Bone tumours

the

nucleate cells. These features could result in a misdiagnosis of myxoid chondrosarcoma. However, in CMF the pseudolobulation, with blood vessels and giant cells at the periphery, and the rarity of mitoses combined with the radiological features should allow an accurate diagnosis. Both the chondroid and the myxoid cells react positively with SlOO protein and it is felt that these tumours are genuinely of cartilaginous origin. Eight of the 36 cases developed local recurrence following curettage, some with soft tissue extension, and so en bloc excision is the recommended treatment. Malignant transformation of CMF, although reported rarely, has not been convincingly documented except in one case. Chondroblastomas are rare benign tumours that are associated with the epiphysis or apophysis of long bones in children and young adults. The authors had the unique opportunity of being able to review an astounding 495 cases and so were able to describe unusual and atypical histological features and sites. However, independent of the site and irrespective of histologically atypical features, all of the tumours behaved in a benign fashion with only limited local aggressiveness. Approximately half the cases occurred in the epiphyses of long bones and a quarter affected bones of the hands and feet. Eighty per cent of cases presented in the age group 1&30 years with a sex ratio of 2:l in favour of males. Radiologically the well-demarcated, predominantly lucent appearance of the chondroblastoma, its epiphyseal/apophyseal site, and the age of the patient often combine to give a correct preoperative diagnosis. Typically, chondroblastoma is composed of sheets of oval or round cells with grooved nuclei and abundant cytoplasm associated with foci of chondroid substance, osteoclast-like giant cells, and areas of 'lace-like' calcification. However, this latter feature, often considered a classic feature, was present in only a third of cases.

Chondroblastoma of bone.
Mitoses were found in 77 per cent of cases with up to 3 mitoses/lO HPF and about a quarter of cases had foci of necrosis. Mitotic activity did not correlate with risk of local recurrence. Cytological atypia was present in 28 per cent of cases, especially in chondroblastomas of the skull, mandible, and maxilla. Secondary aneurysmal bone cyst changes were seen in 38 per cent of cases, especially in tumours of the hands and feet. Irrespective of the histology, and despite the presence of vascular invasion in 1.4 per cent of cases, the tumours behaved in an inherently benign or rarely locally aggressive fashion.
Adamantinoma is a rare tumour often misdiagnosed as a metastasis or a benign fibroblastic lesion. The next two papers outline the range of histological features that can be found in adamantinoma, and emphasize the value of immunohistochemistry in the differential diagnosis of fibroblastic lesions of the tibia. One paper alludes to a good prognostic variant of adamantinoma. This is a clinicopathological review of 85 cases confirming that the overwhelming majority of adamantinomas occur in the tibia (70) with the remainder in other long bones (femur, ulna, humerus, fibula, and radius). The patients are usually 10-30 years of age and present with pain. Twenty-six patients developed local recurrences; 1 1 patients died of tumour; and 7 are alive with metastases. This underlines the necessity for wide en bloc excision or amputation for these tumours. The patients often have a long history of symptoms; the tumours are usually diaphyseal and affect the cortex with or without medullary involvement. The histological appearances were those of epithelial islands set in a fibrous stroma with a prominent vascular pattern. The importance of the paper is its emphasis on the remarkable variability of the histological appearances not only from tumour to tumour, but also in different areas within the same tumour. The histology often showed tiny epithelial islands in an extensive fibrous stroma. This stroma often had a storiform arrangement and contained vascular channels with plump endothelial cells giving an appearance similar to the epithelial nests. It was common to find fibrous dysplasia-like or osteofibrous dysplasia-like areas, particularly at the periphery of the lesion. Cytological atypia was an uncommon feature and there was always a low mitotic rate (0-2 mitoses/lO HPF). The authors emphasize that adamantinoma should be considered in the differential diagnosis of any lowgrade spindle cell tumour which involves the tibia. Adamantinoma is most often misdiagnosed as fibrous or osteofibrous dysplasia. Unfortunately the paper does not shed any light on the origin of the epithelial islands. This clinicopathological review of 25 cases of adamantinoma emphasizes the morphological diversity of the tumour, as in the study by Beabout and co-workers (see above). However, this paper discusses the value of immunohistochemical staining of all fibroblastic lesions of the tibia using antibodies to cytokeratin to demonstrate the presence or absence of any epithelial elements. Some cytokeratin-positive cells were demonstrated in all adamantinomas that were available for study irrespective of the pattern of the tumour. The authors then divided the 25 tumours into two groups. The first group (1 7 cases) had classical histology with basaloid, spindled, tubular, or squamous patterns, and the second group (8 cases) had a predominantly osteofibrous dysplasia pattern with very scanty islands of epithelial cells often requiring immunohistochemistry to demonstrate them. This latter group was all intracortical and in younger patients. The authors suggest that this second group be called 'differentiated adamantinoma' and that these tumours and osteofibrous dysplasia represent a different stage of the same process. In other words, the concept of 'differentiated adamantinoma' implies a continuum of lesions with metastasizing adamantinoma at one end and osteofibrous dysplasia at the other end. However, the authors do not rule out the possibility of the existence of de now osteofibrous dysplasia. They further suggest that 'differentiated adamantinoma' may be prognostically more favourable but admit that this, as yet, remains 'not proven'.
The relatively static situation in the diagnosis of primary malignant round cell tumours of bone has given way to exciting rapid progress with the advent of cytogenetic, tissue culture, and immunohistochemical studies of these tumours. In the past, two eminent pathologists (James Ewing and Rupert Willis) were unable to distinguish these tumours from one another and could not even agree on the very existence of Ewing's tumour as a distinct clinicopathological entity. It is now accepted that primary malignant round cell tumours of bone do exist, the controversy, being whether Ewing's tumour (ET) and primitive neuroectodermal tumour of bone (PNET) are histogenetically and prognostically part of a spectrum of the same tumour or are two distinct tumour entities. Cytogenetic studies' have shown an (1 1;22) translocation in both ET and PNET which lends support to the concept that these tumours are part of a spectrum of one tumour that is capable of neural differentiation. The next paper discusses the immunohistochemical and electron microscopic evidence. This is the broadest of the many recent papers discussing the relationship between Ewing's tumour (ET) and other primary malignant round cell tumours of bone such as primitive neuroectodermal tumours (PNET). The authors compared the immunocytochemical reaction of skeletal and extraskeletal ET (18 cases) with PNET (9 cases), primary neuroblastoma (9 cases), and olfactory neuroblastoma (2 cases) using antibodies to SlOO protein, neurone specific enolase (NSE), neuroblastoma cell surface antigen (NCSA), and neurofilament (NF). Electron microscopy (EM) was also performed. In addition, routine stains for glycogen (PAS stain) were used. The patterns of immunoreactivity for ET and PNET were virtually identical, with the vast majority of both tumours being positive with antibodies to NSE, NCSA, and NF, while a small minority of these tumours reacted with SlOO protein. This was in contrast to all four antibodies reacting positively with all neuroblastomas and olfactory neuroblastomas. The presence of glycogen was unreliable as a diagnostic marker, as 16/18 ETs, 7/9 PNETs, and 3/9 neuroblastomas were positive with PAS stain. The EM findings of rare dense core granules, microtubules, and cell processes in PNET separate these tumours from classical ETs, which have none of these features. In this paper the prognosis for ET and PNET showed no difference. However, Pinto et af. ' have suggested that localized ET that reacts with NSE and/or leu 7 carries a better prognosis than localized ET that is immunonegative for these antibodies. The authors of this paper suggest that there is an intimate histogenetic relationship between ET and PNET, and that it might be concluded that ET represents an immature form of PNET. There is a current prospective trial for primary malignant round cell tumours of bone (organized by the UKCCSG and MRC) in which the chemotherapy and surgery of ET and PNET are identical. Examination of the biopsy material using cytogenetics, EM, and immunohistochemistry may help to clarify whether ET and PNET form a spectrum of differentiation, with or without prognostic implications, or whether they are distinct separate entities with therapeutic implications.
Osteosarcoma is not a single tumour entity. It has a variety of subtypes, some of which are simply histological variants without therapeutic or prognostic implications, while others have radically different biological behaviour and so require different treatment. The small cell osteosarcoma falls into the former group, its importance being its similarity to Ewing's tumour. The low-grade central intraosseous osteosarcoma belongs to the latter group. Its importance is in its possible confusion with a benign lesion, on the one hand, and its overtreatment as a high-grade sarcoma, on the other. Small cell osteosarcoma (SCO) is a rare histological variant of high-grade intraosseous osteosarcoma and accounts for about 4 per cent of all osteosarcomas. This study indicates that the age, sex, site, and prognosis of SCO is identical to that of other high-grade intraosseous osteosarcomas. However, the importance of this variant of osteosarcoma lies in its accurate diagnosis and its separation from Ewing's tumour, lymphoma, and mesenchymal chondrosarcoma, all of which require strikingly different treatments. The authors identified three histological patterns: Ewing's-like (1 8), lymphoma-like (4), and spindle cell (5). Only three cases contained malignant cartilage and this always had a high-grade appearance which helped to separate these cases of SCO from mesenchymal chondrosarcoma. The fact that 10 of 21 tumours contained glycogen when stained by the PAS method helped to separate these from lymphoma, but the presence of glycogen added difficulty in separating SCO from Ewing's tumour. There was scanty but definite tumour osteoid in all cases which allowed confirmation of the diagnosis of osteosarcoma. In addition, the radiographic presence of a mineralized matrix outwith the bone would be exceptionally rare in untreated Ewing's tumour but very common in osteosarcoma. Unfortunately, the authors were not able to make use of enzyme histochemistry for alkaline phosphatase (positive in osteosarcomas) nor employ immunohistochemistry using common leucocyte antigen to exclude lymphoma. These techniques are most useful in helping to establish a definitive diagnosis in the 'malignant round cell tumour' group. This is another never-to-be-repeated large collection of a rare tumour entity to be reported from the Mayo Clinic. Eighty cases of low-grade (welldifferentiated) intraosseous osteosarcoma are described with their clinicopathological correlation. This is a very rare variant of osteosarcoma (1 -9 per cent of all osteosarcomas) and its importance is in separating it from fibrous dysplasia, on the one hand, and in recognizing its low-grade nature, on the other. The age, sex, and site are similar to those of conventional high-grade intraosseous osteosarcoma but the prognosis is very much better. All the tumours showed bundles of spindle-shaped cells of uniform appearance with only slight atypia of the nuclei. Most tumours had 1-2 mitoses/l0 HPF with none having more than 4 mitoses/l0 HPF. Periosteal infiltration occurred in 85 per cent of cases with 41 per cent of tumours invading soft tissues. Small clusters of giant cells (36 per cent of tumours) and scattered small foci of atypical cartilage (18 per cent of tumours) were noted. Forty per cent of the tumours had a histological pattern identical to that of a parosteal osteosarcoma with heavy irregular seams of osteoid, whereas 33 per cent had a fibroblastic or desmoid appearance with scanty osteoid. Osteoid in the classic 'Chinese characters' appearance of fibrous dysplasia was present in 14 per cent of cases. Undoubted invasion either in the radiograph or in the biopsy is the single most helpful feature in separating low-grade intraosseous osteosarcoma from fibrous dysplasia. Local excision was almost always associated with local recurrence while wide excision was usually curative. Recurrent tumour presented as a high-grade conventional osteosarcoma in 15 per cent of patients and this was often associated with metastases and death. The vast majority of cases of low-grade intraosseous osteosarcoma are cured by wide local excision alone.

Low
The last two papers do not really fall within the remit of selected summaries in that they are a little dated. However, I have included them because the whole concept of histological assessment of the response of a tumour to chemotherapy is still contentious. Some centres are applying the same approach of pre-operative chemotherapy not only to bone sarcomas, but also to other tumours including soft tissue sarcomas. These two papers remain the most valuable contribution to understanding the validity of post-chemotherapy changes and the methodology that surrounds such assessments. Both papers will give the reader access to an extensive list of references. Pre-operative chemotherapy (PCT) has become the standard treatment for patients with skeletal osteosarcoma. The effect of PCT has been evaluated and expressed as the degree of necrosis seen in microscopic slides of the resected tumour. This paper addresses the question of how much of the tumour necrosis can be attributed to the therapy rather than to spontaneous changes. Sixteen morphological tumour characteristics were analysed qualitatively (3), quantitatively (9, or semiquantitatively (8) in 22 osteosarcomas from patients who had received PCT and 22 osteosarcomas from patients who had not received PCT. Stepwise logistic analysis showed eight observed differences between the two groups. The morphological changes which were most pronounced in the PCT group were (a) damaged tumour cells (necrotic or degenerative); (b) engorged, cystic, or damaged blood vessels; (c) reparative changes involving fibroblasts, angioblasts, and osteoblasts; and (d) the number of 'bizarre' cells. The two most important variables in discriminating between PCT and non-PCT cases were minimal amounts of viable tumour (less than 10 per cent of the total tumour) and strong fibroblastic proliferation. These two variables in combination clearly separated PCT from non-PCT tumours. The discrimination between PCT and non-PCT tumours based on statistical analysis was superior to discrimination based on 'overall impression'. This paper reviews the whole concept of the role and value of pre-operative chemotherapy (PCT) in skeletal osteosarcoma and describes a technique for analysing post-chemotherapy resection specimens. This is essentially to block out an entire longitudinal slab of the resected tumour and assess the presence or absence of viable tumour, expressing this as a percentage of the total surface area of tumourbearing tissue. The authors admit that it is expensive on technical and medical time but claim that it is a highly reproducible method with a variance of only 1-2 per cent when tumour necrosis is greater than 90 per cent, although when there is less than 70 per cent necrosis the reproducibility falls. Using their method of assessment on 40 cases of high-grade intraosseous osteosarcomas it was shown that patients with 90 per cent or more tumour necrosis had a continuous disease-free survival of 9 1 per cent while those with less than 90 per cent necrosis had a disease-free survival of only 14 per cent. The authors then assessed the prognostic value of this with all the other known prognostic factors such as tumour size, etc., and found that percentage necrosis was the single most important prognostic factor. It may be true to say that the amount of tumour necrosis in the 2 12-23 6.

An analysis of spontaneous and chemotherapy
PCT specimen serves as an accurate biological 'marker' for patient prognosis but there is no guarantee that the effects of chemotherapy on the primary tumour will be reflected in metastases, the ultimate cause of death. Indeed, Nachman et aL3 report disparate histologic responses to chemotherapy in simultaneously resected primary and metastatic osteosarcomas. It will be of interest to see whether PCT histological changes are as predictive of prognosis when PCT is applied to a large multicentre trial such as the current European Osteosarcoma Study (MRC Trial 80861). ARCHIE J. MALCOLM University Department of Pathology Royal Victoria Infirmary Newcastle upon Tyne N E l 4 L P , U.K.