MRI Findings of Giant Cell Tumors of the Spine

MRI of Spinal Tumors Musculoskeletal Imaging Clinical Observations Jong Won Kwon 1 Hye Won Chung 1,2 Eun Yoon Cho 3 Sung Hwan Hong 4 Sang-Hee Choi 1 Young Cheol Yoon 1 Sang Kyu Yi 1 Kwon JW, Chung HW, Cho EY, et al. Keywords: giant cell tumors, MRI, musculoskeletal imaging, spine DOI:10.2214/JR.06.1472 Received November 6, 2006; accepted after revision February 7, 2007. 1 Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul 135-710, Korea. 2 Present address: Department of Radiology, san Medical Center, University of Ulsan College of Medicine, 388-1, Pungnap-2 dong, Songpa-ku, Seoul 138-736, Korea. ddress correspondence to H. W. Chung (chung@amc.seoul.kr). 3 Department of Pathology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea. 4 Department of Radiology and Institute of Radiation Medicine, Seoul National University College of Medicine, Seoul, Korea. JR 2007; 189:246 250 0361 803X/07/1891 246 merican Roentgen Ray Society MRI Findings of Giant Cell Tumors of the Spine OJECTIVE. The purpose of this article is to describe the MRI features of giant cell tumors of the spine in 10 patients. CONCLUSION. One of the tumors was located in C7. The other nine tumors were located in the thoracic spine, lumbar spine, and sacrum, three in each site. The characteristic findings included an expansile mass with heterogeneous low to intermediate signal intensity on the T2-weighted images (10/10), a curvilinear area of signal void on T1- and T2-weighted images (9/10), and cystic changes within the mass (4/10). lthough no imaging feature was pathognomonic, MRI was found to be valuable in identifying the tumor, revealing its extent, and defining its relationship with the intraspinal structures. iant cell tumors involving the G spine are rare. In a literature review by Shankman et al. [1], only 2.7% of the 1,277 giant cell tumors reported were located in the spine. Sanjay et al. [2] reported that from 1955 to 1990 there were only 24 patients with giant cell tumors of the spine at the Mayo Clinic. The tumor usually develops in the vertebral body, and the posterior elements are frequently involved in patients with advanced lesions. lthough the features of giant cell tumors of the appendicular skeleton have been well described in the literature, there are few reports of MRI of vertebral tumors, which are limited to case reports and a few examples included in general reviews of spinal lesions [1 3]. In this study, the clinical records and MRI findings of 10 patients with giant cell tumors of the spine were reviewed retrospectively to define the typical MRI features of these unusual lesions. Materials and Methods Patient Population This study retrospectively analyzed the MR images of 10 patients diagnosed with giant cell tumor of the spine that was confirmed at surgical exploration and histology from 1999 to 2006. Eight patients were transferred from other hospitals along with their MR images for further evaluation of the tumors. ll patients were evaluated on MRI before biopsy or surgery. The clinical information available included age at presentation, sex, history, duration of signs and symptoms, and surgical or biopsy findings. The institutional review board waived approval for this retrospective study. Informed consent was not required. MRI Technique The imaging equipment and protocols varied according to the time and location at which the study was performed. The unenhanced fast spin-echo T1- weighted images (TR/TE, 422 875/9 23) and T2- weighted images (3,000 4,500/96 120) in the sagittal and axial planes were available for all patients. The images obtained after the IV administration of gadolinium were available in each case, and the imaging parameters used for enhanced imaging were similar to those used for unenhanced imaging. Image nalysis Two experienced musculoskeletal radiologists evaluated the imaging studies, and the results were determined by consensus. The lesion size was measured at the maximum diameter in centimeters. lso recorded were the vertebral segments involved and the number of vertebral segments spanned. The following aspects of the MR images were analyzed: the signal intensity on the T1- and T2- weighted images, enhancing pattern, presence or absence of curvilinear area of low signal intensity in the mass, tumor extent, presence or absence of spinal canal involvement, compression fracture of the involved vertebrae, cystic changes within the mass, and fluid fluid level within the tumor. Vertebral compression was defined as a loss of height of the vertebral body compared with the adjacent normal vertebrae. 246 JR:189, July 2007

MRI of Spinal Tumors TLE 1: Clinical Features and MRI Findings of Giant Cell Tumors of the Spine Patient No. ge (y) Sex Site Symptoms Duration of Symptoms (m) Curvilinear rea of Signal Void Fracture or Collapse Cystic Change Spinal Canal Involvement Expansile Pattern Maximum Diameter (cm) Cortical Destruction on CT 1 28 M S1 Radiating pain to right lower 8 + + + + 10.5 N extremity 2 31 F L5 Radiating pain to right lower 4 + + + + 6.0 N extremity 3 48 M S3 Sacral pain 3 + + + + 9.6 + 4 29 F L1 Lower back pain 12 + + + + + 8.6 + 5 19 M T2 Upper back pain 24 + 2.7 6 36 M T5 T7 No symptoms Incidental + + + + 5.6 + detection 7 14 F L5 Lower back pain with weakness 2 + + + + 6.2 + 8 27 M C7 Hoarseness 3 + + + + 7.9 + 9 33 M S1 Pain on right posterior thigh 1 + + + 6.4 N 10 54 M T8 Lower extremity weakness 5 + + + + 5.7 N Note Plus sign (+) = present, minus sign ( ) = absent, N = not applicable. Fig. 1 Giant cell tumor of L5 in 31-year-old woman (patient 2)., Sagittal T2-weighted fast spin-echo MR image (TR/TE, 4,000/102) shows tumor as heterogeneous hyperintensity and curvilinear low signal area (white arrows) within mass. There is decrease of vertebral height of L5. Spinal canal extension is indicated by open arrow., Contrast-enhanced axial T1-weighted MR image (592/14) shows moderate and heterogeneous enhancement. Tumoral mass involves vertebral body, right transverse process, pedicle, and articular facet with epidural and perivertebral extension. Curvilinear low signal area on T1- and T2-weighted images did not reveal any enhancement. On each pulse sequence, the signal intensity of the tumor was graded as low, similar, or high relative to the normal spinal cord. The signal pattern was evaluated as either homogeneous or heterogeneous. homogeneous signal pattern consisted of mainly uniform signal intensity throughout the lesion. heterogeneous signal pattern consisted of a mixture of signal intensities. On the gadoliniumenhanced images, each tumor was evaluated for the degree (graded as nonenhancement, mild, moderate, and marked) and pattern (homogeneous or heterogeneous) of enhancement. The CT images of the tumors, if available, were analyzed for the density of the tumor, the presence or absence of contrast enhancement, and the destruction of cortical bone. Results Clinical Features There were seven male and three female patients with an average age of 31.9 years (age range, 14 54 years) (Table 1). The patients symptoms were lower back, upper back, or sacral pain; weakness of the lower extremities; radiating pain to the lower extremities; and hoarseness. The symptom duration was 1 month to 3 years (mean, 6.2 months). One patient (patient 6) had no significant symptoms, and the tumor was detected incidentally on chest CT for evaluation of pulmonary tuberculosis. MRI Results Table 1 and Figures 1 3 summarize and illustrate the MRI features. One tumor was located in the C7 vertebral body, three in the thoracic spine, three in the lumbar spine, and three in the sacrum. The maximum measured diameter ranged from 2.7 to 10.5 cm (mean, 8.6 cm). The tumors located in the sacrum showed a larger size than the other tumors. T1-weighted MR images showed homogeneous and similar signal intensity to the normal spinal cord in eight cases and heterogeneous low signal intensity in two cases. The T2-weighted images showed heterogeneous signal intensity within the masses in all cases. Three cases showed high signal intensity on the T2-weighted images compared with the normal spinal cord, four cases showed similar signal intensity to the normal spinal cord, and three cases showed low intensity. Curvilinear low signal areas within the tumors were ob- JR:189, July 2007 247

C Fig. 2 Giant cell tumor of sacrum in 48-year-old man (patient 3)., Sagittal T2-weighted fast spin-echo MR image (TR/TE, 3,000/105) shows sacrococcygeal involvement of mass with anterior and posterior soft-tissue extension and heterogeneous high signal intensity., xial T2-weighted MR image (4,010/117) shows fluid fluid level (asterisk) in tumor. C, Sagittal T1-weighted MR image (613/13) shows homogeneous low signal intensity except curvilinear low signal area (arrowheads). D, Enhanced sagittal T1-weighted MR image (576/13) shows marked and heterogeneous enhancement and multiple focal cystic changes (arrows). (Fig. 2 continues on next page) served in nine cases on both the T1- and T2-weighted MR images (Fig.1). Cystic changes within the masses were observed in four cases and the fluid fluid level in only one case (Figs. 2 and 2). The tumors enhanced in a variety of patterns after the administration of gadolinium: mild in two, moderate in five, marked in three; and homogeneous in four and heterogeneous in six (Figs. 2C and 2D). Nine tumors D showed an expansile pattern that extended to the posterior column or paravertebral area (Fig. 1). One case in the T2 vertebral body (patient 5), which did not have a curvilinear area of low signal intensity within the mass, showed no expansile pattern. Eight tumors showed intraspinal extension. Only one case (patient 10) had a compressed spinal cord and showed high signal intensity in the spinal cord, which was indicative of compressive myelopathy. Compression fractures were observed in seven cases, which were located in the cervical (n = 1), thoracic (n =3), or lumbar (n = 3) spine. The sacral masses did not show any compression fracture. CT scans were available in six cases. On the CT scans, the masses were typically isodense with respect to the paraspinal muscle (Fig. 2E). Osteolytic tumors were well shown by CT in all six cases. Destruction of cortical bone was observed in five cases within the tumors (Figs. 3 and 3). Contrast-enhanced CT scans, which were available in three cases, showed enhancement of the lesion compared with the paraspinal muscle (Fig. 2F). Discussion Giant cell tumors of bone in the vertebrae are quite rare. Most of these lesions occur in the sacrum, followed in order of decreasing frequency by the thoracic, cervical, and lumbar segments [3]. Spinal lesions are more frequently found in women and affect patients in their second to fourth decades of life [4]. The clinical symptoms are primarily pain (often with a radicular distribution), weakness, and sensory deficits. dramatic increase in lesion size can occasionally be associated with pregnancy and is presumably related to hormonal stimulation [3]. However, in our series, only one case was detected in the cervical spine. More men were affected than women (M:F = 7:3). Pathologically, giant cell tumor consists of abundant osteoclastic giant cells intermixed throughout the spindle cell stroma. Mononuclear cells in a giant cell tumor are of more significance than the giant cells [5]. The other features of this lesion include occasional erythrocyte lakes (secondary aneurysmal bone cystlike change) and xanthomatous changes within the focal collections of histiocytes. neurysmal bone cyst formation may be encountered as a secondary feature in a variety of osseous lesions including giant cell tumor [6]. In addition, prominent areas of fibrous tissue with a high collagen level are a frequent finding, and giant cells are uncommon in these regions. 248 JR:189, July 2007

MRI of Spinal Tumors Fig. 2 (continued) Giant cell tumor of sacrum in 48- year-old man (patient 3). E, CT scan shows welldefined soft-tissue density with destruction of cortical bone. No calcification is seen within mass. F, Contrast-enhanced CT scan shows that mass enhances. G, Photomicrograph shows that tumor is composed of round-tooval mononuclear cells and multinucleated giant cells. Giant cells with varying numbers of nuclei are arranged more or less uniformly (arrows). (H and E, 100) The presence of a single marrow lesion in the axial skeleton on MR images poses a diagnostic problem of primary bone tumor or metastasis. lthough a biopsy of the lesion remains the mainstay of an ultimate diagnosis, the imaging characteristics of a lesion E F G might be helpful in directing the most appropriate further investigations. Radiologic studies of spinal giant cell tumors usually show an expansile lesion, with osteolysis observed on radiographs [7]. s with appendicular giant cell tumors, spinal lesions show no evidence of mineralized matrix [4]. In the sacrum, the lesions are frequently large with destruction of the sacral foraminal lines, but this nonspecific finding is also observed with other large sacral lytic lesions. sacral giant cell tumor commonly involves both sides of the midline, and an extension across the sacroiliac joint is frequent [8]. In this study, invasions to the sacroiliac joint were also detected in two cases of the three sacral masses. Paradoxically, giant cell tumors of the long bones do not share this invasive feature of spinal lesions and rarely extend across or through the articular cartilage. When a giant cell tumor occurs in the spine above the sacrum, it usually affects the vertebral body as opposed to the posterior elements and is accompanied by many other neoplasms (aneurysmal bone cyst, osteoid osteoma, or osteoblastoma). Extension into the posterior elements and paraspinal soft tissues and associated vertebral collapse are often apparent [7]. The MR images of a giant cell tumor often show heterogeneous signal intensity regardless of the pulse sequence used [9]. Generally, the tumor has low to intermediate signal intensity on the T1-weighted MR images. Interestingly, giant cell tumors have low to similar signal intensity to the normal spinal cord on the T2-weighted MR images in 63 96% of cases [9]. This appears to be caused by the relative collagen content of fibrous components and hemosiderin within the tumor [6]. When bone tumors do not contain a mineralized matrix, the low signal intensity areas observed on MR images are not always the result of hemosiderin deposition and can be caused by a dense collagen matrix [6]. lthough this feature is not unique to giant cell tumors of the spine, it is quite helpful in making a differential diagnosis because most other spinal neoplasms (metastases, myeloma, lymphoma, and chordoma) show high signal intensity on the long-tr MR images. This appearance may be the initial imaging characteristic suggestive of a correct diagnosis. Evidence of hemorrhage may also be apparent with high signal intensity on the T1- and T2- weighted images or fluid fluid levels on the MR images. Cystic areas (similar to those seen in aneurysmal bone cyst) and regions of previous hemorrhage with hemosiderin deposits are JR:189, July 2007 249

Fig. 3 Giant cell tumor of T5 T7 in 36-year-old man (patient 6)., Sagittal T2-weighted fast spin-echo MR image (TR/TE, 2,500/120) shows collapse of T6 vertebra and tumor spread to T5 and T7 vertebrae. Extradural intraspinal mass effect can be seen., Sagittal reconstructed CT scan shows destruction of cortical bone. common [4]. In this study, cystic changes were also noted in four cases, and a fluid fluid level was observed in one case. MRI performed after an IV injection of contrast material also shows enhancement of the lesion, which reflects its increased vascular supply. In the present case series, nine patients showed the unusual appearance of low signal curvilinear areas on the T1- and T2-weighted images within the vertebral body, which presumably corresponds to a multicystic lesion with a thickened trabeculae, fibrous septae, or hemosiderin deposit. The curvilinear area of low signal intensity was described previously for plasmacytoma and vertebral hemangioma [10]. solitary bone plasmacytoma needs to be considered in a differential diagnosis. However, heterogeneous signal intensity on the T2-weighted images, heterogeneous enhancement, and cystic changes were not usually observed in vertebral hemangioma. Disease control after a complete surgical excision for a giant cell tumor is 85 90% successful in the tubular bones. However, giant cell tumors in the axial skeleton, particularly in the sacrum, are associated with a poorer prognosis because the tumors are larger and more difficult to excise completely [11]. In many cases, a complete resection is only possible in limited sacral lesions, and imaging plays a key role in a presurgical evaluation. In general, sacral lesions that spare the majority of the S1 segment and the sacroiliac joint are amenable to a complete excision. Giant cell tumors that cannot be excised entirely are often treated with a combination of partial curettage and radiation therapy [12]. The estimated frequency of malignant giant cell tumors or a malignant transformation of benign giant cell tumors is 10% [12]. In most cases, the malignant transformation develops after irradiation. Radiation therapy, though controversial due to the potential risk of causing the sarcomatous degeneration of benign tumors, may provide a reasonable method for controlling the tumors in these cases [12]. In conclusion, the MRI findings of a giant cell tumor in the spine are often characteristic, allowing for an accurate diagnosis and preoperative evaluation of the extent of the mass. The characteristic findings include an expansile mass with heterogeneous low to intermediate signal intensity on the T2- weighted image, curvilinear area of low signal intensity on the T1- and T2-weighted images, and cystic changes within the mass. References 1. Shankman S, Greenspan, Klein MJ, Lewis MM. Giant cell tumor of the ischium: a report of two cases and review of the literature. Skeletal Radiolo 1988; 17:46 51 2. Sanjay K, Sim FH, Unni KK, McLeod R, Klassen R. Giant-cell tumors of the spine. J one Joint Surg r 1993; 75:148 154 3. idwell JK, Young JW, Khalluff E. Giant cell tumor of the spine: computed tomography appearance and review of the literature. J Comput Tomogr 1987; 11:307 311 4. Murphey MD, ndrews CL, Flemming DJ, Temple HT, Smith WS, Smirniotopoulos JG. From the archives of the FIP: primary tumors of the spine radiologic pathologic correlation. RadioGraphics 1996;16:1131 1158 5. Huvos G. one tumors: diagnosis, treatment, and prognosis, 2nd ed. Philadelphia, P: W Saunders, 1991:429 467 6. oki J, Tanikawa H, Ishii K, et al. MRI findings indicative of hemosiderin in giant cell tumor of bone: frequency, cause, and diagnostic significance. JR 1996;166:145 148 7. Kumar R, Guinto FC Jr, Madewell JE, David R, Shirkhoda. Expansile bone lesions of the vertebra. RadioGraphics 1988; 8:749 769 8. atnitzky S, Soye I, Levine E, Price HI, Hart KZ. CT in the evaluation of lesions arising in and around the sacrum. RadioGraphics 1982; 2:500 528 9. Resnick D. Diagnosis of bone and joint disorders, 3rd ed. Philadelphia, P: Saunders, 1995:3785 3806 10. Shah K, Saifuddin, Price GJ. MRI of spinal plasmacytoma. Clin Radiol 2000; 55:439 445 11. Randall RL. Giant cell tumor of the sacrum. Neurosurg Focus 2003; 15:E13 12. Seider MJ, Rich T, yala G, Murray J. Giant cell tumors of bone: treatment with radiation therapy. Radiology 1986; 161:537 540 250 JR:189, July 2007