EXPRESSION OF GROWTH FACTORS AND THEIR RECEPTORS IN ADAMANTINOMA OF LONG BONES AND THE IMPLICATION FOR ITS HISTOGENESIS

Transcription

1 JOURNAL OF PATHOLOGY, VOL. 184: (1998) EXPRESSION OF GROWTH FACTORS AND THEIR RECEPTORS IN ADAMANTINOMA OF LONG BONES AND THE IMPLICATION FOR ITS HISTOGENESIS JUDITH V. M. G. BOVÉE 1, LAMBERT J. C. M. VAN DEN BROEK 1, WILLEM I. DE BOER 1,2 AND PANCRAS C. W. HOGENDOORN 1 * 1 Department of Pathology, Leiden University Medical Centre, Leiden, The Netherlands 2 Department of Pulmonology, Leiden University Medical Centre, Leiden, The Netherlands SUMMARY Adamantinoma of long bones is a rare bone tumour with (immuno-) histological features of epithelial cells, surrounded by various amounts of osteofibrous tissue. Recent studies have indicated that cells with an epithelial phenotype are most probably the malignant element. There is still debate as to whether the fibrous part should be designated as a benign neoplastic element of a biphasic tumour or as a reactive non-neoplastic tissue next to an epithelioid bone tumour. The expression of fibroblast growth factor type 2 (FGF-2), epidermal growth factor (EGF), and their respective receptors FGFR-1 and EGFR, as well as the proliferation marker Ki-67, was studied in both constituents of adamantinoma in serial sections of 25 cases by immunohistochemistry. Expression of FGF-2 and its receptor was present in both constituents of adamantinoma, but predominated in the epithelial component. Expression of EGF and its receptor was restricted to the epithelial component of adamantinoma. Comparing osteofibrous dysplasia (OFD)-like adamantinoma with classic epithelial cell-rich adamantinoma, the expression of FGF-2, EGF, and EGFR was more intense and in a higher percentage of cells in classic adamantinoma. Proliferative activity was found nearly exclusively in the epithelial component. These data further substantiate the hypothesis that epithelial cells constitute the proliferating tumour cell population responsible for the malignant behaviour of adamantinoma. The data indicate that during progression, the epithelial cells acquire expression of FGF-2, EGF, and EGFR, accompanied by a higher proliferative activity. Within the epithelial cell population, there exists an autocrine pathway of growth stimulation. Furthermore, these data point to an interaction between the epithelial and fibrous components, in which the epithelial cells additionally stimulate fibrous cell growth via a paracrine pathway involving FGF John Wiley & Sons, Ltd. J. Pathol. 184: 24 30, KEY WORDS adamantinoma; bone neoplasms; growth substances; fibroblast growth factor, basic; receptors, fibroblast growth factor; epidermal growth factor-urogastrone; receptors, epidermal growth factor-urogastrone INTRODUCTION Adamantinoma of long bones is a rare bone tumour which histologically consists of abundant fibrous or fibro-osseous tissue in which epithelial cell nests are embedded. 1,2 The occurrence of an epithelioid tumour arising within mesenchymal tissue is intriguing and therefore the histogenesis of adamantinoma has been a matter of debate. Various theories have been suggested, such as misplaced nests of ectodermal epithelium during embryogenesis, trauma with implantation of epithelial cells, low-grade mesenchymal sarcoma differentiating towards an epithelial phenotype, or a synovial or angioblastic origin. 1 Nowadays, the epithelial nature of adamantinoma is beyond challenge, based on ultrastructural and immunohistochemical studies. 3 6 It was shown that aneuploidy and p53 overexpression could be detected in the epithelial component of adamantinoma, whereas in the fibrous part none of these features was found. 7 This suggests that the epithelial part of adamantinoma is neoplastic. *Correspondence to: Pancras C. W. Hogendoorn, MD, PhD, Department of Pathology, Leiden University Medical Centre, P.O. Box 9600, Building 1, L1-Q, 2300 RC Leiden, The Netherlands. P.C.W.Hogendoorn@Pathology.MedFac.LeidenUniv.NL This work was presented in part at the 1996 Annual Meeting of the United States and Canadian Academy of Pathology, Washington, DC. There is still debate as to whether to designate the fibrous component of adamantinoma as a benign neoplastic element, adamantinoma thus being described as a biphasic tumour arising from one stem cell with a dual potential of mesenchymal and epithelial differentiation; or as a reactive tissue proliferation to a monophasic epithelioid bone tumour. Since many studies have indicated an important role for growth factors in tumour growth, we studied the expression of fibroblast growth factor type 2 (FGF-2 or basic FGF), epidermal growth factor (EGF), and their respective receptors FGFR-1 and EGFR in both elements of adamantinoma by immunohistochemistry, to investigate the putative induction of proliferation of fibrous cells by these growth factors. Furthermore, proliferative activity in both components was studied using Ki-67 immunohistochemistry. MATERIALS AND METHODS Patients and tissues Tissue specimens and patients data were obtained from the files of the Netherlands Committee on Bone Tumours, which contains more than bone tumours, collected over a period of more than 40 years. Clinicopathological data have been described earlier. 8 CCC /98/ $17.50 Received 17 July John Wiley & Sons, Ltd. Accepted 9 June 1997

2 GROWTH FACTORS IN ADAMANTINOMA 25 The specimens used for this study included 17 primary tumours, seven local recurrences, and one lung metastasis, from 25 patients. Staining results were compared with clinical data. All histological subtypes were represented: osteofibrous dysplasia (OFD)-like adamantinoma (n=7) as well as classic adamantinoma: tubular adamantinoma (n=6); spindle cell adamantinoma (n=4); mixed adamantinoma (n=3); basaloid adamantinoma (n=2); squamous adamantinoma (n=2); and one lung metastasis. After decalcification, formalinfixed and paraffin-embedded tissue was processed according to standard laboratory practice. Paraffin sections (5 μm) were mounted on 3-aminopropylethoxysilane (APES) (Sigma, St. Louis, MO, U.S.A.) and glutaraldehyde-coated slides and dried overnight at 37 C. Antibodies and controls Monoclonal antibodies directed to bovine FGF-2, type II (Upstate Biotechnology Inc., Lake Placid, NY, U.S.A.), human EGF (Sigma, St. Louis, MO, U.S.A.), EGFR (BioGenex, San Ramon, CA, U.S.A.), and Ki-67, clone MIB-1 (Immunotech SA, Marseille, France) were used. The FGFR-1 antibody was kindly provided by Dr J. Walters (Oxford Brookes University, Oxford, U.K.). The characteristics of this antibody have been described. 9 Immunohistochemistry using antipankeratin, clone 80 (Sanbio, Uden, The Netherlands) was performed to identify the epithelial elements in adamantinoma. For FGF-2 osteoblasts, blood vessel walls and mast cells served as an internal control, whereas osteoblasts and vessel walls also react with anti-fgfr Normal blood vessel walls also served as an internal control for EGF and EGFR immunoreactivity. 13,14 As external positive controls, for FGF-2, specimens of normal tonsil and skin were used, whereas for FGFR-1, specimens of normal skin and for EGF, specimens of normal skin and stomach were used. A sample of normal colon and of a high-grade, decalcified osteosarcoma were used as controls in Ki-67 immunohistochemistry. For pankeratin immunohistochemistry, a specimen of a normal colon was used as positive control. As negative controls, slides were incubated with phosphate-buffered saline containing 1 per cent bovine serum albumin, instead of primary antibodies. Immunohistochemistry Immunohistochemical reactions were performed according to standard laboratory methods as described earlier. 15 After deparaffinization, rehydration, and blocking of endogenous peroxidase, antigen retrieval was performed by incubation with pronase E (Sigma, St. Louis, MO, U.S.A.) (EGFR) or by microwave treatment in a 0 01 M citrate buffer, ph 6 0 (FGF-2, FGFR-1, EGF, pankeratin, MIB-1), followed by an overnight incubation with the primary antibodies against FGF-2 (1:2000), FGFR-1 (1:1200); EGF (1:50), EGFR (1:40), pankeratin (1:5), or MIB-1 (1:200). Biotin-labelled rabbit anti-mouse immunoglobulins and a biotinylated HRP streptavidin complex (DAKO, Glostrup, Denmark) were applied. Visualization was carried out in a diaminobenzidine solution (Sigma, St. Louis, MO, U.S.A.), with the addition of 0 01 M imidazole for FGFR-1. Haematoxylin was used for counterstaining the slides. Evaluation and scoring Staining for FGF-2, FGFR-1, EGF, and EGFR was evaluated by two observers in both the epithelial and the fibrous part of adamantinoma. To differentiate between these two elements, the