| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Musculoskeletal Imaging |
1 From the Departments of Radiologic Pathology (M.D.M.) and Orthopedic Pathology (F.H.G.), Armed Forces Institute of Pathology, 6825 16th St NW, Bldg 54, Rm M-133A, Washington, DC 20306; Departments of Radiology and Nuclear Medicine (M.D.M., D.J.F.) and Surgery (H.T.T.), Uniformed Services University of the Health Sciences, Bethesda, Md; Department of Radiology, University of Maryland School of Medicine, Baltimore, Md (M.D.M.); Department of Surgery, Orthopedic Service, Walter Reed Army Medical Center, Washington, DC (H.T.T.); Department of Radiology, Washington Cancer Institute, Washington Hospital Center, Washington, DC (J.S.J.); and Department of Radiology, National Naval Medical Center, Bethesda, Md (D.J.F.). Received February 25, 2003; revision requested May 23; revision received February 2, 2004; accepted March 2. Address correspondence to M.D.M. (e-mail: murphey@afip.osd.mil).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Data for 40 pathologically confirmed periosteal osteosarcomas were retrospectively reviewed. Patient demographic data were recorded, and radiographs (n = 40), bone scintigrams (n = 10), angiograms (n = 2), and computed tomographic (CT) (n = 11) and magnetic resonance (MR) (n = 12) images were evaluated for lesion location and size, cortical changes, marrow involvement, and intrinsic characteristics by two musculoskeletal radiologists, with agreement by consensus. Pathology reports were reviewed for presence and predominance of histologic components (fibrous, chondroid, and osteoid), tumor grade, and marrow involvement.
RESULTS: There were 25 male (62%) and 15 female (38%) patients with an age range of 10–37 years (average age, 20 years). The most frequent lesion locations were the diaphysis of the tibia (16 patients) or of the femur (15 patients). Radiographs showed a broad-based soft-tissue mass that was attached to the cortex (all patients) and showed cortical thickening (33 patients), cortical scalloping/erosion (37 patients), and/or perpendicular periosteal reaction (38 patients) extending into the soft-tissue mass. Soft-tissue masses were well defined in 91%–100% of cases and surrounded a median of 50%–55% of the cortex. Lesions commonly showed low attenuation at CT (10 patients) and high signal intensity on T2-weighted MR images (10 patients), reflecting the high water content of these largely chondroblastic lesions. Focal areas of adjacent marrow replacement were common at MR imaging (nine patients) but represented reactive changes unless they were in direct continuity with the overlying soft-tissue mass (this was rare, occurring in only one patient, and represented marrow invasion). Review of pathology reports revealed that all lesions contained chondroid tissue, which predominated in 34 patients.
CONCLUSION: The radiologic appearance of periosteal osteosarcoma is a broad-based surface soft-tissue mass causing extrinsic erosion of thickened underlying diaphyseal cortex and perpendicular periosteal reaction extending into the soft-tissue component. Reactive marrow changes are commonly seen at MR imaging, but true marrow invasion is rare.
Index terms: Bone neoplasms, 416.3221, 428.3221, 438.3221, 458.3221, 49.3221 • Bone neoplasms, CT, 416.1211, 428.1211, 438.1211, 458.1211, 49.1211 • Bone neoplasms, diagnosis, 416.3221, 428.3221, 438.3221, 458.3221, 49.3221 • Bone neoplasms, MR, 416.1214, 428.1214, 438.1214, 458.1214, 49.1214 • Osteosarcoma, 416.3221, 428.3221, 438.3221, 458.3221, 49.3221
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Parosteal osteosarcomas have an excellent prognosis (80%–95% long-term survival), and low-grade lesions are usually treated with surgical resection and no neoadjuvant chemotherapy or radiation (1–15). Because of their frequent metaphyseal location, large parosteal osteosarcomas or those with deep medullary invasion may require limb salvage, including joint replacement. In contradistinction, periosteal and high-grade surface osteosarcomas are often treated with neoadjuvant chemotherapy and/or radiation therapy, reflecting their higher degree of anaplasia. The diaphyseal location of the majority of these lesions enables limb salvage with segmental resection that spares the joint surface.
