Incidental Enchondromas of the Knee

Musculoskeletal Imaging Original Research Musculoskeletal Imaging Original Research Michael J. Walden 1 Mark D. Murphey 1,2,3 Jorge. Vidal 1,2 Walden MJ, Murphey MD, Vidal J Keywords: appendicular musculoskeletal system, biomedical statistics, enchondromas, MRI, radiologic pathologic correlation DOI:10.2214/JR.07.2796 Received June 29, 2007; accepted after revision December 19, 2007. 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 Department of the rmy, the Department of the Navy, or the Department of Defense. 1 Department of Radiology, Walter Reed rmy Medical Center, Washington, DC 20306. 2 Department of Radiologic Pathology, rmed Forces Institute of Pathology, 6825 16th St. NW, ldg. 54, Washington, DC 20306. ddress correspondence to M. D. Murphey (murphey@afip.osd.mil). 3 Department of Radiology, Uniformed Services University of the Health Sciences, ethesda, MD. CME This article is available for CME credit. See www.arrs.org for more information. JR 2008; 190:1611 1615 0361 803X/08/1906 1611 merican Roentgen Ray Society Incidental Enchondromas of the Knee OJECTIVE. The purpose of our study was to determine the prevalence of incidental enchondromas on routine MR knee imaging. MTERILS ND METHODS. We retrospectively reviewed 449 consecutive routine knee MR examinations for the presence of enchondromas. MRI was considered positive when a focal geographic area of lobular marrow replacement (nonsubchondral) was identified on T1 weighting and high signal intensity was seen on T2 weighting. Patients with enchondromas were further evaluated for demographics; lesion site, size, and relationship to the physeal plate; aggressive imaging features described with chondrosarcoma; concurrent internal derangement; and study indication. RESULTS. The prevalence of incidental enchondromas was 2.9% on routine knee MR examinations. The prevalence was highest in the distal femur (2.0%), followed by the proximal tibia (0.7%) and the proximal fibula (0.2%). The average lesion size was 1.9 1.2 1.3 cm (57% of lesions were < 1 cm). Most lesions were located in the metaphysis (71%) or diaphysis (21%). Enchondromas were within 1.5 cm of the physeal plate in 72% of cases. No aggressive imaging features to suggest chondrosarcoma were seen. ll patients had evidence of internal derangement as the cause of symptoms and the request for imaging. CONCLUSION. Incidental enchondromas can be identified on 2.9% of routine MR knee examinations, most frequently in the distal femur (2.0%). This significant prevalence is much higher than in an autopsy series (0.2%), likely reflecting the increased sensitivity of MRI for detecting small lesions, and is important to recognize to avoid confusion with other pathologic entities. E nchondroma is a benign neoplasm of the medullary canal com posed of mature hyaline cartilage. Enchondroma is one of the most common osseous neoplasms, representing 12 24% of all benign bone tumors and 3 10% of all bone tumors [1, 2]. Of benign chondroid lesions, it is second only to osteochondroma in frequency [1, 2]. Intramedullary chondrosarcoma is also relatively common, accounting for 20 27% of all primary bone sarcomas and 8 17% of all bone tumors [3 6]. Enchondroma and intramedullary chondrosarcoma can be difficult to distinguish on imaging, although certain features, including the lesion size, degree of endosteal scalloping, a rim of surrounding edema, and pain referable to the lesion, favor the diagnosis of intramedullary chondrosarcoma [7 13]. In our anecdotal experience, incidental enchondromas are commonly seen on routine MRI of the knee. ecause of the rela- tively high frequency of enchondromas and their similarity to intramedullary chondrosarcomas, we believe it is important to establish the prevalence of incidental enchondroma. lthough multiple studies have described the prevalence of these lesions among bone tumors [1, 2, 14 16], to our knowledge the literature describing the overall incidental prevalence of these lesions (0.2%) is limited to a single autopsy series [17]. It is therefore our purpose to establish the frequency of enchondroma about the knee on routine MR examinations. Materials and Methods Two radiologists retrospectively reviewed 463 consecutive routine knee MR examinations from October 2004 to February 2005 for identification of enchondromas, with agreement by consensus. Fourteen of the examinations were eliminated as a result of extensive metallic artifact, leaving a total study group of 449 MR examinations (434 pat ients, JR:190, June 2008 1611