subsequent slides stained with anti-pankeratin were used to identify epithelial cells. For both elements, the distribution of the staining (focal, which we defined as immunoreactivity only in parts of the slide, or diffuse) and the cellular element that showed positivity (nucleus, cytoplasm or cell membrane) were scored. Ki-67 immunohistochemistry was analysed by counting the number of positive cells per 200 epithelial and per 200 fibrous cells. In OFD-like adamantinoma, quantitative scoring for Ki-67 was not possible because the epithelial elements were too sparse. Statistical analysis Results were correlated with histological subtype and clinical data using Fisher s exact test. RESULTS Immunohistochemistry scoring FGF-2 In slides negative for FGF-2, a positive internal control was found except for five cases. These samples were excluded from further analysis, since prolonged decalcification may have destroyed the immunoreactivity of the antigen in these cases. Two cases could not be evaluated because of repeated loss of attachment of tissue during microwave procedures, leaving 18 tumours for study. Fourteen of 18 tumours showed immunoreactivity for FGF-2 (Table I). In 13 of these 14 positive tumours, staining was found in epithelial cells. Staining patterns were diffuse (n=4) as well as focal (n=9), both nuclear (n=5) and cytoplasmic (n=9). In five tumours, both epithelial and fibrous cells expressed FGF-2. In one tumour, only fibrous cells showed focal immunoreactivity. Immunoreactivity was found in the scattered epithelial cells of two out of five (40 per cent) OFD-like adamantinomas, whereas in classic adamantinomas immunoreactivity was seen in the epithelial cells in 85 per cent (P=0 25, not statistically significant). FGFR-1 Of 25 samples, three were excluded because of technical problems and one sample showed a negative internal control. Of the remaining five OFD-like adamantinomas, the epithelial elements could not be evaluated, since a positive fibrous component in four of them made identification of scattered epithelial cells impossible. One OFD-like adamantinoma, with a positive internal control, did not shown any immunoreactivity. The epithelial element in 16 classical adaman-

3 26 J. V. M. G. BOVÉE ET AL. cells were seen in 50 per cent (P=0 038, statistically significant). Ki-67 In classic adamantinoma, nuclear staining for Ki-67 predominated in epithelial cells, ranging from 2 to 41 positive cells per 200 cells (mean 13 6). In fibrous cells, 0 to 6 positive cells per 200 cells were found (mean 1 7). In OFD-like adamantinoma, scattered positive cells were seen. Correlation with clinicopathological data Expression of FGF-2 was not correlated with previous trauma (P=0 485 (Table II). Expression of FGF-2, FGFR-1, EGF, and EGFR was not correlated with patient outcome (data not shown). DISCUSSION Fig. 1 Light micrograph of an example of an adamantinoma of long bones, showing nests of epithelial cells embedded in (osteo-) fibrous tissue tinomas showed cytoplasmic staining for FGFR-1 in all cases (100 per cent), both focal (n=10) and diffuse (n=6). A fibrous part was present in 13 classic-type tumour samples and staining was seen in 11 of them (85 per cent), focal (n=9) more than diffuse (n=2). Staining tended to be more intense in the epithelial component. EGF Five samples were excluded because of technical problems and one sample showed a negative internal control. Of the remaining 19 samples, the epithelial element of 14 samples showed staining of the cytoplasm for EGF (74 per cent), diffuse (n=11) more than focal (n=3), with variable staining intensities. The fibrous part was negative in all cases. Only one of four OFD-like adamantinomas showed scattered positive epithelial cells, whereas the epithelial element of the classic adamantinomas stained in 13 out of 15 cases (87 per cent) (P=0 036, statistically significant). EGFR Twenty of 24 tumours were positive for EGFR (Table I). One case could not be evaluated due to technical problems. Immunoreactivity for EGFR could be detected only in cell membranes of epithelial cells of adamantinoma diffuse (n=17) more than focal (n=3), with variable staining intensities. All fibrous cells were negative. In classic adamantinomas more intense staining patterns were found in 94 per cent of the cases, whereas in OFD-like