High-grade surface osteosarcomas are more likely to invade the medullary canal (43%) than periosteal lesions (which rarely do so), so high-grade surface osteosarcomas require more extensive surgical resection (1–4,16–28,42–45). In addition, high-grade surface osteosarcomas have a worse prognosis (22% long-term survival) than periosteal lesions (55%–83% long-term survival) (1–4,16–28,42–45).
Periosteal osteosarcoma was described in 1976 by Unni and colleagues (24). Pathologically, these lesions are intermediate- to high-grade tumors that are thought to arise from the inner layer of the periosteum. Histologic assessment reveals largely chondroblastic tissue with smaller areas of osteoid formation.
Periosteal osteosarcoma accounts for approximately 25% of all juxtacortical osteosarcomas (1–5). Like high-grade intramedullary (conventional) osteosarcomas, these lesions affect young patients (in the 2nd and 3rd decades of life) and most frequently (85%–95%) involve the femur and tibia, followed by the ulna and humerus (5%–10%) (1–5,16–41). However, unlike conventional osteosarcomas, these lesions show a strong predilection to arise in the diaphysis.
The classic radiographic findings associated with periosteal osteosarcoma have included thickening of the diaphyseal cortex with scalloping and a perpendicular periosteal reaction extending into a broad-based soft-tissue mass (25). In contradistinction to parosteal osteosarcoma, these lesions only rarely invade the medullary canal (12–15). Previous series have included only a small number of patients with limited quantitative radiologic evaluation or were primarily reviewed from a pathologic perspective. In addition, the computed tomographic (CT) and magnetic resonance (MR) imaging appearances of periosteal osteosarcoma have not been extensively evaluated.
The purpose of our study was to review the imaging appearance of periosteal osteosarcoma, with pathologic comparison.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Clinical Information and Image Evaluation
Clinical characteristics recorded included patient sex, age, and presenting symptoms (eg, pain, swelling, presence of a soft-tissue mass, rapid enlargement of a soft-tissue mass if one was present); anatomic site involved (including longitudinal location [diaphysis, metaphysis, or epiphysis] and transverse location [anterior, posterior, medial, and/or lateral] for long bone lesions); and tumor size (determined in two perpendicular dimensions defining the largest and smallest extents by reviewing images obtained with the modality that best depicted the lesion). Two musculoskeletal radiologists (M.D.M., with 17 years of experience, and J.S.J., with 15 years of experience) with complete knowledge of the pathologic diagnosis reviewed the images, and agreement was reached by consensus. Images reviewed included radiographs (n = 40), angiograms (n = 2), bone scintigrams (n = 10), and CT (n = 11) and MR (n = 12) images.
Radiographs (n = 40) were evaluated for the presence of a soft-tissue mass attached to the cortex, any associated abnormality in the medullary canal, the presence and type of periosteal reaction (either nonaggressive [solid cortical thickening] or aggressive [showing a Codman triangle at the lesion margin or having an irregular or laminated appearance]), the presence of periosteal reaction perpendicular to the bone long axis and extending into the soft-tissue mass, cortical scalloping (including depth, extent, and involvement [whether involving only cortical thickening or affecting the underlying native cortex]), and the presence and extent (mild, moderate, marked) of mineralization in the soft-tissue mass (in addition to perpendicular periosteal reaction) before any therapy.
If available, radiographs that were obtained after initiation of therapy (radiation and/or chemotherapy) were also evaluated for an increase or decrease in the extent and maturity of matrix mineralization compared with the extent and maturity of matrix mineralization on pretherapy images; the extent and maturity of mineralization were considered increased if the appearance of the involved bone progressed toward that of normal bone.
Bone scintigrams (n = 10) were evaluated for the presence, degree (mild, moderate, or marked), homogeneity (homogeneous, mildly heterogeneous, or markedly heterogeneous), and location in bone (central or eccentric) of radionuclide uptake. Angiograms (n = 2) were evaluated for the presence and degree (mild, moderate, or marked) of tumor staining, displacement of native vessels, and the presence of early draining veins.