in 15 of whom bilateral MR knee exami nations were evaluated). This study was approved by the Department of Clinical Investigations of the Walter Reed rmy Medical Center in compli ance with HIP. Informed consent was not required. MRI examinations were performed on four different imaging units that operated at high field strength (1.5 T). Images available for review were from three institutions and included T1-weighted (TR range/te range, 350 700/13 40) spin-echo or fast spin-echo, T2-weighted (3,650 4,350/ 80 110) spin-echo or fast spin-echo, gradientrecalled echo (400 650/10 15), and proton density weighted with fat-saturation (1,500 4,850/13 50) images in the axial, coronal, and sagittal oblique planes. The field of view of all images ranged from 140 to 160 mm. Slice interval was 3.5 or 4 mm on all images. MRI was considered positive for enchondroma on identification of a focal geographic (masslike) area of marrow replacement on T1-weighted images and corresponding high signal intensity (similar to fluid) on T2-weighted images with lobular margins on either pulse sequence. Lesions with similar characteristics that were subchondral in location or had associated overlying hyaline cartilaginous defects were excluded as representing subchondral cysts, intraosseous ganglia, or subchondral edema or contusion. Patients with lesions that met these imaging criteria were further evaluated for lesion site (femur, tibia, or fibula), patient demographics (sex and age and compared with the total patient population), lesion size (in three dimensions and distinction of lesions into two groups as those less than or greater than 1 cm in maximal dimension), longitudinal and axial location in the marrow (epiphyseal, metaphyseal, or diaphyseal; and cen tral or eccentric), relationship of the lesion to the physeal plate (abutting the physeal plate, not adjacent to but within 1.5 cm of the physeal plate, or > 1.5 cm from the physeal plate [and average lesion size for enchondromas within 1.5 cm of the plate vs farther than 1.5 cm]), evidence of endo steal scalloping (greater than or less than two thirds of the normal cortical thickness, if present), presence of an associated soft-tissue component, presence or absence of surrounding marrow edema, and concurrent findings of internal de rangement (e.g., meniscal tears, cruciate liga ment injuries, hyaline cartilage defects, or colla teral ligament injuries). In studies positive for the presence of enchondroma, the indication for ordering each examination was reviewed to deter mine whether the primary suspicion was internal derangement of the knee or whether a primary osseous lesion was suspected. search for other imaging techniques such as radiography, bone scintigraphy, and CT was made in all cases in which an enchondroma was identified on MRI for purposes of correlation. Only three patients had correlative images, all of which were radiographs. These were evaluated for the following features: the presence or absence of a mixed lytic and sclerotic lesion, presence or absence of matrix mineralization showing a typical chondroid arc and ring appearance as the cause of the lesion sclerosis, lesion size, and endosteal scalloping less than or greater than two thirds of the normal cortical thickness (if present). Results We found 14 (3.1%) enchondromas about the knee in 13 (2.9%) patients (one patient had two enchondromas, both in the distal femoral diaphysis). Nine cases were in the femur (2.0%) (Figs. 1 4), three in the tibia (0.7%) (Fig. 5), and one in the fibula (0.2%). Of the 13 patients with lesions, seven were male from 254 males in the total patient population. Six female patients had lesions of 180 females in the total sample population. Fig. 1 Incidental enchondroma in distal femur of 51-year-old woman with medial meniscal tear as cause of clinical symptoms. and, Sagittal T1-weighted (, TR/TE, 367/20) and fat-suppressed proton density weighted (, 2,250/13) MR images show focal area of marrow replacement with high signal intensity on long-tr image (arrows) in distal femoral metaphysis. Lesion is centrally located in medullary canal, has mildly lobulated borders, and is juxtaposed to old epiphyseal plate (arrowheads, ). Fig. 2 Incidental enchondromas in distal femoral diaphysis of 24-year-old woman with anterior knee pain caused by patellofemoral disease., Sagittal scout gradient-echo (TR/TE, 92/1.6; flip angle, 30 ) MR image shows two foci of marrow replacement in distal femoral diaphysis (arrows)., xial fat-suppressed T2-weighted (4,650/36) MR image through more inferior lesion reveals high signal intensity in eccentric intramedullary lesion and mildly lobulated margins (arrowsheads). No endosteal scalloping is present. Patient s symptoms and cause for MRI was patellofemoral disease (not shown). 1612 JR:190, June 2008