adamantinomas scattered positive The expression of FGF-2, EGF, and their receptors in both elements of adamantinoma was studied, in order to investigate whether growth factors play a role in the interaction between the two components of adamantinoma of long bones, with regard to its histogenesis. Both FGF-2 and FGFR-1 were found to be expressed in both constituents of adamantinoma, with a preference for the epithelial component, showing a more intense staining in a higher percentage of samples. Expression of EGF and EGFR was restricted to the epithelial component. Both growth factors and their receptors showed a preference for classic adamantinoma over OFD-like adamantinoma. Proliferative activity predominated in the epithelial component as well. Activation of genes encoding growth factors or growth factor receptors is one of the molecular mechanisms thought to be involved in malignant transformation. 16 FGF-2, a member of the fibroblast growth factor family, affects the growth and function of a wide variety of mesenchymal, endothelial, epithelial, and neural cells. 17,18 The effects of FGF-2 on mesenchymal cells include differentiation of mesenchyme during embryonic development, stimulation of neovascularization, and repair of connective tissue Since a considerable number of patients with adamantinoma (28 50 per cent) have had some significant trauma (including fracture) months or years before development of the tumour, 1,8 we correlated the expression of FGF-2 with previous trauma, in order to exclude FGF-2 immunoreactivity resulting from tissue repair. No correlation was found. Regarding the in vivo cellular distribution of FGFR-1, there is limited information. In normal adult tissues, FGFR-1 immunoreactivity tended to be confined to the tissue microvasculature and some epithelia, 9,12 as well as some other cell types including fibroblasts. 23 The co-expression of FGF-2 and FGFR-1 in both constituents of adamantinoma suggests that both autocrine and paracrine interactions between FGF-2 and FGFR-1 contribute to the progression of this tumour. Epithelial cell-derived FGF-2 probably stimulates both epithelial and fibrous cell proliferation and perhaps also extracellular matrix synthesis and deposition. In addition, the

4 GROWTH FACTORS IN ADAMANTINOMA 27 Fig. 2 Light micrograph of immunohistochemical staining for FGF-2 (A) and FGFR-1 (B). Immunoreactivity for FGF-2 was most prominent in the epithelial component, both nuclear and cytoplasmic. The fibrous component showed scattered immunoreactivity as well. Cytoplasmic immunostaining for FGFR was present in both components, with a higher intensity in the epithelial component immunoreactivity for FGF-2 in the fibrous part in six cases suggests an autocrine pathway by fibrous cellderived FGF-2 in these cases as well. A paracrine action of FGF-2 has also been described for its role in angiogenesis, although the mechanism of secretion is not yet fully understood. 21,22 FGF-2 may therefore also be involved in progression of adamantinoma, through its ability to induce neovascularization. Vascularization of tumour tissue allows increasing tumour size and provides a route for metastases. 21,22 Epidermal growth factor (EGF) and transforming growth factor alpha (TGF-α) are potent mitogens for epithelial, mesenchymal, and glial cells. They both exert their effects in target cells by binding to the EGF-receptor (EGFR), a transmembrane protein tyrosine kinase. 24,25 EGF and TGF-α are known to play an Table I Results of FGF-2, FGFR-1, EGF, EGFR, and Ki-67 immunohistochemistry: a comparison of OFD-like adamantinoma versus classic adamantinoma OFD-like adamantinoma Classic adamantinoma Epithelial Fibrous Epithelial Fibrous* FGF-2 2/5 (40%) 2 /5 (40%) 11/13 (85%) 4/13 (31%) FGFR-1 Non-informative 4/5 (80%) 16/16 (100%) 11/13 (85%) EGF 1/4 (25%) 0/4 (0%) 13/15 (87%) 0/13 (0%) EGFR 3/6 (50%) 0/6 (0%) 17/18 (94%) 0/17 (0%) Ki-67 Scattered positivity Mean 13 6/200 Mean 1 7/200 *In some tumour samples, a fibrous component was lacking, because in metastases and soft tissue extension only the epithelial part is present. One OFD-like adamantinoma showed focal nuclear staining in the fibrous element only, whereas one tumour showed expression of FGF-2 in both elements. Staining was more intense in the epithelial component than in the fibrous component.

5 28 J. V. M. G. BOVÉE ET AL. Fig. 3 Light micrograph of immunohistochemical staining for EGF (A) and EGFR (B). Expression of both EGF and its receptor was confined to the epithelial component, suggesting an autocrine pathway of growth stimulation. EGF staining was present in the cytoplasm and EGFR staining was restricted to the cell membranes of the epithelial cells important role in the development and regulation of a wide variety of malignancies via elevated EGFR levels In adamantinoma, immunoreactivity for EGF and EGFR was found in a high percentage of cases, restricted to the epithelial cells, suggesting a strong autocrine growth stimulation via EGF. This suggests that EGF acts on the epithelial component of adamantinoma, which is consistent with the hypothesis that the epithelial part is the proliferating malignant element, the fibrous part having no malignant potential in itself. 7 Ki-67 is a cell cycle antigen whose expression is closely associated with the proliferation phase and which is absent during the resting phase of the cell cycle. Expression was mainly found in epithelial cells. In a previous study it was shown that 48 per cent of adamantinomas show p53 staining by immunohistochemistry restricted to the epithelial component. 7 As transcription of the p53 gene is cell cycle-dependent, it may be that the low proliferation rate of the fibrous component that we found is responsible for the lack of detectable p53 protein in the fibrous cells. 29 The higher proliferation rate in epithelial cells also indicates that the suggested autocrine growth stimulation of epithelial cells, by EGF and FGF-2, is more effective than the paracrine growth stimulation of fibrous cells by epithelial cell-derived FGF-2. In soft tissue sarcomas, increasing Ki-67 expression correlated positively with tumour size, malignancy grade, necrosis, vascular invasion, S-phase fraction, and metastasis. 30,31 In our study, the number of cases is too small to correlate Ki-67 expression with prognostic factors. Concerning the histogenesis of adamantinoma, the two foremost hypotheses state that adamantinoma arises from an ectopic embryonal epithelial cell rest, with the fibrous part being reactive stroma; or that adamantinoma arises from a mesenchymal stem cell, from which clonal proliferating osteofibrous tissue transforms to an epithelial phenotype, with a subsequent potential for malignant behaviour. The clonal nature of the fibrous part has been suggested by cytogenetic studies Evidence has been proposed for the epithelial part being the malignant element and this malignant behaviour may be the result of secondary genetic events. 7 Our results strongly support the latter theory, since co-expression of FGF-2, EGF, their receptors, and Ki-67 predominated in cells with an epithelial phenotype. Autocrine growth stimulation of epithelial cells via FGF-2 and EGF may be effective in the progression of adamantinoma. The differences in staining percentages between OFD-like adamantinoma and classic adamantinoma are of interest, considering the possible histogenetic relationship between osteofibrous dysplasia (OFD) and adamantinoma. Czerniak et al. 35 and Ueda et al. 36

6 GROWTH FACTORS IN ADAMANTINOMA 29 adamantinomas, more intense staining was found in 87 and 94 per cent for EGF and EGFR, respectively. Similarly, FGF-2 staining was present in the epithelial component of classic adamantinomas in 85 per cent of the cases, as opposed to 40 per cent of OFD-like adamantinomas. This suggests that the epithelial cells in adamantinoma acquire expression of FGF-2, EGF, and EGFR during progression, before developing into classic adamantinoma, supporting the precursor theory. In conclusion, our data further substantiate the hypothesis that the epithelial cells in adamantinoma constitute the proliferating tumour cell population which is responsible for malignant behaviour. During progression, the epithelial cells acquire expression of FGF-2, EGF, and