CT images (n = 11) reviewed included those obtained with both bone and soft-tissue windows at variable thicknesses (n = 1 at 3 mm, n = 4 at 5 mm, n = 6 at 10 mm). CT images were evaluated for the presence of (a) a soft-tissue mass attached to the cortex, (b) any associated abnormality in the medullary canal (eg, replacement of the yellow marrow by soft-tissue attenuation; if replacement was present, it was further characterized as being either continuous with the soft-tissue mass or discontinuous and separated by intervening normal cortex), (c) cortical thickening, (d) cortical scalloping, and (e) periosteal reaction perpendicular to the bone long axis and extending into the soft-tissue mass. CT images were further assessed for the presence and extent (mild, moderate, or marked) of mineralization in the soft-tissue mass (in addition to perpendicular periosteal reaction) and the percentage of the circumference (as graded from 0% to 100% in 10% intervals) of the cortex surrounded by the soft-tissue mass on transverse images.
CT images were also assessed for the predominant attenuation (lower than that of muscle, similar to that of muscle, or higher than that of muscle) and homogeneity (whether homogeneous or heterogeneous) of the nonmineralized component of the soft-tissue mass. The lesion margin was determined as being either defined with a pseudocapsule (ie, a complete rim of high attenuation around the mass), defined without a pseudocapsule, or infiltrative (ie, ill defined). In cases in which images obtained both before and after the administration of intravenous contrast material were available, the degree (mild, moderate, or marked) and the predominant pattern of contrast enhancement (peripheral and septal thin [rim and septa 
2 mm], peripheral and septal thick [rim and septa > 2 mm] with nodularity, peripheral and septal thick without nodularity, peripheral nodular, central nodular, or diffuse) were evaluated.
MR imaging (n = 12) was performed with various imaging units that operated at low field strengths of 0.3–0.5 T (n = 2) or at higher field strengths of 1.0–1.5 T (n = 10). MR imaging sequences used to obtain the images available for review included a standard (spin-echo) T1-weighted sequence (repetition time msec/echo time msec, 400–900/10–20) (n = 10), an intermediate-weighted/T2-weighted sequence (1500–2500/70–100) without fat suppression (n = 12), a short inversion time inversion-recovery sequence (n = 2), and T1-weighted sequences performed before and after injection of a gadolinium chelate (n = 4). Imaging planes included the transverse plane and at least one long axis in all patients.
MR images were evaluated for the presence of (a) a soft-tissue mass attached to the cortex, (b) any associated abnormality in the medullary canal (eg, replacement of the normal yellow marrow signal intensity on T1- and/or T2- weighted MR images; if replacement was present, it was further characterized as being either continuous with the soft-tissue mass or discontinuous and separated by intervening normal cortex), (c) cortical thickening, (d) cortical scalloping, and (d) periosteal reaction perpendicular to the bone long axis and extending into the soft-tissue mass (appearing as low-signal-intensity "rays" on images obtained with all MR pulse sequences). MR images were further evaluated for the presence and extent (mild, moderate, or marked) of mineralization (appearing as nodular foci of low signal intensity on images obtained with all MR pulse sequences) in the soft-tissue mass (in addition to perpendicular periosteal reaction) and the percentage of the circumference (as graded from 0% to 100% in 10% intervals) of the cortex surrounded by the soft-tissue mass on transverse images.
MR images were also assessed for the predominant signal intensity and homogeneity (homogeneous or heterogeneous) of the nonmineralized component of the soft-tissue mass on T1-weighted images (on which the signal intensity of the mass was considered to be low when it was lower than that of muscle, intermediate when it was similar to that of muscle, and high when it was similar to that of fat) and T2-weighted images (on which the signal intensity of the mass was considered to be low when it was similar to that of muscle, intermediate when it was similar to that of fat, and high when it was greater than that of fat).