Fig. 3 Minute enchondroma or cartilage rest in distal femur of 33-year-old man with knee pain associated with quadriceps and patellar tendon tendinopathy. and, Coronal T1-weighted (, TR/TE, 400/18) and fat-suppressed T2-weighted (, 4,417/41) MR images show minute focal area of marrow replacement with prominent high signal intensity on long-tr image in distal femoral metadiaphysis (arrows). Lesion reveals minimally lobulated margins on T2-weighted image and is more than 1.5 cm from physeal scar (arrowheads). Cause for MRI was quadriceps and patellar tendon tendinopathy (not shown). C D ge varied from 10 to 83 years (average, 40.1 years) in the total sample population, whereas the age of patients with lesions varied from 24 to 53 years (average, 36.4 years). Size range of enchondromas varied from 0.3 0.3 0.4 cm to 4.0 2.2 2.6 cm (average, 1.9 1.2 1.3 cm). The maximal dimension of the lesions was less than 1 cm in 57% (n = 8) and larger than 1 cm in 43% (n = 6). The lesions were located in the metaphysis in 71% (n = 10) (Figs. 1, 4, and 5) of cases, in the epiphysis in 7% (n = 1), and in the diaphysis in 21% (n = 3) (Fig. 2). The lesions were located centrally in the medullary canal in 57% (n = 8) and eccentrically in 43% (n = 6). The lesions abutted the physeal plate in 43% (n = 6) of cases (Figs. 1, 4, and 5) and were within 1.5 cm of the physeal plate in 29% (n = 4) of lesions. The remaining 29% of lesions (n = 4) were more than 1.5 cm from the physeal plate (Figs. 2 and 3). The average lesion size of enchondromas less than 1.5 cm from or abutting the physeal plate was 1.2 0.8 0.9 cm. The average lesion size of enchondromas more than 1.5 cm from the epiphyseal plate was 1.2 0.9 1.0 cm. None of the lesions had evidence of endosteal scalloping, a soft-tissue component, or surrounding marrow edema. ll patients with enchondromas had accompanying abnormalities of internal derangement identified on MRI that corresponded to the patient s complaints and were the cause of the imaging being performed. Fig. 4 Incidental enchondroma of distal femur and small subchondral cyst or intraosseous ganglion in 53-year-old woman with knee pain associated with medial and lateral meniscal tears. and, Coronal T1-weighted (, TR/TE, 550/15) and fat-suppressed T2-weighted (, 4,867/50) MR images reveal proximal tibial epiphyseal lesion with marrow replacement and high signal intensity on long-tr image (arrows). These intrinsic MR characteristics simulate enchondroma. C and D, Sagittal T1-weighted (C, 550/15) and gradient-echo (D, 577/10; flip angle, 30 ) MR images reveal cleftlike extension to subchondral bone (arrowheads) that shows high signal intensity on long-tr and gradient-echo images. This feature is not seen in enchondromas and suggests correct diagnosis of subchondral cyst or intraosseous ganglion. Note distal femoral enchondroma (arrows) with typical features on sagittal images. JR:190, June 2008 1613

Fig. 5 Incidental enchondroma in proximal tibia of 32-year-old woman with meniscal tear. and, Coronal T1-weighted (, TR/TE, 500/15) and sagittal fat-suppressed T2-weighted (, 4,000/103) MR images show small focal area of marrow replacement (arrows). Lesion reveals high signal on T2 weighting, has mildly lobular borders, is juxtaposed to epiphyseal plate scar (arrowheads, ), and is slightly eccentrically located in medullary canal. Radiographs were available for correlation in three cases. In one patient, a mixed lytic and sclerotic lesion with typical arc and ring chondroid matrix mineralization (causing the lesion sclerosis) was seen and measured 4.0 2.2 2.6 cm. In the other two patients, the radiographs were normal and no lesion could be identified corresponding to the MR abnormality. No evidence of endosteal scalloping was seen in any of these three lesions at radiography. Discussion Our study confirms the significant prevalence of enchondroma affecting the osseous structures about the knee. These lesions were identified on 2.9% of routine MR knee examinations. The significant frequency of these lesions about the knee should be recognized in