The value of fat-suppressed T2 or STIR sequences in distinguishing lipoma from well-differentiated liposarcoma

Transcription

1 Eur Radiol (2003) 13: DOI /s MUSCULOSKELETAL J. Galant L. Martí-Bonmatí F. Sáez R. Soler R. Alcalá-Santaella M. Navarro The value of fat-suppressed T2 or STIR sequences in distinguishing lipoma from well-differentiated liposarcoma Received: 18 October 2001 Revised: 11 March 2002 Accepted: 18 March 2002 Published online: 24 July 2002 Springer-Verlag 2002 J. Galant ( ) M. Navarro Servicio de Radiodiagnóstico, Hospital Universitario San Juan de Alicante, Ctra. Nacional 332 Alicante Valencia s/n, San Juan de Alicante, Spain galant_joa@gva.es L. Martí-Bonmatí Department of Radiology, Hospital Universitario Dr. Peset, Valencia, Spain F. Sáez Department of Radiology, Hospital Cruces de Baracaldo, Vizcaya, Spain R. Soler Department of Radiology, Hospital Juan Canalejo, A Coruña, Spain J. Galant Resonancia Magnética del Sureste, Murcia, Spain R. Alcalá-Santaella Department of Traumatology, Hospital Universitario San Juan de Alicante, Ctra. Nacional 332 Alicante Valencia s/n, San Juan de Alicante, Spain Abstract The objective of this study was to evaluate the diagnostic value of fat-suppressed T2-weighted (FS-T2) images or short tau inversion recovery (STIR) imaging in distinguishing lipoma from lipoma-like subtype of well-differentiated liposarcoma. Spin-echo T1-weighted and STIR or fat-suppression T2-weighted sequences were performed in 60 lipomas and 32 lipoma-like well-differentiated liposarcomas, histologically proven, looking for thick septa or nodules in T1-weighted images and linear, nodular, or amorphous hyperintensities on FS-T2/STIR sequences. Fourteen lipomas (23.3%) showed thick septa and/or nodules on T1, whereas on FS-T2 or STIR sequences only seven (11.7%) displayed hyperintense nodules and/or septa. All well-differentiated liposarcomas contained these signs on FS-T2 or STIR sequences. The presence of hyperintense septa or nodules in a predominantly lipomatous tumor on FS-T2/STIR sequences helps to differentiate malignant tumors from lipomas. Employing the presence of hyperintense nodules and/or septa as criteria of malignancy specificity was 76.6% and sensitivity 100%. Overdiagnoses of welldifferentiated liposarcoma can occur due to the presence of non-lipomatous areas within lipomas. Keywords Lipoma Well-differentiated Liposarcoma Fat-suppression Introduction Tumors with fat origin constitute an important group within soft tissue masses that can be characterized with MRI because of their predominant signal intensity similar to that of subcutaneous fat on all the pulse sequences [1]. Malignant tumors of the group, named liposarcomas, represent approximately 12 14% of soft tissue sarcomas [2, 3]. In 1962 Enzinger and Winslow classified liposarcomas into four types: well-differentiated; myxoid; round-cell type; and pleomorphic with acknowledging of mixed forms [4]. More recently, dedifferentiated liposarcoma has been considered a fifth type [5]. Actually, it is well known that myxoid liposarcoma can be accompanied by a varying number of round cells in such a way that so-called round cell liposarcoma is the extreme ex-

2 338 pression of this phenomenon rather than a specific type of liposarcoma [6]. Well-differentiated liposarcomas are placed on the less aggressive side of the spectrum resembling ordinary lipomas grossly, making occasionally difficult to distinguish both. The terms well-differentiated liposarcomas and atypical lipoma are synonyms, although the latter is actually preferred to describe tumors arising in the extremities on the grounds that such lesions, although capable of local recurrence, neither behave in a locally destructive fashion nor metastasize [6, 7, 8, 9]. On the other hand, to complicate even more the differential diagnosis with well-differentiated liposarcomas, some lipomas can contain mesenchymal non-fatty elements and thus differ from their typical homogeneous MR appearance. Our purpose is to evaluate the value of fat-suppressed T2-weighted (FS-T2) or STIR sequences in the differentiation between lipomas and lipoma-like subtype of welldifferentiated liposarcomas. Patients and methods From a series of 435 patients with musculoskeletal soft tissue tumors of the extremities or trunk wall examined with MR imaging, 155 lesions with extensive fatty component were retrospectively selected. Only surgically excised tumors with histologic diagnosis of lipoma and lipoma-like well-differentiated liposarcomas were included. Three tumors biopsied before imaging studies were excluded, as well as a tumor with foci of dedifferentiation. Finally, the study group was composed of 60 lipomas and 32 atypical lipomas. Patients were studied at five different hospital centers. Imaging studies were performed on 0.5- and 1.5-T Gyroscan NT (Philips Medical System, Eindhoven, The Netherlands) and 0.5-T Signa (General Electric, Milwaukee, Wis.). Spin-echo T1-weighted (TR/TE: ms/15 30 ms) and fat-suppression (FS) T2- weighted spin-echo (SE; TR: ms/te: 80, 120) or short tau inversoin recovery (STIR; TR/TE/TI: ms/40 ms/ ms) images were systematically obtained. Five tumors (3 atypical lipoma and 2 lipomas) were also imaged with FS T1-weighted sequences after the intravenous bolus administration of a paramagnetic gadolinium chelate (0.1 mmol/kg of Gd-DTPA- MBA). Two radiologists, who were masked to the final diagnosis, reviewed the images independently. Results were compared when disagreement occurred. In those instances, the two observers discussed the findings and reached a consensus opinion. Reading protocol included analysis of both T1-weighted and FS T2/STIR sequences. On T1-weighted images the presence of nodules and/or septa with non-fatty signal intensity within tumors was analyzed. Septa were further classified into thick and faint, the latter being considered when a regular and fine pencil lines appearance was found. In the analysis of FS-T2/STIR images, the presence of hyperintense septa (faint or thick) or nodules were recorded. The presence on FS images of weak hyperintense, poorly defined, nonlinear, and non-nodular areas with amorphous shape, was also registered. A semi-quantitative scale of enhancement (none, moderate, or intense) was applied for the analysis of the five tumors studied after contrast administration. The T1-weighted criteria for well-differentiated liposarcoma were thick septation and/or nodules with non-fatty signal intensity within tumors. When imaging in FS-T2 or STIR sequences, a lesion was considered well-differentiated liposarcoma when linear or nodular well-defined hyperintensities were detected. Sensitivity and specificity were calculated. Pathologists were aware of the age and gender of the patients as well as location of tumors. Recognition of lipoblasts was considered necessary to diagnose well-differentiated liposarcomas. Pathologic descriptions of the macro and microhistopathology of the tumors were reviewed for each lesion. Lipomas without significant mesenchymal components other than adipocytes were termed classic lipomas. The relation between the pathologic diagnosis and the presence of septa and/or nodules was analyzed by contingency tables (Fisher s exact test). Descriptive parameters of sensitivity, specificity, as well as positive and negative predictive values were obtained. Results Results are summarized in Table 1. In T1-weighted images 14 lipomas (23.3%) differed from the standard description of lipomas (homogeneous appearance with or without faint septa (Fig. 1a). Thick septation was present in 8 of these masses (13.3%; Fig. 2a), nodules were seen in 4 (6.7%; Fig. 3a), and thick septa and nodules were seen in 2 (3.3%) cases. In FS-T2/STIR images, hyperintense thick septa were found in 3 lipomas (Fig. 4b), nodules in 2 (Fig. 3b), and both together in 2 lipomas. Regarding well-differentiated liposarcomas in T1- weighted images, thick septa, nodules, or both (Figs. 5b, Table 1 Relationship between histology and septa thickness on T1-weighted images and hyperintensity on fat-suppressed T2/short tau inversion recovery (FS-T2/STIR) images of non-fat components. WDLS well-differentiated liposarcomas T1 FS-T2/STIR Septa Nodules Nodules Frank hyperintensities Weak and thick hyperintensities Fine Thick septa Septa Nodules Nodules and thick Fine Thick septa Lipomas (n=60) 41 (68.3) 8 (13.3) 4 (6.7) 2 (3.3) 0 3 (5) 2 (3.3) 2 (3.3) 11 (18.3) WDLS (n=32) 21 (65.6) 19 (59.4) 4 (12.5) 8 (25) 7 (21.9) 17 (53.1) 5 (15.6) 10 (31.3) 20 (62.5) Numbers in parentheses are percentages

3 339 Fig. 1a, b Classic lipoma. a Sagittal T1-weighted image shows homogeneous lesion with signal intensity equal to that of subcutaneous fat. b Coronal short tau inversion recovery (STIR) image completely nullifies signal from fat Fig. 2a, b Fibro-osteo-lipoma. a Sagittal T1-weighted image shows internal low intensity septa that remain hypointense on b sagittal FS-T2-weighted image 6b) were found in all but one tumor (Fig. 6a). On FS- T2/STIR images, hyperintense nodules, septa, or both were present in all of the malignant tumors (hyperintense septa in 17 lesions, hyperintense nodules in 5, and both in 10 lesions). Of the 14 lipomas with nodules or thick septa on T1- weighted images, 7 were histologically classified as classic lipomas, whereas other 7 tumors contained different mesenchymal components (5 fibrolipomas, Fig. 4a; 1 osteolipoma, Fig. 2a; and 1 angiomyolipoma). Within one classic lipoma there was a nodular area reflecting a histologically confirmed focal fat necrosis (Fig. 7). The histologic study of the remaining 6 lipomas did not explain the presence on imaging of the nodules or septa. Five classic lipomas, an osteolipoma and a fibrolipoma showing nodules or septa on T1-weighted images, did not exhibit high signal intensity abnormalities on FS- T2/STIR (this sequence nullified the signal from 6 thick septa and 1 nodule detected on T1-weighted images; Fig. 2b). T1-weighted images diagnosed well-differentiated liposarcomas by the presence of thick septa, nodules, or both with a sensitivity and specificity of 96.9 and 76.7%, respectively. Positive and negative predictive values were 68.9 and 97.9%, respectively. In diagnosing well-differentiated liposarcomas, the presence of hyperintense nodules, septa, or both on FS-T2/STIR images had a sensitivity of 100%, specificity of 88.3%, and positive and negative predictive values of 82.1 and 100%, respectively. Hyperintense weak areas on FS-T2/STIR images displayed a sensitivity of 62.5%, specificity of 18.3%, positive predictive value of 64.5%, and negative predictive value of 80.3% in diagnosing well-differentiated liposarcoma. Moderate enhancement, evaluated on FS-T1-weighted sequences after gadolinium administration, was found within the septa of 3 well-differentiated liposarcomas, whereas two lipomas did not show any significant enhancement of the septa.

4 340 Fig. 3 Lipoma with a histologically proven area of fibrolipoma differentiation with nodular appearance on a sagittal T1-weighted and b STIR images Fig. 4a, b Paravertebral intramuscular fibrolipoma. a Axial T1-weighted image depicts septa within tumor. b A STIR image shows diffuse and nodular hyperintensities Fig. 5a, b Well-differentiated liposarcoma. a Coronal T1-weighted and b FS-T2 images demonstrate internal thick septa with high signal in the FS image

5 341 Fig. 6a, b Well-differentiated liposarcoma. a Whereas axial T1-weighted image shows a nearly completely homogeneous tumor with thin scattered septa, b the coronal STIR image depicts internal hyperintense nodules and lines Fig. 7 Lipoma containing a nodular focus of fat necrosis. Axial T1-weighted image shows a nodule with intermediate signal intensity between fat and muscle with its center of lower signal intensity. Histologically it was composed of inflammatory cells, adipocytes, and granulomatous changes. T2-weighted images (not shown) display a high signal intensity nodule with its center of higher signal intensity Discussion Benign lipoma constitutes by far the most common soft tissue tumor. Although in clinical practice most superficial lipomas are diagnosed solely by clinical examination and no further studies are required, some superficial lipomas and many of the deep located lipomas are presently studied with MR imaging. By doing this, lipoma represents the most commonly diagnosed soft tissue mass on MR imaging [10]. On the other hand, liposarcoma is second in frequency only to malignant fibrous histiocytoma among mesenchymal sarcomas, with an approximated incidence between 12 and 14% of all soft tissue sarcomas [2, 3]. Liposarcomas have been classified into five types: well-differentiated; myxoid; round-cell type; pleomorphic; and dedifferentiated [5]. While myxoid liposarcoma is the most common malignant type, well-differentiated liposarcoma is the second largest group. Both types are low-grade tumors and have a high disease-free survival rate [11]. Well-differentiated liposarcomas may be further subdivided, and being lipoma-like are the subtype that more closely resembles benign lipoma [6]. Well-differentiated liposarcomas have a very uncommon tendency to metastasize in such a way that some authors have proposed the term atypical lipoma for these tumors [6, 8, 9, 12, 13]. Although there is a low risk of recurrence for the extremity well-differentiated liposarcoma when wide surgical margin is achieved, lesions of large size cannot always be completely excised and tumors often recur if only marginal surgery is performed [13, 14]. Furthermore, recurrences can be associated with dedifferentiation which implies a poorer prognosis [13]. This allows to safely resect well-differentiated liposarcoma, and thus differential diagnosis between lipoma and well-differentiated liposarcoma is of utmost importance in the preoperative planning. Magnetic resonance signal characteristics of fat-containing neoplasm allows a reliable diagnosis by detecting signal intensity areas within the mass equal to that of subcutaneous and interfascial fat on all the pulse sequences, with loss of signal on FS techniques. It has been pointed out that difficulties in differentiating welldifferentiated liposarcoma from benign lipomas may exists in many cases [14, 15, 16, 17, 18, 19]. This is probably related to the close histologic similarity of low-grade liposarcomas and benign lipomas [20]. Some parameters, traditionally employed in differentiating benign from malignant soft tissue tumors with variable results, are not useful in distinguishing lipomas from well-differentiated liposarcomas. Considerable overlapping exists between sizes, due to the tendency of many deep lipomas to reach large diameters [5, 21]. Intramuscular lipomas not infrequently appear as infiltrative masses with intermingled muscle fibers within tumors [21], and many of the well-differentiated liposarcomas have well-defined margins [2]. Due to the fact that most lipomas tend to be homogeneous lesions with lobulated components separated by thin intervening con-

6 342 nective tissue septa, some authors have considered the heterogeneity the most reliable sign in distinguishing benign from malignant fatty tumors [10]; however, it is not unusual at all that some lipomas show a not completely homogeneous appearance [18, 22] related to the presence of other mesenchymal components. Also, the presence of nodules with signal intensity different from fat, which would argue against the diagnosis of lipoma, is occasionally found within lipomas, mostly in relation to fat necrosis. For those reasons, new signs must be investigated with the aim of achieving a higher confidence in diagnosis. Thick septation has been considered one of the hallmarks of well-differentiated liposarcomas [14, 23]. Nevertheless, some lipomas can also contain septa thicker than expected and therefore the parameter of thickness of septa, although useful, can be less reliable than is required. On T1-weighted images, septa and nodules within benign and malignant lesions may display similar linear low signal intensity. Conversely, when FS techniques are employed, benign thick septa tend to remain hypointense, opposing the hyperintense septa of low-grade liposarcomas. Moreover, the FS techniques are sensitive enough in the diagnosis of well-differentiated liposarcomas. In our series, all these malignant lesions presented with hyperintense nodules or septa on FS-T2 or STIR sequences, and only one well-differentiated liposarcoma containing a hyperintense nodular area on FS images was missed on T1-weighted sequences (100% sensitivity for FS techniques and 97% for T1-weighted sequence); however, differences in specificity do exist. Seven of 14 lipomas with nodules or thick septa on T1-weighted images were not hyperintense on FS-T2 or STIR sequences, allowing the correct diagnosis of lipomas; therefore, specificity rises from 77% of T1-weighted images to 88% when combining with FS T2-weighted images or STIR sequences. The reason why septa and nodules of well-differentiated liposarcomas are hyperintense on FS techniques, in contrast to the behavior of septa within lipomas, is not clear; the increased vascularity in the septa of the former, confirmed on pathologic study, has also been proved on gadolinium-enhanced studies [24] and may constitute the mayor contribution to the hyperintensity on FS T2- weighted or STIR sequences. Inflammatory cell infiltration and malignant cellularity may also play a role [25]. Weakly hyperintense amorphous non-nodular, non-linear areas appeared more frequently in malignant lesions in our series, but specificity was low. Previously, Hosono et al. [25] stated that, when studied with contrast-enhanced FS T1-weighted images, septa in well-differentiated liposarcoma enhanced considerably while only slightly in lipoma. This was concordant with a recent report [24] and the behavior of 3 well-differentiated liposarcomas in our series. The cases of our series that displayed a significant enhancement could be easily diagnosed solely on the bases of their appearance on non-enhanced images. The small number of patients examined with gadolinium-enhanced imaging does not allow us to raise a conclusion about whether strongly enhanced septa can constitute a valuable tool favoring diagnosis of malignancy. Probably further studies are necessary to determine the potential of post-gadolinium images in the discrimination between false-positive cases of the non-enhanced image analysis. Finally, it must be highlighted that the selection criteria of our series introduced a bias which makes the differentiation between lipoma and well-differentiated liposarcoma presumably more difficult than in a general population since lesions of the lipoma-like subtype would show an even more similar appearance to lipomas than do others subtypes of well-differentiated liposarcomas. Conclusion In conclusion, differential diagnosis between lipomas and atypical lipomas is not always easy. Despite similarities, differential diagnosis can be reliably established employing some image parameters. Presence of linear or nodular foci of hyperintensities on FS-T2/STIR sequences constitutes a more specific sign than the detection of thick septa or nodules in T1. In our series overall sensitivity and specificity for FS T2 or STIR sequences employing this sign was 100 and 82%, respectively. The presence of other mesenchymal components different from fat can alter the expected absence of hyperintense foci on FS-T2 or STIR sequences within lipomas. In our series, this was found in 4 fibrolipomas and 1 angiolipoma, whereas a fat necrosis focus within lipoma caused a nodular appearance in another case. References 1. Dooms GC, Hricak H, Sollitto RA, Higgins CB (1985) Lipomatous tumors and tumors with fatty component: MR imaging potential and comparison of MR and CT results. Radiology 157: Enzinger FM, Weiss SW (1988) Liposarcoma. In: Enzinger FM, Weiss SW (eds) Soft tissue tumors, 2nd edn. Mosby, St. Louis, pp Kransdorf MJ (1995) Malignant softtissue tumors in a large referral population: distribution of diagnoses by age, sex and location. Am J Roentgenol 164: Enzinger FM, Winslow DJ (1962) Liposarcoma. A study of 103 cases. Virchows Arch 335:

7 Marques MC, Garcia H, Vanhoenacker F (2001) Lipomatous tumors. In: De Schepper AM, Parizel PM, De Beuckeleer L, Vanhoenacker F (eds) Imaging of soft tissue tumors. Springer, Berlin Heidelberg New York, pp Rosay J (1996) Soft tissues. In: Rosay J (ed) Ackerman s surgical pathology, 7th edn. Mosby, St. Louis, pp Evans HL, Soule EH, Winkelmann RK (1979) Atypical lipoma, atypical intramuscular lipoma, and well-differentiated retroperitoneal liposarcoma: a reappraisal of 30 cases formerly classified as well-differentiated liposarcoma. Cancer 43: Evans HL (1988) Liposarcomas and atypical lipomatous tumors. A study of 66 cases followed for a minimum of 10 years. Surg Pathol 1: Azumi N, Curtis J, Kempson RL, Hendrickson MR (1987) Atypical and malignant neoplasms showing lipomatous differentiation. A study of 111 cases. Am J Surg Pathol 11: Sundaram M, Sharafuddin MJA (1995) MR imaging of benign soft-tissue masses. Magn Reson Imaging Clin North Am 3: Springfield D (1993) Liposarcoma. Clin Orthop 289: Lucas DR, Nascimento AG, Sanjay BK, Rock MG (1994) Well-differentiated liposarcoma. The Mayo clinic experience with 58 cases. Am J Clin Pathol 102: Weiss SW, Rao VK (1992) Well-differentiated liposarcoma (atypical lipoma) of the deep soft tissue of the extremities, retroperitoneum, and miscellaneous sites. A follow-up study of 92 cases with analysis of the incidence of dedifferentiation. Am J Surg Pathol 16: London J, Kim EE, Wallace S, Shirkhoda A, Coan J, Evans J (1989) MR imaging of liposarcomas: correlation of MR features and histology. J Comput Assist Tomogr 15: Jelinek JS, Kransdorf MJ, Shmookler MB, Aboulafia AJ, Malawer MM (1993) Liposarcoma of the extremities: MR and CT findings in the histologic subtypes. Radiology 186: Caron KH, Bisset GS III (1990) Magnetic resonance imaging of pediatric atraumatic musculoskeletal lesions. Top Magn Reson Imaging 3: Kransdorf MJ, Jelinek JS, Moser RP Jr (1993) Imaging of soft tissue tumors. Radiol Clin North Am 31: Murphy WD, Hurst GC, Duerk JL, Feiglin DH, Christopher M, Bellon EM (1991) Atypical appearance of lipomatous tumors on MR images: high signal intensity with fat-suppression STIR sequences. J Magn Reson Imaging 1: Munk PL, Lee MJ, Janzen DL, Connell DG, Logan PM, Poon PY, Bainbridge TC (1997) Lipoma and liposarcoma: evaluation using CT and MR imaging. Am J Roentgenol 169: Weatherall PT (1995) Benign and malignant masses. MR imaging differentiation. Magn Reson Imaging Clin North Am 3: Matsumoto K, Hukuda S, Ishizawa M, Chano T, Okabe H (1999) MRI findings of intramuscular lipomas. Skeletal Radiol 28: Gelineck J, Keller J, Myhre Jensen O, Molsen OS, Christensen T (1994) Evaluation of lipomatous soft tissue tumors by MR imaging. Acta Radiol 35: Bush CH, Spanier SS, Gillespy T III (1988) Imaging of atypical lipomas of the extremities: report of three cases. Skeletal Radiol 17: Yang YJ, Damron TA, Cohen H, Hojnowski L (2001) Distinction of well-differentiated liposarcoma from lipoma in two patients with multiple well-differentiated fatty masses. Skeletal Radiol 30: Hosono M, Kobayashi H, Fujimoto R et al. (1997) Septum-like structures in lipoma and liposarcoma: MR imaging and pathologic correlation. Skeletal Radiol 26: