Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

MRI of Adamantinoma of Long Bones in Correlation with Histopathology Henk-Jan Van der Woude1,2 Hans-Marten Hazelbag3 Johan L. Bloem2 Antonie H. M. Taminiau4 Pancras C. W. Hogendoorn3

OBJECTIVE. The purposes of this retrospective study were to assess specific MRI features of adamantinoma, including classic adamantinoma and its osteofibrous dysplasia-like variant, and to assess the role of adamantinoma in surgical planning. MATERIALS AND METHODS. MR images of 22 patients with histologically proven adamantinoma, subtyped according to defined criteria, were analyzed, with emphasis on morphologic features, signal intensities, and enhancement parameters. Intra- and extraosseous tumor extent was determined. In all patients, examination of the corresponding resected specimens was performed with regard to tumor extent and presence of multicentricity. Moreover, radiographs were reviewed, and radiographic features derived from the literature were determined. RESULTS. All tumors were primarily localized in the tibia diaphysis (including one patient with additional lesions in the fibula), most frequently in the anterior cortical bone (19/22) with extension toward the bone marrow in 12 patients. We distinguished two morphologic patterns: a solitary lobulated focus versus a pattern of multiple small nodules in one or more foci. Separated tumor foci, defined as foci of high signal intensity on either T2-weighted images or T1-weighted contrast-enhanced images, interspersed with normal-appearing cortical or spongious bone were seen in six patients. All tumors showed intense and homogeneous static enhancement, but there was no uniform dynamic enhancement pattern. No relationship between MRI features and histologic subtype of adamantinoma was found. CONCLUSION. Some uniform MRI characteristics, along with those of radiography, may contribute to the diagnosis of adamantinoma; however, these are not related to the histologic subtype. MRI is pivotal for precise locoregional staging, especially for depiction of distant cortical foci, soft tissue, and intramedullary extension and thus is useful for determining tumor-free margins and strategies for reconstructive surgery.

A

Received February 17, 2004; accepted after revision May 11, 2004. 1 Department of Radiology, Onze Lieve Vrouwe Gasthuis, PO Box 95500, Amsterdam 1090 HM, The Netherlands. Address correspondence to H.-J. Van der Woude. 2

Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands. 3 Department of Pathology, Leiden University Medical Center, Leiden, The Netherlands. 4 Department of Orthopedic Surgery, Leiden University Medical Center, Leiden, The Netherlands.

AJR 2004;183:1737–1744 0361–803X/04/1836–1737 © American Roentgen Ray Society

AJR:183, December 2004

damantinoma of bone is a rare malignant tumor that is almost exclusively seen in the diaphysis of the tibia. The tumor is characterized by slow clinical progression, with the potential to metastasize, mainly to the lungs and usually long after presentation of the primary tumor, with metastasis occurring with increasing frequency after local recurrence [1–5]. This relatively indolent behavior necessitates longterm clinical and radiologic follow-up after primary intervention. Because of the substantial differences in likelihood of metastases, it is of clinical importance to distinguish classic adamantinoma, with an abundant epithelial component, and osteofibrous dysplasia-like adamantinoma, in which only scarce epithelial elements are found in predominant fi-

broosseous tissue [6]. On the basis of a number of tumor biologic and clinical studies, the latter is regarded to be a precursor lesion of the former [6, 7]. Moreover, adamantinoma must be differentiated from completely benign fibroosseous lesions such as fibrous dysplasia also occurring in the tibia [8]. Careful wide en bloc resection of an adamantinoma is pivotal because marginal or intralesional removal significantly increases the risk of local recurrence (≤ 32%) and of lung metastases (≤ 25%) [5, 6]. In this respect, the surgeon is often challenged by the multifocal appearance of the adamantinoma, which may hamper the objective of optimal functional recovery. An optimal preoperative staging procedure is therefore essential. In this respect, findings of radiography alone

1737

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

Van der Woude et al. may underestimate the true tumor extent. In general, because of optimal soft-tissue contrast and multidirectional imaging, MRI is the preferred technique for the staging of musculoskeletal tumors [9]. In the literature, only incidental reports describe the MRI characteristics of adamantinoma [10–16]. In this article, MRI features based on a series of 22 patients and the role of MRI in the staging of adamantinoma are described. Special attention was paid to the multifocal appearance of the tumor and the differential diagnosis between the osteofibrous dysplasia-like and classic adamantinoma subtypes. The latter may be important because of the different clinical behavior of the histologic subtypes.

Materials and Methods Study Group Between 1985 and 2001, 30 patients with adamantinoma were treated at our hospital. MR images were available in 28 patients. Three of the 28 patients were excluded from further analysis be-

cause MRI was performed within a short time after curettage or tumor excision. In three other patients, a recent pathologic fracture prohibited reliable interpretation of the images. As such, 22 patients were included for this retrospective study. Clinical details of eight of these patients were documented previously [6, 17, 18]; a summary is given in Table 1. Radiography Radiographs with both posteroanterior and lateral projections were available in 20 patients. The radiographic features that were determined were derived from the literature [8], and the results are given in Table 2. Histopathology The histologic diagnosis of adamantinoma was made in all patients after biopsy, including immunohistochemical analysis using antibodies directed against a broad variety of cytokeratins [19]. Six patients had a classic adamantinoma (median age of adamantinoma, 26.8 years), whereas in the other 16 patients, a diagnosis of osteofibrous dysplasia-like adamantinoma (median age of osteofibrous dysplasia-like adamantinoma, 16.1 years) was established. Diagnosis of classic adamanti-

TABLE 2

TABLE 1

Clinicopathologic Data of 22 Patients with Adamantinoma in MRI Series

Characteristic Median age (range) (yr) Sex Male Female Histologic subtype Osteofibrous dysplasia-like Classic Primary treatment En bloc resection Curettage Follow-up (yr) Mean (range) No evidence of disease Local recurrence Died of disease

Finding 19 (4–38) 9 13 16 6 20 2a 5.7 (0.5–13) 18 4b 1

aCurettage was performed elsewhere; disease recurred in both patients. After undergoing en bloc resection in our center, both patients are disease-free. bTwo patients are those referred to in footnote “a.” The two other patients had recurrence after en bloc resection, and one died of metastatic disease.

1738

Radiographic Features of Lesions in 20 Patients with Adamantinoma Finding

Anterior bowing Multifocal lesions in tibia Additional lesions in fibula Lesions in Diaphysis only Diaphysis and metaphysis Metaphysis and epiphysis Eccentric localization Geographic destruction Local moth-eaten destruction Predominantly osteolysis Osteolysis with ossification or septation Ground-glass opacity Margins Smooth Lobulated Sclerotic Cortical destruction Regular Irregular Predominantly intracortical location Ballooning

No. of Patients 1 2 1 19 1 0 18 18 1 3 11 1 1 4 8 19 1 17 2

noma was made when clusters of epithelial cells in various differentiation patterns predominated in an osteofibrous background (Fig. 1). The predominance of an osteofibrous dysplasia-like pattern with the presence of scattered keratin-positive epithelial tumor cells led to a diagnosis of osteofibrous dysplasia-like adamantinoma [6]. In all resected specimens, special attention was paid to the potential multicentricity of the lesion, cortical involvement of the tumor, and the rate of bone marrow involvement. In all patients, MRI sections could be compared with histologic macrosections. MRI All patients finally selected for this study underwent MRI before surgery of the primary tumor. The examinations were performed on a 0.5-T (n = 7) or a 1.5-T (n = 15) superconductive scanner. Spin-echo T1-weighted (TR range/TE, 500–600/ 20) sequences were always available. Spin-echo T2-weighted (TR/TE, 2,000/100) or turbo spinecho (TR range/TE, 3,000–4,000/120) sequences with (n = 15) or without fat-selective presaturation were performed in 21 of 22 patients. In 19 of 22 patients, spin-echo T1-weighted sequences after IV injection of contrast medium (gadopentetate dimeglumine, Magnevist, Schering) with or without fat-selective presaturation were performed, preceded by a dynamic contrast-enhanced gradient-echo sequence with a temporal resolution of 3 sec in 14 patients. The following characteristics were evaluated in consensus by two observers and the results are given in Table 3. Tumor size.—The length of the tumors in the longitudinal direction, assessed on coronal or sagittal spin-echo T1-weighted images, was measured in centimeters. Tumor localization.—The exact tumor site within a long bone (epi-, meta-, or diaphysis or all three) was assessed. Cortical involvement and tumor margins.— Cortical involvement, defined as the presence or absence of abnormal signal intensity foci (intermediate on T1-weighted images and high on T2weighted images) as opposed to the low signal intensity of the normal cortical bone, was assessed. Moreover, cortical thickening was recorded, and cortical breach was determined when outer cortical integrity was disturbed by the presence of abnormal (tumor) signal intensity. Bone marrow involvement and tumor margins.—The absence or presence of bone marrow involvement, defined as abnormal signal intensity areas and foci (intermediate on T1-weighted images and opposed to the high signal intensity of normal bone marrow and high on T2-weighted images) was determined, and bone marrow involvement relative to cortical involvement was assessed in each patient. Soft-tissue extension.—The presence or absence of tumor extending ahead of the margins of cortical bone and periosteum was recorded.

AJR:183, December 2004

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

MRI of Adamantinoma of Long Bones

A

B

D

C

E

Fig. 1.—38-year-old man with classic adamantinoma of tibia. Radiologic diagnosis of fibrous dysplasia was established at 2 years old in 1961. At 6 years old, he had fracture, and anterior bowing developed after cast removal. Three years after en bloc resection in 1996, he developed recurrent tumor and lung metastases and died. A, Sagittal spin-echo T1-weighted image reveals anterior bowing of tibia due to presence of multinodular tumor with intermediate signal intensity showing extensive corticomedullary involvement and anterior soft-tissue extension (arrow). B, Sagittal spin-echo T1-weighted image obtained after administration of contrast medium shows homogeneous enhancement (same orientation as A). C, Axial spin-echo T1-weighted gadopentetate dimeglumine–enhanced image with fat suppression shows complete cortical destruction and soft-tissue extension (arrow). F = fibula. D and E, Photomicrographs of lesion after resection show bulging and thinning of cortical bone (C, D) by tumor (T, D) and cortical breakthrough and invasion of surrounding thickened periosteum (P, E). (H and E, ×50) F, Photomicrograph of center of tumor after resection shows classic adamantinoma, strings of epithelial cells (arrow) embedded in fibrous tissue. (H and E, ×100)

F AJR:183, December 2004

1739

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

Van der Woude et al. MRI macromorphology and multicentricity.— Distinction was made between a solitary lesion and the presence of a multifocal appearance (two or more tumor foci within one bone separated by normal-appearing cortical or spongious bone). MRI micromorphology.—A distinction was made between a lobulated pattern of a solitary focus and a pattern of multiple small nodules in a single focus. Signal intensity characteristics.—Signal intensities of tumor relative to muscle on spin-echo T1and T2-weighted images (higher, equal, or lower) were determined.

Static enhancement.—The pattern of tumor enhancement on T1-weighted gadopentetate dimeglumine–enhanced images (absent, peripheral, or homogeneous) was recorded. Dynamic enhancement.—Lesional enhancement within or after 6 sec relative to arterial enhancement on dynamic contrast-enhanced gradient-echo images was recorded; moreover, signal intensity–time curves were analyzed. A distinction was made between a pattern of rapidly progressive enhancement during the first pass of the contrast medium, according to the steepest part of the signal intensity–time curve, followed

by an early plateau phase or washout versus gradual increase of enhancement [20].

Results MRI

Tumor size and localization in bone.—All tumors were located in the tibia. One patient had multiple tumor sites in both the tibia and fibula (Fig. 2). All tumors were located in the diaphyseal part of the tibia. Besides, one patient had a separate lesion in the cortical meta-

A

B

C

D

Fig. 2.—6-year-old boy with classic adamantinoma of tibia and fibula. A, Sagittal spin-echo T1-weighted images show multiple lesions in both anterior and posterior cortical bone and in bone marrow compartment of tibia (arrows). Note lesion in distal fibula (arrowhead). B, On coronal spin-echo T1-weighted images with fat-selective presaturation after contrast medium administration, lesions show intense enhancement and are well demarcated. C, Photograph of histologic macrosection in this specific sagittal plane shows similar anterior cortical bowing of tibia with two tumor foci; one (middle part) with intramedullary extension and one (distal) mainly intracortical. D, Photomicrograph of histopathologic specimen reveals invasion of tumor (T) in bone marrow (BM). (H and E, ×100)

1740

AJR:183, December 2004

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

MRI of Adamantinoma of Long Bones physis, and one patient, in the intramedullary epiphysis (Fig. 3). Longitudinal tumor extension ranged from 2 to 21 cm (mean, 9 cm). Cortical involvement.—Cortical thickening was seen in all patients. In all except three patients, tumor was seen within the anterior cortical bone (Fig. 4). In three patients, only the posterior or posteromedial cortical bone was involved. In 12 (55%) of 22 patients, intracortical tumor without signs of cortical breach was seen on both MR images and corresponding histologic macrosections. Cortical breach, according to the predefined criteria, was seen on both MR images and corresponding histologic sections in seven (32%) of 22 patients. In the remaining three patients (14%), histology showed discrete cortical breakthrough, but this was not scored as such on the MR images. When related to histologic subtype, cortical breakthrough was seen in half of the patients with osteofibrous dysplasia-like adamantinoma and in two of six patients with classic adamantinoma. Bone marrow involvement.—In 20 (91%) of 22 patients, findings on MR images corresponded well with those of the histologic sections with regard to bone marrow extension of the intracortical tumor localization. In eight (40%) of these 20 patients, no bone marrow involvement at all was found. In 12 (60%) of 20 patients, both MR images and histology revealed the presence of tumor in the bone marrow compartment, usually as an extension of the intracortical tumor component (Figs. 1 and 2). In one patient, there were no signs of bone marrow involvement based on the MR images, whereas discrete extension to the bone marrow compartment was shown on microscopy. In one other patient, no histopathologic macrosection of the specimen was available; thus, histologic examination could not reliably be assessed regarding this topic, although the MR images did not show bone marrow involvement. Regarding the histologic subtype, bone marrow extension was seen in four (67%) of six patients with classic adamantinoma and in nine (56%) of 16 patients with osteofibrous dysplasia-like adamantinoma. Soft-tissue extension.—In only two (9%) of 22 patients, tumor extension into the soft tissues was seen (Fig. 1). In the remaining 20 patients (91%), the tumor was limited to the cortical bone and the bone marrow compartment. MRI macromorphology and multicentricity.—Sixteen (73%) of 22 patients had only one tumor focus within the long bone (Fig. 4). Separated tumor foci, defined as foci of

AJR:183, December 2004

Fig. 3.—20-year-old woman with adamantinoma of tibia. Gadolinium-enhanced sagittal T1-weighted images with fat-selective presaturation show lesion consisting of multiple small nodules within anterior cortical bone of diaphysis (arrow). Separate focus is seen in proximal epiphysis (arrowhead).

high signal intensity on either T2-weighted images or T1-weighted contrast-enhanced images, interspersed with normal-appearing cortical or spongious bone were seen in the remaining six patients (Figs. 2 and 3). The latter were three (of six) patients with classic and three (of 16) patients with osteofibrous dysplasia-like adamantinoma. MRI micromorphology.—In 10 (45%) of 22 patients, a pattern of multiple small nodules in one or more foci was seen (Figs. 1 and 3). A lobulated pattern of a solitary focus was found in (41%) of 22 patients (Figs. 2 and 4), and a mixed pattern, in the remaining three patients (14%). No relationship was found between MRI morphology and histologic subtype. Signal-intensity characteristics.—All lesions showed intermediate signal intensity on T1-weighted images (equal to muscle) that was usually homogeneous. On T2weighted images, signal intensity was always high (equal to that of fat in examinations without fat suppression). The signal was homogeneous in 13 (62%) of the 21 patients with images available and heterogeneous in the other eight patients. Static enhancement.—Static contrast-enhanced spin-echo T1-weighted images were available in 19 patients, with fat-selective presaturation in 15 patients. Enhancement was always intense and homogeneous. There was no relationship between enhancement pattern and histologic subtype.

Dynamic enhancement.—Dynamic sequences during and after contrast administration with a high temporal resolution were performed in 14 of 22 patients. In four (29%) of 14 patients, enhancement started more than 6 sec after the earliest arterial enhancement. In most of the remaining patients (10/ 14, 71%), the earliest lesional enhancement was seen within 0–6 sec after arterial enhancement. In the latter 10 patients, rapidly progressive enhancement with an early plateau phase or washout, as reflected by the signal intensity–time curve, was seen. Imaging of the remaining patients revealed a more gradual increase of signal intensity during the dynamic sequence. No relationship existed between the dynamic enhancement pattern and histologic subtype. Radiography

In this series of radiographs comprising imaging of osteofibrous dysplasia-like and classic adamantinomas, the most common radiographic features encountered were eccentric intracortical localization in the tibia diaphysis, a geographic pattern of bony destruction and regular cortical destruction, and a mixed pattern of osteolysis with ossification or septation (Fig. 5 and Table 2). Radiographs showed multifocal lesions in two patients. One satellite lesion in the bone marrow of the epiphysis, shown on MR images, could not be depicted on radiographs.

1741

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

Van der Woude et al.

A

C

B

D

Fig. 4.—29-year-old woman with osteofibrous dysplasia-like adamantinoma of tibia. A, Axial spin-echo T2-weighted image with fat suppression shows well-demarcated single lobulated lesion in thickened anterior cortical bone with high signal intensity. No cortical breakthrough is noted. B, Sagittal spin-echo T1-weighted image shows lesion to be well-demarcated relative to bone marrow compartment by sclerotic rim. C, Sagittal spin-echo T1-weighted image obtained after contrast medium injection shows intense and homogeneous enhancement (same orientation as B). D, Photomicrograph of resected specimen shows intracortical osteofibrous dysplasia-like adamantinoma composed of osteofibrous tissue with woven bone trabeculae (center) rimmed by osteoblasts. No epithelial component is visible at plain section. (H and E, ×50)

1742

AJR:183, December 2004

MRI of Adamantinoma of Long Bones

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

Discussion

Adamantinoma of the long bones comprises only 0.1–0.5% of all primary bone tumors [1–5]. Because of its rarity, reports on the MRI features of the tumor are scarce and restricted to case reports [10–16]. In these publications, the importance of MRI with respect to obtaining information about the extent of the tumor was stressed. Most case reports, however, were restricted to native T1- and T2-weighted images without static and dynamic contrast enhancement. In the present retrospective study, the results of MRI studies, including static and dynamic contrast enhancement analysis, performed before surgery in 22 patients with histologically proven adamantinoma are reported. The intention to distinguish osteofibrous dysplasia-like adamantinoma from classic adamantinoma on the basis of MRI features was one of the goals of this study because the subtypes may have different clinical behaviors [6]. However, neither signal intensity characteristics nor static or dynamic enhancement features contributed to this differentiation. Some consistent morphologic features, in addition to radiographic features, may lead to a likely diagnosis of either classic or osteofibrous dysplasia-like adamantinoma. All tumors were primarily localized in the tibia diaphysis (including tumors in one patient with additional lesions in the fibula), most frequently in the anterior cortical bone. There was, however, discrete or more pronounced extension toward the bone marrow in 60% of patients, as opposed to 40% of patients with tumor localization in the cortical bone only. MRI revealed optimal correlation with corresponding histologic sections regarding cortical involvement and breakthrough in 86% of patients and regarding bone marrow extension in 91%. Two main morphologic tumor patterns were distinguished on MRI: a solitary lobulated focus (in 41% of patients) versus a pattern of multiple small nodules in one or more foci (in 45%). Separated tumor foci, defined as foci of high signal intensity on either T2-weighted or T1-weighted contrast-enhanced images, interspersed with normal-appearing cortical or spongious bone were seen in six (27%) of 22 patients. Although the number is small, this multicentric appearance occurred in three of six patients with classic adamantinoma versus only 18% of the patients with osteofibrous dysplasia-like adamantinoma.

AJR:183, December 2004

Fig. 5.—Radiograph of adamantinoma of tibia in 22-year-old man. Eccentric relatively well-demarcated intracortical lesion in anterior tibia diaphysis shows pattern of mixed lysis and sclerosis and geographic destruction.

This emphasizes the role of MRI for locoregional staging of adamantinoma. Optimal multidirectional imaging of the entire tumor-bearing bone is pivotal, particularly when reconstructive surgery with auto- or allograft transplantation is planned. Because of the preferential manifestation of adamantinoma in the anterior cortical bone of the tibia, sagittal images of the entire lower leg are usually optimal to show the exact cranial–caudal tumor extension and to identify small distant foci within the cortical bone or bone marrow compartment that can be missed on radiographs. Because all small nodules or lobulated tumor areas showed intense static enhancement, contrast-enhanced spin-echo T1-weighted images with fat-selective presaturation images were especially suitable. Ideally, a complete 3D data set is preferable to deal with multiple foci of different sizes and at different sites within the bone. Software is now available that allows the performance of reconstructions based on 3D volume data sets or even on 2D sets with interslice gaps. In regard to reconstructive surgery, axial images, either T1- or T2weighted, are particularly useful to assess the status of the posterior cortical bone. The intense enhancement on static contrastenhanced T1-weighted images, which was seen on images in all patients evaluated with contrast medium, suggests high vascularity of the stroma. This has indeed been encountered when applying immunohistochemistry with vascular markers. The patterns of dynamic enhancement did not correspond with the histologic subtype of adamantinoma. There was also no association between the dynamic enhancement features and a more or less aggressive growth pattern of the osteofibrous dysplasia-like adamantinoma. Probably the amount of capillary vessels and the density of the osteofibrous stroma in osteofibrous dysplasia-like and classic adamantinoma are responsible for the different types of enhancement curves seen in the

TABLE 3

MRI Features of Lesions in 22 Patients with Adamantinoma of Long Bones Finding

Location Tibial cortex Tibia and fibula Diaphysis Diaphysis and metaphysis Diaphysis and epiphysis Anterior cortex Posterior cortex Cortical involvement Intracortical only Cortical breakthrough Marrow involvement Yes No Soft-tissue extension Yes No Multicentricity Yes No Micromorphology Multiple nodules Solitary lobulated nodule Mixed pattern Signal intensity T1: intermediate, homogeneous T2: high, homogeneous T2: high, heterogeneous Static enhancement Homogeneous, intense Dynamic enhancement Temporal resolution in 3 sec ≤ 6 sec > 6 sec Rapidly progressive curve Gradual curve

No. of Patients 21 1 21 1 1 19 3 15 7 12 10 2 20 6 16 10 9 3 20/22 13/21 8/21 19

10 4 10 4

1743

Downloaded from www.ajronline.org by 37.44.207.22 on 01/21/17 from IP address 37.44.207.22. Copyright ARRS. For personal use only; all rights reserved

Van der Woude et al. study population. A high number of vessels may result in a rapidly progressive early enhancement, as opposed to a more gradual increase in signal intensity if a smaller number of capillary vessels are present. The presence of a loose vascularized osteofibrous stroma may result in a continuous increase in signal intensity during the second phase of the dynamic sequence, whereas a more dense stroma corresponding to a tighter interstitial space, may affect the plateau level of the signal intensity–time curve. A hallmark of adamantinoma is its zonal architecture. The typical histopathologic feature of classic adamantinoma is the presence of abundant epithelium within the center, which gradually disappears toward the periphery of the lesion. Thus, at the periphery, the classic adamantinoma may mimic an osteofibrous dysplasia-like adamantinoma, in which the osteofibrous component predominates and isolated small groups or individual keratin-positive cells are scarcely encountered. On the other hand, individual epithelial cells may be completely absent at the periphery; this feature results in the appearance of osteofibrous dysplasia or predominantly reactive osseous tissue. Biopsy samples from these ossified peripheral parts may thus result in the wrong final diagnosis and probably surgical undertreatment. Osteofibrous dysplasia should at least be followed up thoroughly by clinical and radiologic examinations. Clinical or radiographic evidence of growth warrants a repeat biopsy. Although some authors advocate a similar wait-and-see approach for osteofibrous dysplasia-like adamantinoma (which behaves less aggressively than classic adamantinoma), the most appropriate treatment for both histologic subtypes of adamantinoma is considered to be an en bloc resection of the tumor. Biopsies taken from the central part of the tumor, represented by the strongly enhancing foci on T1-weighted images and high signal intensity on T2-weighted images, thus seem to be optimal. This advantage emphasizes the role of MRI before diagnostic surgery. Another entity frequently included in the

1744

differential diagnosis of adamantinoma is fibrous dysplasia. Fibrous dysplasia is in general an intramedullary lesion, with a typical ground-glass pattern of bony destruction on radiographs [8]. Fibrous dysplasia may be treated conservatively, whereas intralesional surgery by curettage is allowed. In contrast, this treatment is absolutely prohibited in adamantinoma because of the high risk of recurrence and potential to metastasize [6]. In conclusion, some uniform MRI characteristics, in addition to those of radiography, may contribute to the diagnosis of adamantinoma; however, these are independent of the histologic subtype, either osteofibrous dysplasia-like or classic. A multicentric appearance seems to occur more frequently in classic adamantinoma. Furthermore, in contrast to a popular opinion that adamantinoma is a purely intracortical tumor, we found extension toward the bone marrow compartment in 60% of patients. MRI is pivotal for precise locoregional staging, especially for depiction of distant cortical foci, soft-tissue extension, and intramedullary extension and, as such, for determination of tumor-free margins and strategy for reconstructive surgery.

8.

9.

10.

11.

12.

13.

14.

15.

References 1. Mulder JD, Schutte HE, Kroon HM, Taconis WK. Radiologic atlas of bone tumors. Amsterdam, The Netherlands: Elsevier, 1993 2. Mirra JM. Adamantinoma and osteofibrous dysplasia. In: Bone tumors, clinical, radiologic and pathologic correlations. Philadelphia, PA: Lea & Febiger, 1989:1204–1231 3. Moon NF. Adamantinoma of the appendicular skeleton: a statistical review of reported cases and inclusion of 10 new cases. Clin Orthop 1965;43:189–213 4. Moon NF. Adamantinoma of the appendicular skeleton: updated. Clin Orthop 1986;204:217–237 5. Kahn LB. Adamantinoma, osteofibrous dysplasia and differentiated adamantinoma. Skeletal Radiol 2003;32:245–258 6. Hazelbag HM, Taminiau AH, Fleuren GJ, Hogendoorn PC. Adamantinoma of the long bones: a clinicopathological study of thirty-two patients with emphasis on histological subtype, precursor lesion, and biological behavior. J Bone Joint Surg Am 1994;76:1482–1499 7. Springfield DS, Rosenberg AE, Mankin HJ, Mindell ER. Relationship between osteofibrous dys-

16. 17.

18.

19.

20.

plasia and adamantinoma. Clin Orthop 1994;309:234–244 Bloem JL, van der Heul RO, Schuttevaer HM, Kuipers D. Fibrous dysplasia vs adamantinoma of the tibia: differentiation based on discriminant analysis of clinical and plain film findings. AJR 1991;156:1017–1023 Bloem JL, Taminiau AHM, Eulderink F, Hermans J, Pauwels EK. Radiologic staging of primary bone sarcoma: MR imaging, scintigraphy, angiography and CT correlated with pathologic examination. Radiology 1988;169:805–810 Brain EC, Raymond E, Goldwasser F, Extra JM, Marty M. Adamantinoma of the proximal end of the tibia: a case [in French]. Presse Med 1994;23:1522–1526 Garces P, Romano CC, Vellet AD, Alakija P, Schachar NS. Adamantinoma of the tibia: plainfilm, computed tomography and magnetic resonance imaging appearance. Can Assoc Radiol J 1994;45:314–317 Judmaier W, Peer S, Krejzi T, Dessl A, Kuhberger R. MR findings in tibial adamantinoma: a case report. Acta Radiol 1998;39:276–278 Tehranzadeh J, Fanney D, Ghandur-Mnaymneh L, Ganz W, Mnaymneh W. Case report 517. Skeletal Radiol 1989;17:614–619 Vanhoenacker F, Van Ongeval C, Ceulemans R, Wouters W, Lateur L. Adamantinoma of the tibia: MRI documentation. J Belge Radiol 1993;76:154–156 Vandermarcq P, Defaux F, Ferrie JC, et al. Adamantinoma of the tibia: X-ray computed tomographic and MRI study—apropos of a case [in French]. J Radiol 1993;74:35–38 Zehr RJ, Recht MP, Bauer TW. Adamantinoma. Skeletal Radiol 1995;24:553–555 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;147:1770–1779 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 1997;97:5–11 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;17:1225–1233 Van der Woude HJ, Verstraete KL, Hogendoorn PCW, Taminiau AHM, Bloem JL. Musculoskeletal tumors: does fast dynamic contrast-enhanced subtraction MR imaging contribute to the characterization? Radiology 1998;208:821–828

AJR:183, December 2004