order to avoid confusion with or misinterpretation of other pathologic entities. The most frequently encountered site of incidental enchondroma was the distal femur (2.0%, n = 9), followed by the proximal tibia (0.7%, n = 3), and the proximal fibula (0.2%, n = 1). The distal femoral prevalence of 2.0% is 10 times higher than that in an autopsy series that showed only two enchondromas in 1,125 right distal femurs specimens (0.2%) of patients older than 25 years [17]. In this autopsy series, the overall prevalence of incidental enchondromas in the femur was 1.8%, with most lesions described as proximal between the femoral head and neck [17]. The reason for the higher prevalence of enchondromas on imaging as compared with the autopsy series is likely related to the increased sensitivity of MRI in identifying small lesions that would not be apparent at typical gross pathologic sectioning thickness and intervals [17]. Only those lesions visible to the naked eye on the gross pathologic sections underwent further histologic analysis in the autopsy series [17]. This concept is supported by the small size of many of our lesions; 57% of the lesions were smaller than 1 cm in maximal dimension. We believed that most of the lesions would be closely related to the physeal plate. This was based on the theory, with which we agree, that enchondromas originate as cartilage rests derived from the physeal plate [8, 17]. Indeed, 71% (n = 10) of enchondromas were within 1.5 cm of the physeal plate and 43% (n = 6) abutted the physeal plate. Only 29% of enchondromas were more than 1.5 cm from the physeal plate. In addition, we believed that lesions of increased distance from the physeal plate might be larger because they arose earlier and had more time available for potential growth. However, we found no significant difference in the average size of enchondromas less than 1.5 cm from or abutting the physeal plate (1.2 0.8 0.9 cm) compared with those more than 1.5 cm from the physeal plate (1.2 0.9 1.0 cm). This finding probably reflects the indolent natural history of these lesions, most of which have a limited potential for growth. None of the lesions in our study had imaging or clinical features that could have been described as suggestive of intramedullary chondrosarcoma. The average maximal diameter of enchondromas in our study was 1.9 cm, a figure that agrees with previous studies, indicating that size is a significant discriminator between enchondromas and intramedullary chondrosarcomas [7, 8, 13]. In particular, lesions greater than 4.0 cm were more likely to be intramedullary chondrosarcomas in the study by Kendell et al. [13]. The small size of the lesions is also the probable reason that only one of the three enchondromas was visible on radiographic evaluation. In our series, 57% (n = 8) of lesions were smaller than 1 cm in maximal dimension; in dictating our reports, we refer to lesions of this size as cartilaginous rests. In our opinion, this terminology emphasizes the indolent nature of these lesions. There is also limited or no value in recommending correlation with radiographs in these lesions. We reserve the designation of enchondroma for lesions larger than 1 cm in maximum dimension, which constituted 43% (n = 6) of the lesions in our study. No lesions were seen to have peritumoral marrow edema, another imaging finding that suggested chondrosarcoma in the series by Janzen et al. [12]. Deep endosteal scalloping greater than two thirds of the normal cortical thickness has also been described as an imaging feature that allows distinction of enchondroma from intramedullary chondrosarcoma in long bone lesions [7, 8]. Only one of the lesions in our study was adjacent to the endosteum, and no significant endosteal scalloping was seen in this patient. Pain referable to the lesion is a clinical feature that is helpful to suggest the diagnosis of intramedullary chondrosarcoma as opposed to enchondroma of long bones, which is frequently asymptomatic [7, 8, 11]. ll patients with lesions in our series had concurrent abnormal findings on MRI that corresponded to the indication for imaging. We therefore presumed that each lesion was asymptomatic and incidentally detected [7, 8, 11]. None of the patients with enchondromas was imaged for evaluation of an osseous lesion. Some limitations of this study include its retrospective nature and the lack of uniformity in imaging parameters because of the multiplicity of imaging centers. In addition, the diagnosis of the presumed enchondroma was not confirmed pathologically, although such confirmation is naturally limited to the realm of autopsy series. However, we were stringent in our criteria for designating a lesion an enchondroma (Fig. 5). Only medullary lesions with lobular borders, geographic 1614 JR:190, June 2008

(masslike) marrow replacement, and typical intrinsic characteristics were included (Fig. 5). Lesions in a subchondral location or with overlying hyaline cartilaginous defects were excluded. We believe this eliminated inclusion of any intraosseous contusions, subchondral cysts, intraosseous ganglia, or subchondral edema that could show similar intrinsic characteristics. Despite these limitations, we believe our results accurately represent the prevalence of these lesions. In conclusion, we found that incidental enchondromas can be identified in 2.9% of routine MR knee examinations. They are most frequently encountered in the distal femur (2.0%), followed by the proximal tibia (0.7%) and the proximal fibula (0.2%). The prevalence in the distal femur is much higher than that seen in an autopsy series (0.2%), which is likely related to the increased sensitivity of MRI in identifying small lesions. These lesions show no aggressive features that have been associated with intramedullary chondrosarcoma. In our opinion these lesions do not require imaging follow-up, although radiographs of larger enchondromas may serve as a useful inexpensive baseline examination unless new symptoms related to the lesion become clinically apparent. The significant frequency of these lesions about the knee is important to recognize in order to avoid confusion with other pathologic entities. FOR YOUR INFORMTION References 1. Dorfman DD, ogdan C. one tumors. St. Louis, MO: Mosby, 1998:253 276 2. Unni KK, Inwards CY, ridge J, et al. FIP atlas of tumor pathology series 4: tumors of the bones and joints. Silver Spring, MD: RP Press, 2005:46 52 3. Mulder JD, Kroon HM, Schutte HE, Taconis WK, eds. Radiologic atlas of bone tumors. msterdam, The Netherlands: Elsevier, 1993:7 421 4. Unni KK. Chondroma. In: Unni KK, ed. Dahlin s bone tumors: general aspects and data on 11,087 cases, 5th ed. Philadelphia, P: Lippincott-Raven, 1996:25 45 5. Mirra JM. Intramedullary cartilage and chondroidproducing tumors. In Mirra JM, ed. one tumors: clinical, radiologic, and pathologic correlations. Philadelpia, P: Lea & Febiger, 1989: 439 535 6. Campanacci M. one and soft tissue tumors. New York, NY: Springer-Verlag, 1990 7. Murphey MD, Flemming DJ, oyea SR, et al. Enchondroma versus chondrosarcoma in the appendicular skeleton: differentiating features. Radio- Graphics 1998; 18:1213 1237 8. Flemming DJ, Murphey MD. Enchondroma and chondrosarcoma. Semin Musculoskelet Radiol 2000; 4:5 71 9. rien EW, Mirra JM, Kerr R. enign and malignant cartilage tumors of bone and joint: their anatomic and theoretical basis with an emphasis on radiology, pathology and clinical biology. Skeletal Radiol 1997; 26:325 353 10. Robinson P, White LM, Sundaram M, et al. Pe- This article is available for CME credit. See www.arrs.org for more information. riosteal chondroid tumors: radiologic evaluation with pathologic correlation. JR 2001; 177:1183 1188 11. Weiner SD. Enchondroma and chondrosarcoma of bone: clinical, radiologic, and histologic differentiation. Instr Course Lect 2004; 53:645 649 12. Janzen L, Logan PM, O Connell JX, et al. Intramedullary chondroid tumors of bone: correlation of abnormal peritumoral marrow and softtissue MRI signal with tumor type. Skeletal Radiol 1997; 26:100 106 13. Kendell SD, Collins MS, dkins MC, Sundaram M, Unni KK. Radiographic differentiation of enchondroma from low-grade chondrosarcoma in the fibula. Skeletal Radiol 2004; 33:458 466 14. arbosa CS, raujo, Mirando D. Incidence of primary benign and malignant neoplasms and bone pseudotumoral lesions: an epidemiologic analysis of 585 cases diagnosed at the Faculdade de Medicina of the Universidade Federal de Minas Gerais [in Portuguese]. M Rev ssoc Med ras 1991; 37:187 192 15. Pongkripetch M, Sirikulchayanonta V. nalysis of bone tumors in Ramathibodi Hospital, Thailand during 1977 1986: study of 652 cases. J Med ssoc Thai 1989; 72:621 628 16. Valdespino-Gómez VM, Cintra-McGlone E, Fiqueroa-eltrán M. one tumors: their prevalence [in Spanish]. Gac Med Mex 1990; 126:325 334 17. Scherer E. Exostosen, enchondrome und ihre beziehung zum periost, Frankfurt. Ztschr F Path 1928; 36:587 605 JR:190, June 2008 1615