EGFR, accompanied by a higher proliferative activity. Our data indicate that in the epithelial cells, an autocrine pathway of growth stimulation occurs via FGF-2 and EGF. Furthermore, we found evidence for an interaction between the fibrous and the epithelial components of adamantinoma of long bones. The epithelial cells additionally stimulate fibrous cell growth via a paracrine pathway in which FGF-2 is presumed to play a dominant role. Fig. 4 Light micrograph of immunohistochemical staining for Ki-67. Immunoreactivity predominated in the nuclei of the epithelial component postulated that the predominance of an OFD-like pattern in adamantinoma is the result of a secondary reparative process overgrowing matured and regressing tumour tissue. Our results contradict such a fibrous overgrowth, since proliferative activity was about eight times higher in the epithelial component than in the fibrous component. Hazelbag et al. 8 and Springfield et al. 37 regard OFD and OFD-like adamantinoma as precursor lesions, since in OFD and OFD-like adamantinoma the mean age at the time of diagnosis is lower and no metastases have been reported in cases of OFD-like adamantinoma. In our study, expression of EGF and EGFR was only weak in the scattered epithelial cells of a small number of OFD-like adamantinomas. In contrast, in the epithelial cells of classic Table II FGF-2 immunohistochemistry results correlated with previous trauma in order to exclude FGF-2 immunoreactivity resulting from tissue repair; no statistical correlation was found (Fisher s exact test: P=0 485) FGF-2 positive FGF-2 negative Total Trauma No trauma Total ACKNOWLEDGEMENTS We gratefully thank Dr J. Walters (Oxford Brookes University, Oxford, U.K.) for providing the FGFR-1 antibody; The Netherlands Committee on Bone Tumours for providing the cases studied; Dr H. M. Hazelbag for critical comments on an earlier version of the manuscript; and E. Vink, C. B. J. Vos, and N. T. ter Haar for expert technical assistance. REFERENCES 1. Mirra JM. Adamantinoma and osteofibrous dysplasia. In: Bone Tumors. Clinical, Radiologic, and Pathologic Correlations. Philadelphia. London: Lea & Febiger, 1989; Weiss SW. Dorfman HD. Adamantinoma of long bone. An analysis of nine new cases with emphasis on metastasizing lesions and fibrous dysplasia-like changes. Hum Pathol 1977; 8: Rosai J. Adamantinoma of the tibia: electron microscopic evidence of its epithelial origin. Am J Clin Pathol 1969; 51: Pieterse AS, Smith PS, McClure J. Adamantinoma of long bones: clinical, pathological and ultrastructural features. J Clin Pathol 1982; 35: Knapp RH, Wick MR, Scheithauer BW, Unni KK. Adamantinoma of bone. An electron microscopic and immunohistochemical study. Virchows Arch A 1982; 398: Perez-Atayde AR, Kozakewich HPW, Vawter GF. Adamantinoma of the tibia. An ultrastructural and immunohistochemical study. Cancer 1985; 55: Hazelbag HM, Fleuren GJ, Cornelisse CJ, Van den Broek LJCM, Taminiau AHM, Hogendoorn PCW. DNA aberrations in the epithelial cell component of adamantinoma of long bones. Am J Pathol 1995; 147: Hazelbag HM, Taminiau AHM, Fleuren GJ, Hogendoorn PCW. Adamantinoma of long bones. A clinicopathological study of thirty-two cases with emphasis on histological subtype, precursor lesion and biological behavior. J Bone Joint Surg Am 1994; 76A: De-Boer WI, Vermeij M, Gil Diez de Medina S, et al. Functions of fibroblast and transforming growth factors in primary organoid-like cultures of normal human urothelium. Lab Invest 1996; 75: Reed JA, Albino AP, McNutt NS. Human cutaneous mast cells express basic fibroblast growth factor. Lab Invest 1995; 72: Qu Z, Huang X-N, Ahmadi P, et al. Expression of basic fibroblast growth factor in synovial tissue from patients with rheumatoid arthritis and degenerative joint disease. Lab Invest 1995; 73: Hughes SE, Hall PA. Immunolocalization of fibroblast growth factor receptor 1 and its ligands in human tissues. Lab Invest 1993; 69:

7 30 J. V. M. G. BOVÉE ET AL. 13. Perosio PM, Brooks JJ. Expression of growth factors and growth factor receptors in soft tissue tumors; implications for the autocrine hypothesis. Lab Invest 1989; 60: Wilson SE, Schultz GS, Chegini N, Weng J, He Y-G. Epidermal growth factor, transforming growth factor alpha, transforming growth factor beta, acidic fibroblast growth factor, basic fibroblast growth factor, and interleukin-1 proteins in the cornea. Exp Eye Res 1994; 59: Hazelbag HM, Fleuren GJ, Van den Broek LJCM, Taminiau AHM, Hogendoorn PCW. Adamantinoma of the long bones: keratin subclass immunoreactivity pattern with reference to its histogenesis. Am J Surg Pathol 1993; 17: Sporn MB, Todaro GJ. Autocrine secretion and malignant transformation of cells. N Engl J Med 1980; 303: Baird A, Walicke PA. Fibroblast growth factors. Br Med Bull 1989; 45: Cross M, Dexter TM. Growth factors in development, transformation, and tumorigenesis. Cell 1991; 64: Joseph-Silverstein J, Consigli SA, Lyser KM, Pault CV. Basic fibroblast growth factor in the chick embryo: immunolocalization to striated muscle cells and their precursors. J Cell Biol 1989; 108: Fu Y-M, Spirito P, Yu Z-X, et al. Acidic fibroblast growth factor in the developing rat embryo. J Cell Biol 1991; 114: Slavin J. Fibroblast growth factors: at the heart of angiogenesis. Cell Biol Int 1995; 19: Bikfalvi A. Significance of angiogenesis in tumour progression and metastasis. Eur J Cancer 1995; 31A: Johnson DE, Lu J, Chen H, Werner S, Williams LT. The human fibroblast growth factor receptor genes: a common structural arrangement underlies the mechanisms for generating receptor forms that differ in their third immunoglobulin domain. Mol Cell Biol 1991; 11: Boonstra J, Rijken H, Cremers F, Verkleij A, Van Bergen en Henegouwen P. The epidermal growth factor. Cell Biol Int 1995; 19: Burgess AW. Epidermal growth factor and transforming growth factor α. Br Med Bull 1989; 45: Gusterson B, Cowley G, McIlhinney J, Ozanne B, Fisher C, Reeves B. Evidence for increased epidermal growth factor receptors in human sarcomas. Int J Cancer 1985; 36: Lemoine NR, Jain S, Silvestre F, et al. Amplification and overexpression of the EGF receptor and c-erbb-2 proto-oncogenes in human stomach cancer. Br J Cancer 1991; 64: Duda RB, Cundiff D, August CZ, Wagman LD, Bauer KD. Growth factor receptor and related oncogene determination in mesenchymal tumors. Cancer 1993; 71: Vogelstein B, Kinzler KW. p53 function and dysfunction. Cell 1992; 70: Choong PFM, Akerman M, Andersson C, et al. Prognostic value of Ki-67 expression in 182 soft tissue sarcomas. Proliferation a marker of metastasis? APMIS 1994; 102: Choong PFM, Akerman M, Willen H, et al. Expression of proliferating cell nuclear antigen (PCNA) and Ki-67 in soft tissue sarcoma. Is prognostic significance histotype-specific? APMIS 1995; 103: Bridge JA, Dembinski A, DeBoer J, Travis J, Neff JR. Clonal chromosomal abnormalities in osteofibrous dysplasia. Implications for histopathogenesis and its relationship with adamantinoma. Cancer 1994; 73: Hazelbag HM, Wessels JW, Mollevangers P, Van den Berg E, Molenaar WM, Hogendoorn PCW. Cytogenetic analysis of adamantinoma of long bones. Further indications for a common histogenesis with osteofibrous dysplasia. Cancer Genet Cytogenet (in press). 34. Mandahl N, Heim S, Rydholm A, Willen H, Mitelman F. Structural chromosome aberrations in an adamantinoma. Cancer Genet Cytogenet 1989; 42: Czerniak B, Rojas-Corona RR, Dorfman HD. Morphologic diversity of long bone adamantinoma. The concept of differentiated (regressing) adamantinoma and its relationship to osteofibrous dysplasia. Cancer 1989; 64: Ueda Y, Blasius S, Edel G, Wuisman P, Bocker W, Roessner A. Osteofibrous dysplasia of long bones. A reactive process to adamantinomatous tissue. J Cancer Res Clin Oncol 1992; 118: Springfield DS, Rosenberg AE, Mankin HJ, Mindell ER. Relationship between osteofibrous dysplasia and adamantinoma. Clin Orthop 1994; 309: