Takayuki Ohguri 1 Takatoshi Aoki 1 Masanori Hisaoka 2 Hideyuki Watanabe 1 Katsumi Nakamura 1 Hiroshi Hashimoto 2 Toshitaka Nakamura 3 Hajime Nakata 1

Transcription

1 Takayuki Ohguri 1 Takatoshi Aoki 1 Masanori Hisaoka 2 Hideyuki Watanabe 1 Katsumi Nakamura 1 Hiroshi Hashimoto 2 Toshitaka Nakamura 3 Hajime Nakata 1 Received July 1, 2002; accepted after revision November 12, Department of Radiology, University of Occupational and Environmental Health, Iseigaoka 1-1, Yahatanisi-ku, Kitakyushu-shi , Japan. Address correspondence to T. Ohguri. 2 Department of Pathology and Oncology, University of Occupational and Environmental Health, Yahatanisi-ku, Kitakyushu-shi , Japan. 3 Department of Orthopedic Surgery, University of Occupational and Environmental Health, Yahatanisi-ku, Kitakyushu-shi , Japan. AJR 2003;180: X/03/ American Roentgen Ray Society Differential Diagnosis of Benign Peripheral Lipoma from Well-Differentiated Liposarcoma on MR Imaging: Is Comparison of Margins and Internal Characteristics Useful? OBJECTIVE. Our objective was to evaluate the reliability of MR imaging in distinguishing between benign lipoma and well-differentiated liposarcoma. MATERIALS AND METHODS. The MR images of 35 pathologically proven benign lipomas in 35 patients and 23 well-differentiated liposarcomas in 17 patients were retrospectively reviewed. T1-, T2-, and fat-suppressed T1-weighted images were obtained after administration of gadopentetate dimeglumine. Margins and internal characteristics revealed on the MR images and the degree of contrast enhancement of septa were evaluated. These MR imaging findings were compared for well-differentiated liposarcomas and benign lipomas. RESULTS. Completely irregular margins were recognized only in benign lipomas with a pathologic diagnosis of infiltrating lipoma. All tumors without a recognizable nonadipose component were benign lipomas (p < 0.05). As for the well-differentiated liposarcomas, thick septa and nodular or patchy nonadipose components were present more frequently in deep and retroperitoneal lesions than in subcutaneous lesions (p < 0.01). No cases showed only thin septa in the deep lesions of well-differentiated liposarcoma, and all cases showed thick septa or nodular or patchy nonadipose components. The septa in well-differentiated liposarcomas enhanced more strongly than did those in benign lipomas. The septa showed no enhancement relative to muscle in 11 of 19 benign lipomas, whereas the septa showed moderate or marked enhancement in all well-differentiated liposarcomas (p < 0.01). CONCLUSION. Careful assessment of margins and internal characteristics on MR imaging can be a useful aid in further distinguishing between biologically different benign lipoma and well-differentiated liposarcoma. W ell-differentiated liposarcomas of the soft tissue closely mimic benign lipomas, although both tumors are different in biologic nature. Well-differentiated liposarcomas often recur and must be treated with extensive excision [1 3]. Many studies have reported the MR findings of benign lipomas and liposarcomas [4 11]. Both well-differentiated liposarcomas and lipomas have been reported to show MR signal intensity equal to that of fat. In addition, well-differentiated liposarcomas of the extremities can have septa and minor nodular components consisting of nonadipose tissue on MR imaging [6]. Benign lipomas occasionally contain other mesenchymal elements. The most common of these is fibrous connective tissue that may appear as septa showing linear areas of decreased signal intensity on MR imaging, regardless of the pulse sequence [4]. To our knowledge, only a few reports have compared the nonadipose tissues of these tumors on MR imaging [8, 11], and the differentiation between benign lipomas and well-differentiated liposarcomas has not been sufficiently discussed. In our study, we retrospectively reviewed the MR findings of benign lipomas and well-differentiated liposarcomas to evaluate the possibility of their differentiation. Materials and Methods We retrospectively reviewed the records of 35 patients with 35 benign lipomas (17 men and 18 women; age range, years; average age, 55 years) and 17 patients with 23 well-differentiated liposarcomas (11 men and six women; age range, years; average age, 62.7 years) from 1991 to All patients underwent preoperative MR imaging and had their tumors excised at our institution. The benign lipomas were smaller than 3 cm (n = 7), 3 5 cm (n = 14), 6 10 cm (n = 12), and AJR:180, June

2 Ohguri et al. larger than 10 cm (n = 2); and the well-differentiated liposarcomas were smaller than 3 cm (n = 0), 3 5 cm (n = 4), 6 10 cm (n = 7), and larger than 10 cm (n = 12). The locations of the benign lipomas were subcutaneous (n = 19) and deep somatic (n = 16); and the locations of the well-differentiated liposarcomas were subcutaneous (n = 9), deep somatic (n = 9), and retroperitoneal (n = 5). The histopathologic diagnoses of the benign lipomas were lipomas (n = 26), infiltrating lipomas (n = 4), parosteal lipomas (n = 2), fibrolipomas (n = 2), and spindle cell lipoma (n = 1). The histopathologic subtypes of the well-differentiated liposarcomas were lipomalike (n = 19) and sclerosing (n = 4). All surgical specimens of the tumors were reviewed and diagnosed by two pathologists who were experienced in the diagnosis of bone and soft-tissue tumors. MR imaging was performed with a 1.5-T superconductive unit (VISART, Toshiba Medial Systems, Tokyo, Japan; Signa, General Electric Medial Systems, Milwaukee, WI). Slice thickness varied from 4 to 10 mm, matrix size was , and pulse sequences were spin-echo T1-weighted images (TR range/te range, /15 20) and T2- weighted images ( /80 120). Fat-suppressed T1-weighted imaging was performed after administration of IV gadopentetate dimeglumine (Magnevist, Schering, Berlin, Germany) (0.1 mmol/ kg of body weight) on 21 cases of benign lipoma and 10 cases of well-differentiated liposarcoma. The MR images were reviewed by two radiologists who were unaware of the histopathologic diagnoses. The final assessment was reached by consensus. Margins, internal structures, and degrees of enhancement of septa after IV administration of gadopentetate dimeglumine were analyzed and classified as follows. Margins were defined as well-defined and smooth, partially irregular, and completely irregular. Lesions with well-defined and smooth margins were further divided into a uninodular or a multinodular mass independent of shape. Internal structures were classified into five types on the basis of MR signal intensity of the nonadipose components: type I, nonadipose component unrecognizable; type II, only thin septa ( 2 mm) with low signal intensity detectable; type III, one or two thick septa (> 2 mm) with low signal intensity detectable; type IV, three or more thick septa detectable; and type V, nodular or patchy nonadipose component detectable (Fig. 1). The relationship between the location of tumors and the internal characteristics on MR imaging was also evaluated. Because tumors with a completely irregular margin that prominently infiltrated the surrounding muscle at any point could not be classified into any one of these five types, they were excluded from this evaluation. All such lesions were infiltrating lipomas. The degree of enhancement of the septa after administration of gadopentetate dimeglumine was determined relative to muscle: decreased enhancement was considered as no enhancement, similar to moderate enhancement; and increased enhancement was considered as marked enhancement. Type I Type II Type III Type IV Type V For all cases, both macroscopic and microscopic histologic findings of surgical specimens were available and were examined in comparison with the previously mentioned MR findings. The Fisher s exact test was used for statistical analysis to determine whether the differences between lipomas and liposarcomas in terms of margins, internal structures (as evaluated on the basis of nonfatty components), and enhancement of septa on fat-suppressed T1-weighted images after administration of gadopentetate dimeglumine were statistically significant. The Fisher s exact test was also used to determine whether the internal structures of well-differentiated liposarcomas differed significantly on the basis of location (i.e., subcutaneous, deep, and retroperitoneal lesions). TABLE 1 a Fisher s exact test. Characteristics of Tumor Margins Results Fig. 1. Diagram shows five types of tumor based on appearance of nonadipose components on MR imaging: type I, nonadipose component unrecognizable; type II, only thin septa ( 2 mm) with low signal intensity detectable; type III, one or two thick septa (> 2 mm) with low signal intensity detectable; type IV, three or more thick septa detectable; and type V, nodular or patchy nonadipose component detectable. The margins and internal characteristics of benign lipoma and well-differentiated liposarcoma are summarized in Tables 1 and 2. The margins were mostly well defined and smooth in both tumor types. Of the 20 well-differentiated liposarcomas with well-defined and smooth margins, only six were uninodular, and the remaining 14 were multinodular. Twentyone of 30 benign lipomas were uninodular, and the remaining nine were multinodular. The differences in both uninodular and multinodular margins between benign lipomas and well- Tumor Margin Well-Differentiated p a Benign Lipoma (n = 35) Liposarcoma (n = 23) Well-defined and smooth Uninodular 21 (60) 6 (26) < 0.05 Multinodular 9 (25) 14 (61) < 0.05 Partially irregular 2 (6) 3 (13) 0.38 Completely irregular 3 (9) 0.27 TABLE 2 Tumor Type Based on Appearance of Nonfatty Components of Internal Structures on MR Imaging Tumor Type p b Benign Lipoma (n = 32) a Well-Differentiated Liposarcoma (n = 23) Type I 7 (22) < 0.05 Type II 15 (47) 2 (9) < 0.01 Type III 7 (22) 6 (26) 0.76 Type IV 8 (35) < Type V 3 (9) 7 (30) 0.08 a Three infiltrating lipomas that showed prominent infiltrative margin in surrounding muscle tissue were excluded. b Fisher s exact test AJR:180, June 2003

3 MR Imaging of Benign Lipoma and Well-Differentiated Liposarcoma Fig year-old woman with infiltrating lipoma of thigh. Margin of tumor is completely irregular because of neoplastic fatty tissue infiltrating surrounding muscle tissue as shown on T1-weighted sagittal MR image. Fig year-old man with benign lipoma of shoulder (type I, nonadipose component unrecognizable). T1-weighted MR image shows uninodular tumor with well-defined and smooth margin (arrow). No recognizable nonadipose tissue is present. Fig year-old man with benign subcutaneous lipoma of shoulder (type II, only thin septa [ 2 mm] with low signal intensity detectable). T1-weighted MR image shows uninodular hyperintense tumor (arrow) with well-defined and smooth margin and thin septa. differentiated liposarcomas were statistically significant (Fisher s exact test, p < 0.05). Three well-differentiated liposarcomas and two benign lipomas showed partially irregular margins. Completely irregular margins were recognized only in benign lipomas with a pathologic diagnosis of infiltrating lipoma (Fig. 2). In three of four intramuscular (infiltrating) lipomas, the margins were entirely irregular at all points where the neoplastic fatty tissue infiltrated and intermingled with the surrounding muscle tissue. In the remaining intramuscular lipoma, the margin was partially irregular. The assessment of internal characteristics showed that all the type I tumors were benign lipomas (Fig. 3). Most type II tumors were benign lipomas, but two were well-differentiated liposarcomas (Fig. 4). The differences were significant ( p < 0.05, p < 0.01, respectively). Seven benign lipomas and six well-differentiated liposarcomas were found to be type III; there was no statistical significance for type III (Fig. 5). All the type IV tumors were well-differentiated liposarcomas ( p < 0.001) (Fig. 6). Type V tumors were found in both lipomas and liposarcomas. Of the three type V lipomas, two were parosteal lipomas and one was a spindle cell lipoma. The MR images of these parosteal lipomas clearly showed the lipomatous and osteochondromatous components. Of the well-differentiated liposarcomas, two deep lesions and all retroperitoneal lesions were type V (Fig. 7). In two well-differentiated liposarcomas of deep lesions and one spindle cell lipoma, the MR signal intensities of the nodular or patchy nonadipose components were low intermediate or low relative to the muscle on T1- and T2-weighted images. The nodular components in these two well-differentiated liposarcomas histologically showed fibrous tissue and smooth muscle differentiation. The relationship between the locations of tumors and their internal characteristics is shown in Table 3. For well-differentiated liposarcomas, thick septa and nodular or patchy nonadipose components were present more frequently in deep and retroperitoneal lesions than in subcutaneous lesions. Five of eight subcutaneous well-differentiated li- posarcomas were type III, seven of 10 deep lesions were type IV, and all retroperitoneal lesions were type V ( p < 0.01). Two welldifferentiated liposarcomas with thin septa alone were located in the subcutaneous region, and no case showed thin septa alone in deep lesions. All septa were hypointense on both T1- and T2-weighted images, but the septa of well-differentiated liposarcomas tended to be enhanced more prominently than those of lipomas on fatsuppressed T1-weighted images after administration of gadopentetate dimeglumine (Table 4). No enhancement of the septa was observed in 11 A B AJR:180, June 2003 Fig year-old woman with well-differentiated liposarcoma of thigh (type III, one or two thick septa [> 2 mm] with low signal intensity detectable). A, T1-weighted MR image shows hyperintense multinodular tumor with well-defined and smooth margin. Some thick hypointense septa (arrows) are recognized. B, Fat-suppressed T1-weighted MR image obtained after administration of gadopentetate dimeglumine shows strongly enhanced septa. 1691

4 Ohguri et al. A B C D Fig year-old man with well-differentiated liposarcoma of thigh (type IV, three or more thick septa detectable). A, T1-weighted coronal MR image shows hypointense thick septa (arrows) in hyperintense tumor with partially irregular margin. B, Fat-suppressed T1-weighted coronal MR image obtained after administration of gadopentetate dimeglumine shows strongly enhanced septa in hypointense tumor. C, On photomicrograph, adipose tissue is separated by thick fibrous septa. Some large and small or small blood vessels are recognized within septa. (H and E, 5) D, On photomicrograph, vacuolated lipoblasts are frequently observed near septa (arrows). (H and E, 300) of 19 benign lipomas, but the septa in the welldifferentiated liposarcomas showed moderate or marked enhancement. This difference was significant (p < 0.01). Six of eight well-differentiated liposarcomas showed marked enhancement that was seen in only one benign lipoma (p < 0.001). The well-differentiated liposarcomas histologically contained thick fibrous septa with some TABLE 3 large and small or small blood vessels and vacuolated lipoblasts. Inflammatory cells and myxoid areas were frequently observed near the septa. Discussion Well-differentiated liposarcomas may be dismissed as little more than benign but locally aggressive lesions. The tumors consist of Relationship Between Tumor Type Based on Appearance of Internal Structures and Locations of Tumors in Well-Differentiated Liposarcoma Tumor Type Fig year-old man with well-differentiated liposarcoma of thigh (type V, nodular or patchy nonadipose component detectable). T1-weighted coronal MR image shows hypointense nodular tissues (arrows) and many thick septa in hyperintense multinodular tumor with well-defined and smooth margin Subcutaneous (n = 8) Deep (n = 10) Retroperitoneal (n = 5) 2 (25) 5 (62.5)a 1 (12.5) 1 (10) 7 (70)a 2 (20) 5 (100)a Type I Type II Type III Type IV Type V afisher s exact test, p < AJR:180, June 2003

5 MR Imaging of Benign Lipoma and Well-Differentiated Liposarcoma TABLE 4 Enhancement of Septa on Fat-Suppressed T1-Weighted MR Imaging After Administration of Gadopentetate Dimeglumine Degree of Enhancement of Septa p a (Relative to Muscle) Benign Lipoma (n = 19) Well-Differentiated Liposarcoma (n = 8) None 11 (58) < 0.01 Moderate 7 (37) 2 (25) 0.68 Marked 1 (5) 6 (75) < Note. Four cases without septa were excluded. a Fisher s exact test. abundant adipose tissue and grossly mimic benign lipomas but often recur if only marginally excised. A small percentage of these tumors dedifferentiate over time or progress histologically to high-grade lesions. Once dedifferentiation occurs, the tumors are more aggressive, with the potential to metastasize. Appropriate resection of a well-differentiated liposarcoma is thus required. Although radiologic findings of fatty tumors have been documented in many studies, the reliability of MR imaging to distinguish between benign lipoma and well-differentiated liposarcoma has not been fully discussed. Our results concur with a recent report by Kransdorf et al. [11], in which the authors reported the distinguishing features of lipoma and well-differentiated liposarcoma and suggested that the risk of malignancy increases with the advanced age and with the sex (men have a higher rate of liposarcoma) of the patient and the increased lesion size. In the case of our subjects, the average age for patients with benign lipoma was 55 years and for those with well-differentiated liposarcoma, 62.7 years. Of the lesions larger than 10 cm, two were benign lipomas and 12 were welldifferentiated liposarcomas. Well-differentiated liposarcomas were seen in 11 men and six women. Matsumoto et al. [9] reported that an infiltrative nature, which is a general characteristic of a malignant tumor, indicates benignity and not malignancy in an intramuscular (infiltrating) lipoma. In our three cases of intramuscular (infiltrating) lipomas, the margins were completely irregular at all points where the neoplastic fatty tissue infiltrated and intermingled with the surrounding muscle tissue. Completely irregular margins were recognized only in infiltrating lipomas, and therefore they still could be easily distinguished from well-differentiated liposarcomas, although this finding was not statistically significant (p = 0.27) because of the small number of patients with infiltrating lipoma. Multinodular margins were frequently recognized in well-differentiated liposarcomas, and uninodular margins were seen in benign lipomas with statistical significance. However, it may not be advisable to rely on this finding alone because some cases overlapped. Enzinger and Weiss [12] mentioned a tendency of well-differentiated liposarcoma to have more fibrous septa compared with lipoma, and they found that atypical cells or vacuolated lipoblasts admixed with fibroblastlike spindle cells are frequently situated in the septa surrounding irregularly sized lobules of fat. In our study, benign lipomas could be easily distinguished from well-differentiated liposarcomas when benign lipomas were completely composed of adipose tissue (p < 0.05), although nonadipose tissue was recognized in 25 of 35 benign tumors. Welldifferentiated liposarcomas of the extremities recur in nearly half of patients, whereas the recurrence rate in the retroperitoneum approaches 100%. Approximately one third of patients die as a direct result of their disease. Although deep somatic soft-tissue lesions recur frequently, subcutaneous lesions are generally cured by limited excision [12, 13]. Dedifferentiation occurs most frequently in retroperitoneal liposarcomas and in deep somatic lesions, but dedifferentiation is rare in subcutaneous tumors. The location of a tumor strongly influences its biologic behavior. MR findings may reflect the site-dependent differences in the behavior of well-differentiated liposarcoma. In our study, thick septa were more prevalent in deep lesions than in subcutaneous lesions (p < 0.01). In deep lesions, no case showed thin septa alone. Hosono et al. [8] reported that lipoblasts, vessels, myxoid areas, and inflammatory cells in addition to the septa in well-differentiated liposarcoma may contribute to the prominent enhancement after administration of gadopentetate dimeglumine. In our study, the septa in well-differentiated liposarcomas were enhanced more prominently than in benign lipomas on fat-suppressed T1- weighted images after administration of gadopentetate dimeglumine. In addition, cases with septa that showed no enhancement were all benign lipomas (p < 0.01). The well-differentiated liposarcomas in our study contained thick fibrous septa with some large or small blood vessels; and vacuolated lipoblasts, inflammatory cells, and myxoid areas were frequently observed near the septa. When considering these observations, we found that an adipose tumor situated in a deep location with only thin septa showing no enhancement after administration of gadopentetate dimeglumine may be safely regarded as benign. Careful assessment of internal fatty tumors on MR imaging is thus useful in distinguishing between well-differentiated liposarcomas and lipomas and may contribute to a more accurate preoperative assessment for the surgical approach. Fat-suppressed T1-weighted MR imaging was performed after IV administration of gadopentetate dimeglumine in six of seven benign lipomas in which a few thick septa had been recognized. Unlike well-differentiated liposarcomas, no marked enhancement of the septa could be seen in these cases, except one, after administration of gadopentetate dimeglumine. Therefore, it is likely that septa enhancement plays a more important role than septa thickness in differentiating benign lipoma from well-differentiated liposarcoma. Angiolipoma may be difficult to distinguish from well-differentiated liposarcoma by enhancement with gadolinium alone, but it often occurs in the forearm, usually has multiple lesions, and may be tender on palpation. These characteristics can be helpful in differentiating angiolipoma from other tumors. Although our cases of fibrolipoma and lipoma with myxoid change showed less enhancement than did muscle, the enhancement of these types of lipoma may be influenced by the vascular richness of internal structures. Nodular or patchy nonadipose components were recognized in both benign lipomas and well-differentiated liposarcomas. Two cases of parosteal lipoma and one case of spindle cell lipoma showed these components. In two parosteal lipomas, MR imaging clearly showed the lipomatous and osteochondromatous components. A correct preoperative diagnosis of this rare tumor is possible if these characteristic features are confirmed on MR imaging. A spindle cell lipoma is a lipomatous mass in which primitive, benign collagen-forming spindle cells have either partially or totally replaced mature fat. Pleomorphic lipoma may be a variant of a spindle cell lipoma. Kransdorf et al. [5] found that areas of spindle cell prolifera- AJR:180, June

6 Ohguri et al. tion within the fatty tumor show soft-tissue attenuation on CT, and as such, these tumors can mimic liposarcoma. The nodular components in our two cases of well-differentiated liposarcomas histologically showed fibrous tissue and smooth muscle differentiation. In two well-differentiated liposarcomas situated in deep locations and in one spindle cell lipoma, the MR signal intensities of the nodular or patchy nonadipose components that showed low intermediate signal intensity or low signal intensity on T1- and T2-weighted images were not specific. Spindle cell or pleomorphic lipoma is usually encountered in men between the ages of 45 and 65 years and is most frequently located in the posterior neck and shoulder. Although spindle cell or pleomorphic lipoma shares the MR findings of well-differentiated liposarcomas, such clinical information may be of assistance in reaching an accurate diagnosis. In conclusion, although some MR findings are shared by both benign lipomas and well-differentiated liposarcomas, MR imaging of the margins and internal characteristics is useful for the preoperative diagnosis of these tumors. References 1. Evans HL, Soule EH, Winkelmann RK. Atypical lipoma, atypical intramuscular lipoma, and well differentiated retroperitoneal liposarcoma. Cancer 1979;43: Azumi N, Curtis J, Kempson RL, Hendrickson MR. Atypical and malignant neoplasms showing lipomatous differentiation: a study of 111 cases. Am J Surg Pathol 1987;11: Azzopardi JG, Iocco J, Salm R. Pleomorphic lipoma: a tumour simulating liposarcoma. Histopathology 1983;7: Kransdorf MJ, Jelinek JS, Moser RP Jr, et al. Soft-tissue masses: diagnosis using MR imaging. AJR 1989;153: Kransdorf MJ, Moser RP, Meis JM, Meyer CA. Fat containing soft tissue masses of the extremities. RadioGraphics 1991;11: Jelinek JS, Kransdorf MJ, Shmookler BM, Aboulafia AJ, Malawer MM. Liposarcoma of the extremities: MR and CT findings in the histologic subtypes. Radiology 1993;186: Munk PL, Lee MJ, Janzen DL, et al. Lipoma and liposarcoma: evaluation using CT and MR imaging. AJR 1997;169: Hosono M, Kobayashi H, Fujimoto R, et al. Septum-like structures in lipoma and liposarcoma: MR imaging and pathologic correlation. Skeletal Radiol 1997;26: Matsumoto K, Hukuda S, Ishizawa M, Chano T, Okabe H. MRI findings in intramuscular lipomas. Skeletal Radiol 1999;28: Matsumoto K, Takada M, Okabe H, Ishizawa M. Foci of signal intensities different from fat in well-differentiated liposarcoma and lipoma: correlation between MR and histological findings. Clin Imaging 2000;24: Kransdorf MJ, Bancroft LW, Peterson JJ, Murphey MD, Foster WC, Temple HT. Imaging of fatty tumors: distinction of lipoma and well-differentiated liposarcoma. Radiology 2002;224: Enzinger FM, Weiss SW. Liposarcoma: soft tissue tumors, 4th ed. St. Louis: Mosby, 2001: Rosai J, Akerman M, Dal Cin P, et al. Combined morphologic and karyotypic study of 59 atypical lipomatous tumors: evaluation of their relationship and differential diagnosis with other adipose tissue tumors. Am J Surg Pathol 1996;20: AJR:180, June 2003