The lesion margin was determined as being either defined with a pseudocapsule (ie, the lesion showed a low-signal-intensity rim on images obtained with all MR pulse sequences), defined without a pseudocapsule, or infiltrative (ie, ill defined). MR images obtained after intravenous gadolinium chelate injection were evaluated for the degree (mild, moderate, or marked) and the predominant pattern of contrast enhancement (peripheral and septal thin [rim and septa 2 mm], peripheral and septal thick [rim and septa > 2 mm] with nodularity, peripheral and septal thick without nodularity, peripheral nodular, central nodular, or diffuse).
Review of Pathology Reports
Pathology reports (from both outside institutions and the Armed Forces Institute of Pathology) were reviewed (by M.D.M. and F.H.G.) for the presence of various histologic components, including fibrous, chondroid, and osteoid tissue, and for the predominant tissue present. Tumor grade (1–3) was also recorded when available (grade 1 indicated mild increased cellularity and atypia but no bizarre cellular features or necrosis; grade 2, moderate increased cellularity and atypia with readily identifiable mitoses and limited necrosis; and grade 3, marked increased cellularity with pleomorphism, bizarre nuclei, numerous mitoses, and prominent necrosis). Pathology reports were also reviewed for the presence of marrow involvement. If available, evidence of local recurrence or metastatic disease was also recorded.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Osseous locations were the tibia (n = 16), femur (n = 15), phalanx of the hand (n = 2), fibula (n = 2), ulna (n = 2), humerus (n = 1), radius (n = 1), and skull (n = 1). The right side was affected more commonly (n = 23, 58%), and, for tubular bone lesions (n = 39), the diaphysis was involved in 36 patients (92%), while the metaphysis was involved in three patients (8%). The two most common locations were the femur and the tibia, and femoral lesions most frequently occurred anteromedially (n = 5) in the distal (n = 6) or middle (n = 5) portion of the diaphysis; tibial lesions were most frequently seen in the proximal portion of the diaphysis (n = 9) anteromedially (n = 5). Lesions were oval, with an average size of 8.1 x 2.4 cm (range, 23.0 x 5.5 to 2.2 x 1.5 cm).
Imaging Findings
Imaging findings, including the presence of (a) a soft-tissue mass, (b) an abnormality in the medullary canal, (c) cortical thickening and scalloping, (d) perpendicular periosteal reaction extending into a soft-tissue mass, and (e) mineralization in the soft-tissue mass—as well as the circumference of the cortex surrounded by the soft-tissue mass—are summarized in the Table according to the imaging modality (radiography, CT, and MR imaging).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Bone scintigraphy (n = 10) revealed eccentric increased uptake of radionuclide in all patients; this uptake was considered to be of marked extent in nine (90%) patients and of moderate extent in one (10%) patient (Fig 2b). The radionuclide activity was homogeneous in seven (70%) patients and mildly heterogeneous in three (30%) patients. Angiography (n = 2) revealed mild tumor staining and displacement of native vessels but no early draining veins in both cases.
At CT (n = 11), the predominant attenuation of the nonmineralized component of the soft-tissue mass was substantially lower than that of muscle in 10 (91%) patients and similar to that of muscle in one (9%) patient (Figs 3b, 6). The attenuation of the soft-tissue mass was considered to be homogeneous in six (55%) patients and heterogeneous in five (45%) patients. Mineralization in the soft-tissue component was seen in 10 (91%) patients, and the extent of the mineralization was considered to be mild in eight (73%) patients and moderate in two (18%) patients (Fig 3b). Lesion margins were well defined without a pseudocapsule in 10 (91%) patients; in one (9%) patient, an apparent pseudocapsule was identified (Figs 3b, 6). Both pre- and postcontrast CT images were available for only one patient and revealed mild thick (>2 mm) peripheral and septal enhancement with nodularity of the soft-tissue mass (Fig 6).
|
|
|
Lesion margins were well defined with a pseudocapsule in one (8%) patient, well defined without a pseudocapsule in 10 (83%) patients, and appeared infiltrative in one (8%) patient (Figs 2, 3). MR images obtained after intravenous gadolinium chelate administration were available for four patients and showed a mild degree of enhancement in one (25%) patient, a moderate degree of enhancement in one (25%) patient, and a marked degree of enhancement in two (50%) patients. The enhancement pattern was the thick (>2 mm) peripheral and septal enhancement with nodularity pattern in all cases (Figs 3, 6).
Areas of abnormal marrow signal intensity (ie, low signal intensity on T1-weighted MR images and high signal intensity on T2-weighted or short inversion time inversion-recovery MR images) were seen in nine (75%) patients (Fig 2). These foci of abnormal signal intensity were not continuous with the soft-tissue mass (ie, the intervening cortex showed normal low signal intensity) in eight (89%) of these patients, and no pathologic evidence of neoplasm was seen in the medullary canal in these patients (Fig 2). In one patient (11% of the patients with a marrow abnormality at MR imaging) the abnormal marrow signal was continuous with the soft-tissue component (ie, the intervening cortex did not show normal low signal intensity); at pathologic evaluation, these changes were found to represent marrow invasion by tumor (Fig 5).
Histologically, all lesions contained osteoid produced by the malignant cells (Fig 1c). Chondroid elements were present in all cases and predominated in 34 (85%) patients (Fig 1). Fibrous tissue was present in 11 (28%) patients and predominated in six (15%) patients. In cases in which tumor grade was available (n = 18), one (6%) tumor was grade 1, eight (44%) tumors were grade 2, and nine (50%) tumors were grade 3. Marrow involvement was seen in one patient (2%) (Fig 5). Two patients had local recurrences, and two died of metastatic disease to the lungs.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The demographic data and lesion locations in our series, including a male predilection (1.7:1), an average patient age of 20 years, and the tibia (40%) and femur (38%) as the most frequent locations, were similar to those in previous studies (16–41). Unlike parosteal osteosarcomas (which are most commonly located in the posterior and distal portions of the femoral metaphysis), periosteal lesions predominate in the anteromedial portion of the diaphysis, with tibial lesions being most commonly proximal and femoral lesions being more frequent in the middle to distal portions of the diaphysis.
Radiographic findings of periosteal osteosarcoma are often characteristic and have been described in previous series, although they have not been analyzed for their frequency. In our study, periosteal osteosarcoma consisted of a broad-based soft-tissue mass attached to the cortex (100%) with cortical thickening (82%), extrinsic scalloping of the cortex (92%), and periosteal reaction (95%). Several types of periosteal reaction were observed, including both aggressive- and nonaggressive-appearing reactions. The most characteristic periosteal reaction associated with this lesion is aggressive and perpendicular to the osseous long axis (showing a "hair-on-end" appearance), and it extends into the soft-tissue component (95%).
Cortical thickening was more frequently seen in our study (in 82% of cases) than in the study of deSantos and colleagues (25), who observed it in only 67% of their cases, although all of their illustrated cases revealed this finding. In our study, the cortical thickening was often solid and nonaggressive in appearance (51%); a more aggressive periosteal reaction (generating a laminated appearance or a Codman triangle) was seen in only 11% of cases, and both patterns were present in 38% of cases.
Interestingly, in the present study, cortical scalloping was frequently quite shallow, affecting only the thickened cortex in 68% of cases, with involvement of the native cortex in only 32% of cases. Areas of mineralization in addition to periosteal reaction extending into the soft-tissue component were seen on radiographs in 68% of cases. Increasing maturity and extent of ossification in the soft-tissue mass were noted as a response to chemotherapy and/or radiation therapy in all six patients in whom posttherapy radiographs were available, although this finding could not be quantified to reflect the degree of tumor necrosis. Results of bone scintigraphy reflected these radiographic changes; bone scintigraphy revealed eccentric uptake of radionuclide (in 100% of cases) that was usually marked in degree (in 90% of cases) and homogeneous (in 70% of cases).
CT and MR imaging also revealed the changes of cortical thickening (in 91% and 83% of cases, respectively), cortical scalloping (in 91% and 92% of cases, respectively), and periosteal reaction extending into the soft-tissue mass (in 91% and 75% of cases, respectively) in the majority of patients. However, as is the case with other osseous neoplasms, radiography was superior for this evaluation. CT demonstrated additional areas of calcification in the soft-tissue component in 91% of cases; this reflects the superior contrast resolution available with CT versus that available with radiography (which revealed additional areas of calcification in 68% of cases). The finding of calcification corresponded to the histologic results in this matrix-producing tumor; osteoid was seen in all cases. Periosteal reaction extending perpendicularly from the cortex into the soft-tissue mass was seen at MR imaging in only 75% of cases as rays of low signal intensity on images obtained with all pulse sequences. This probably resulted from the high-signal-intensity mass obscuring the periosteal reaction at MR imaging in some cases (as compared with the appearance of this reaction at radiography and CT).
The intrinsic imaging characteristics of the soft-tissue mass reflected the pathologic nature of this largely chondroid neoplasm and the inherent high water content (75%–80%) of hyaline cartilage. In the present study, CT revealed lower attenuation than the attenuation of muscle in 91% of cases, and high signal intensity was seen on T2-weighted MR images in 83% of cases. This corresponds to the results of our pathologic review, in which we found that all lesions had a cartilage component, which was the predominant tissue present in 85% of lesions. This feature is important to recognize, because biopsy of a highly cartilaginous region may lead to an erroneous diagnosis of chondrosarcoma, with substantial treatment implications (ie, chemotherapy would not be offered for a chondrosarcoma but it would for an osteosarcoma).
In our study, lesion margins were frequently well defined at CT (91%) and MR imaging (92%), as demonstrated in previous pathologic studies (4,5,13,14). Contrast enhancement was septal and peripheral in all of our cases at CT and MR imaging; this enhancement pattern has been described as characteristic of cartilaginous lesions (47,48). However, interestingly, the nodular features and prominent degree of enhancement (moderate or marked in 75% of cases at MR imaging) may be related to the more vascularized malignant tissue and the intermixed fibrous and osteoid components of this lesion. The soft-tissue masses surrounded a median of 50%–55% of the bone circumference in our series. In our opinion, this may be helpful in distinguishing these lesions from high-grade surface osteosarcomas, which, in our experience, often surround the entire outer cortex (1–4,42–45).
Involvement of the medullary canal by periosteal osteosarcoma is a controversial subject, with some authors suggesting that it does not occur or that if it is present, the diagnosis of periosteal osteosarcoma should be excluded. We agree with Hall et al (21) and Sonobe et al (26) that, while unusual (one case [2%] in our series), medullary invasion may rarely occur and should not preclude the diagnosis if other features are distinctive. Although periosteal osteosarcoma is slow growing and the cortex acts as a neoplastic barrier, ultimate penetration by this malignancy should not be surprising.
Intramedullary invasion was detected in our series in only one case at MR imaging as direct continuity between the overlying surface tumor and the foci of marrow replacement and was confirmed pathologically. Intramedullary invasion is to be distinguished from areas of marrow that show low signal intensity on T1-weighted MR images and high signal intensity on MR images obtained with water-sensitive pulse sequences (eg, those involving T2 weighting and/or short inversion time inversion recovery) but are not continuous with the overlying surface tumor (as seen in 75% of our cases at MR imaging). These regions presumably represent reactive marrow changes, because no tumor was seen at pathologic examination. The intrinsic MR signal characteristics alone do not enable differentiation of tumor from marrow edema because both have a similar appearance. This fact emphasizes the importance of the relationship of marrow replacement at MR imaging to the surface lesion. The distinction of tumor from marrow edema is very important in directing the extent of surgical resection.
The differential diagnosis of periosteal osteosarcoma includes juxtacortical chondrosarcoma, Ewing sarcoma, parosteal osteosarcoma, and high-grade surface osteosarcoma. Juxtacortical chondrosarcoma usually affects older patients (in their 4th to 5th decades of life), most commonly occurs in the metaphysis, and often shows extensive osteoid and chondroid mineralization (49–52). In addition, although cortical scalloping and thickening are frequent, perpendicular periosteal reaction is usually not apparent.
Ewing sarcoma is only rarely periosteal (3% of cases) without medullary involvement but can closely simulate periosteal osteosarcoma in terms of patient demographics, lesion location, and radiologic appearance (53–57). However, mineralization within the soft-tissue component, perpendicular periosteal reaction (before chemotherapy), and low attenuation in the soft-tissue component at CT and very high signal intensity of the component on T2-weighted MR images are not typical features of periosteal Ewing sarcoma (53–57).
Parosteal osteosarcoma characteristically affects older patients (in the late 3rd or 4th decade of life), involves the metaphysis (particularly the posterior part of the distal portion of the femur) without perpendicular periosteal reaction, and is initially attached to the bone by a narrow stalk (12–15). High-grade surface osteosarcoma is often the most difficult lesion to distinguish from periosteal osteosarcoma because both are usually diaphyseal and show a perpendicular periosteal reaction. In our experience, compared with periosteal osteosarcoma, high-grade osteosarcomas usually surround a much higher percentage of the bone circumference, are more likely to invade the medullary canal, and do not consist of a high-water-content soft-tissue mass that shows low attenuation at CT, very high signal intensity on T2-weighted MR images, and/or peripheral and septal enhancement resulting from the presence of a cartilaginous component.
We acknowledge that there were limitations to this study, including its retrospectively analytic nature, the relatively small number of CT and MR images evaluated, the fact that the observers were not blinded to the diagnosis, the fact that no comparison to other surface osteosarcomas was made, the use of consensus for evaluation, and the inability to control imaging parameters and directly map the imaging findings to the pathologic specimens. Also, the referral nature of our study group made evaluation of patient follow-up findings and outcome difficult to impossible. However, two of our patients had local recurrences that were treated with wide reexcision, and two patients died of metastatic disease to the lungs.
Finally, the classic definition of periosteal osteosarcoma, like the definitions of many osseous neoplasms, is not based strictly on pathologic findings but also encompasses the radiographic appearance. This inclusion of the radiographic appearance in the lesion definition creates inherent bias when one is evaluating the radiologic characteristics of the lesion, although this was also the case in all previous studies of periosteal osteosarcoma. Despite these limitations, we believe our results add marked understanding of the imaging appearance of periosteal osteosarcoma.
In conclusion, periosteal osteosarcomas often show a distinctive imaging appearance—that of a diaphyseal lesion involving the femur or tibia with cortical thickening that is extrinsically eroded (scalloped) by a broad-based soft-tissue mass attached to the cortex. Periosteal reaction (perpendicular to the long axis of the affected bone) extends into the soft-tissue mass, and additional areas of mineralization are frequently seen on radiographs. CT and MR imaging also show these features and reveal the extent of the soft-tissue mass. CT and MR imaging also reflect the largely chondroid tissue seen pathologically, which shows low attenuation at CT and high signal intensity at T2-weighted MR imaging. MR imaging commonly reveals foci of marrow replacement in the region of the tumor. However, medullary invasion is rare and should only be suggested when the marrow replacement is in continuity with the surface soft-tissue component. Marrow signal abnormality that is not contiguous with the remainder of the tumor represents reactive changes, and distinction of this abnormality from medullary invasion is important for determining the extent of tumor resection.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Departments of the Army, Navy, or Defense.
Author contributions: Guarantors of integrity of entire study, all authors; study concepts and design, M.D.M.; literature research, M.D.M.; clinical studies, M.D.M., J.S.J., H.T.T., D.J.F.; data acquisition, M.D.M., J.S.J., H.T.T., F.H.G.; data analysis/interpretation, M.D.M.; statistical analysis, M.D.M., J.S.G., H.T.T.; manuscript preparation, revision/review, and final version approval, M.D.M.; manuscript definition of intellectual content, M.D.M., J.S.J.; manuscript editing, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
