‐surface carbohydrate antigens on colon cancer cells and enhanced liver metastasis. Cancer Res 2002. 621567–1572.1572 [PubMed]
13. Tsujii M, Kawano S, DuBois R N. Cyclooxygenase‐2 expression in human colon cancer cells increases metastatic potential. Proc Natl Acad Sci USA 1997. 943336–3340.3340 [PubMed]
14. Kawaguchi Y, Hasegawa T, Oka S. et al Mechanism of intramedullary high intensity area on T2‐weighted magnetic resonance imaging in osteoid osteoma: a possible role of COX‐2 expression. Pathol Int 2001. 51933–937.937 [PubMed]
15. Kawaguchi Y, Sato C, Hasegawa T. et al Intraarticular osteoid osteoma associated with synovitis: a possible role of cyclooxygenase‐2 expression by osteoblasts in the nidus. Mod Pathol 2000. 131086–1091.1091 [PubMed]
16. Mungo D V, Zhang X, O'Keefe R J. et al COX‐1 and COX‐2 expression in osteoid osteomas. J Orthop Res 2002. 20159–162.162 [PubMed]
17. Dickens D S, Kozielski R, Khan J. et al Cyclooxygenase‐2 expression in pediatric sarcomas. Pediatr Dev Pathol 2002. 5356–364.364 [PubMed]
18. Dickens D S, Kozielski R, Leavey P J. et al Cyclooxygenase‐2 expression does not correlate with outcome in osteosarcoma or rhabdomyosarcoma. J Pediatr Hematol Oncol 2003. 25282–285.285 [PubMed]
19. Naruse T, Nishida Y, Hosono K. et al Meloxicam inhibits osteosarcoma growth, invasiveness and metastasis by COX‐2‐dependent and independent routes. Carcinogenesis 2006. 27584–592.592 [PubMed]
20. Dorfman H D, Czerniak B, Kotz R. et al, eds. World Health Organization classification of tumours. Pathology and genetics of tumours of soft tissue and bone Lyon: IARC Press, 2002. 227–232.232
21. Shinmura K, Ishida T, Goto T. et al Expression of cyclooxygenase‐2 in chondroblastoma: immunohistochemical analysis with special emphasis on local inflammatory reaction. Virchows Arch 2004. 44428–35.35 [PubMed]
22. Endo M, Matsumura T, Yamaguchi T. et al Cyclooxygenase‐2 overexpression associated with a poor prognosis in chondrosarcomas. Hum Pathol 2006. 37471–476.476 [PubMed]
23. Sutton K M, Wright M, Fondren G. et al Cyclooxygenase‐2 expression in chondrosarcoma. Oncology 2004. 66275–280.280 [PubMed]
24. Bertoni F, Unni K K, McLeod R A. et al Osteosarcoma resembling osteblastoma. Cancer 1985. 55416–426.426 [PubMed]
25. Williams C D, Mann M, Dubois R N. The role of cyclooxygenases in inflammation, cancer and development. Oncogene 1999. 187908–7916.7916 [PubMed]
26. Thun M J, Henley S J, Patrono C. Nonsteroidal anti‐inflammatory drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues. J Natl Cancer Inst 2002. 94252–266.266 [PubMed]
27. Gasparini G, Longo R, Sarmiento R. et al Inhibitors of cyclo‐oxygenase 2: a new class of anticancer agents? Lancet Oncol 2003. 4605–615.615 [PubMed]
28. Agarwal B, Swaroop P, Protiva P. et al Cox‐2 is needed but not sufficient for apoptosis induced by COX‐2 selective inhibitors in colon cancer cells. Apoptosis 2003. 8649–654.654 [PubMed]

Articles from Journal of Clinical Pathology are provided here courtesy of BMJ Publishing Group