1992, The British Journal of Radiology, 65, 14-20
Tumour and tumour-like conditions of the soft tissue: magnetic resonance imaging features differentiating benign from malignant masses By George Hermann, MD, Ibrahim Fikry Abdelwahab, MD, Theodore T. Miller, MD, "Michael J. Klein, MD and tMichael M. Lewis, MD Departments of Radiology, *Pathology and tOrthopedics of The Mount Sinai Medical Center, CUNY, New York, USA {Received 5 March 1991 and in revised form 21 June 1991, accepted 23 July 1991)
Keywords: Soft-tissue tumour, MRI, Benign, Malignant
Abstract. 49 primary soft-tissue masses were studied by magnetic resonance imaging (MRI) in an effort to differentiate benign from malignant lesions. There were 24 benign and 25 malignant tumours. Images were evaluated for the presence of low-signal septation, changes in pattern of homogeneity, signal intensity, and margin definition and shape. 20 of the 25 malignant tumours (80%) demonstrated low-signal septation, and two of the 24 benign tumours (8%) also demonstrated this feature. Malignant tumours tended to change pattern from being homogeneous on Trweighted images to being heterogeneous on ^-weighted images (72%). Benign tumours maintained their pattern on both sequences (67%). Only three of the 24 benign tumours (12.5%) changed from being homogeneous to heterogeneous. Our findings suggest that tumours which change pattern, from homogeneous on ^-weighted images to heterogeneous on 7^-weighted images, and tumours which have low-signal septations are likely to be malignant. Moreover benign lesions tend to have well defined margins, and some benign masses have characteristic appearances that aid in their differentiation from malignant processes.
Since the introduction of magnetic resonance imaging (MRI), its diagnostic efficacy has been compared with computed tomography (CT). While the superiority of MRI for visualizing the skeleton and surrounding soft tissues has been documented (Petasnick et al, 1986; Totty et al, 1986; Demas et al, 1988; Sundaram et al, 1988; Kransdorf et al, 1989a; Berquist et al, 1990), its ability to provide a histological diagnosis has not been established. There are no established features with which to predict or distinguish malignant from benign lesions, although tumour signal intensity, margination, homogeneity and the presence of calcification have been used by some authors (Totty et al, 1986; Sundaram et al, 1988; Kransdorf et al, 1989a; Berquist et al, 1990). By modifying these criteria to include tumour morphology, presence of septation, changes in pattern of homogeneity, signal intensity and margin definition, we have attempted to identify certain trends that may be useful in distinguishing benign from malignant tumours. Materials and methods
MR images of 49 primary soft-tissue masses in 49 patients were reviewed. The patients, 26 males and 23 females, ranged in age from 6 to 75 years. The overall Address correspondence to George Hermann, MD, Box 1234, Department of Radiology, Mount Sinai Medical Center, One Gustave L. Levy Place, New York, NY 10029-6574, USA. 14
average age of the patients was 42 years in the benign and 38 years in the malignant groups. In most cases, the MRI studies were performed in our institution, but in several instances the patients were referred to us for second opinions or for surgery. All patients underwent surgery at our institution. The MRI scans were obtained on an Elscint 5000 0.5 T, Siemens Magnetom 1.0 T or a Diasonic 0.3 T system, but we did not evaluate images for possible differences in field strength. Multisection and multiplanar techniques were applied in the majority of cases. Spin-echo Tr and ^-weighted pulse sequences were 450-700/20-30 TR/TE (repetition time/echo time) and 1800-2200/80 TR/TE, respectively. The images were obtained in a combination of coronal, sagittal and/or axial planes. Section thickness varied from 4 to 10 mm. Surface coils were used appropriate to the site being imaged. In all cases diagnosis was confirmed with the pathological specimen except in three cases in which a radiological diagnosis could be made with certainty (e.g. two small lipomas and one popliteal cyst), obviating the need for operation. The MR images were reviewed by two radiologists together (G.H. and T.T.M.) and were evaluated with regard to: (1) shape, (2) morphology, (3) pattern, (4) signal intensity, and (5) definition of margins. The shape of the tumour was considered either round-oval or irregular. Morphology refers to the presence of lowsignal septations which traversed the lesion and divided The British Journal of Radiology, January 1992
MRI of soft-tissue tumours
it into lobules. Tumours could demonstrate low-signal septations while maintaining an overall shape of roundoval, but an external contour of lobulation was not possible without septation, and such a shape by definition was irregular. The pattern was described either as homogeneous if all sections had uniform signal, or heterogeneous if the lesion contained varied signal. Any change in pattern between ^-weighted images (T^-WI) and %-weighted images (7J-WI) was noted. The evaluation of signal intensity was based on comparison with adjacent muscles in both 7J-WI and 7^-WI and was graded as low, equal or high. Tumour margins were evaluated as ill defined, partially or well defined according to their relationship to the surrounding tissue: ill defined when the borders were confluent with the surrounding tissue; partially defined if at least part of the border of the mass was confluent with the surrounding tissue; and well defined when the entire tumour was sharply demarcated from the adjacent tissues. Results
Of the 49 soft-tissue masses, 24 were benign and 25 malignant (Tables I, II).
In the benign group, eight occurred in the thigh, four in the pelvis, three in the shoulder region, three in the calf, two in the forearm, two in the popliteal fossa, one in the ankle and one in the foot. In the malignant group, 10 occurred in the thigh, four in the popliteal fossa, three in the calf, three in the back, two in the foot, two in the elbow, and one in the hip (Fig. I). The size of the tumours ranged from 2 to 20 cm, with an average of 8 cm in greatest diameter. Shape and morphology The overall configurations of the mass in both benign and malignant lesions were equally divided. 27 were round-oval (14 malignant, 13 benign) and 22 irregular (11 malignant, 11 benign). Low-signal septation was present in 20 of 25 malignant lesions (80%) (Fig. 2) and in only two of 24 benign masses (8%). 18 of the 20 (90%) malignant septated masses can be broadly classified histologically as spindle cell tumours: 10 malignant fibrous histiocytomas, three synovial sarcomas, one malignant Schwannoma, one undifferentiated spindle cell sarcoma, one liposarcoma, one haemangiopericytoma and one extraosseous osteosarcoma. The remaining two septated malignancies were extraosseous
Table I. Benign conditions (n = 24) Tumour Pattern Lipoma Lipoma Lipoma Lipoma Lipoma Atypical lipoma Neurofibroma Neurofibroma Neurofibroma PVNS PVNS Ganglion Ganglion Desmoid Desmoid Haematoma Haematoma Abscess Abscess Infection Popliteal cyst Myxoma Haemangioma Benign Schwannoma
Septation
T2-WI
Hm Ht Hm Hm Hm Hm Ht Hm Ht Hm
Ht Ht Hm Ht Hm Ht Ht Ht Ht Ht Ht Hm Ht Hm
SI
Definition
Pattern
SI
Definition
Hm Ht Hm Hm Hm Hm Ht Hm Hm Ht Ht Hm Ht Ht Hm Ht Ht Hm Hm Hm Ht Hm Ht Ht
Pattern: Hm, homogeneous; Ht, heterogeneous. SI: (—) SI Jess than adjacent muscle, (=) SI equa) to adjacent muscle, (-h) SI greater than adjacent musde. Definition: (+) poorly defined margins, (+ +) partially well defined, (+ + +) well defined. Septation: (—) none, ( + ) seen on T2-WI only, ( + ) present on Tr and r 2 -WI. Vol. 65, No. 769
15
G. Hermann, I. F. Abdelwahab, T. T. Miller, M. J. Klein and M. M. Lewis
Ewing's sarcoma and an alveolar soft-part sarcoma. The two benign septated masses consisted of an abscess and an atypical lipoma. None of the benign spindle cell tumours (e.g. neurofibroma, benign Schwannoma, desmoid) showed any evidence of septation.
pattern. Of these 23 masses, 16 (70%) were benign. Nine tumours remained homogeneous, of which eight (89%) were benign. 14 tumours remained heterogeneous: eight benign (57%) and six malignant (43%).
Signal intensity Pattern Compared with adjacent muscle on 7^-WI, 10 of 25 In the malignant group, 18 of the 25 tumours (72%) malignant tumours showed a high signal, one a low changed pattern between Tx- and T^-WI: 16 changed signal and 14 of 25 as isointense signal. However, all from homogeneous to heterogeneous (Fig. 3) and two these tumours showed high signal on TrWI. Overall, 15 vice versa. Seven of 25 (28%) did not change pattern: of 25 (60%) changed signal intensity. one remained homogeneous and six remained heteroThe benign tumours had high signal intensity in 14 of geneous. In contrast, only eight of the 24 benign 24 lesions on TJ'-WI, isointensity in six and low signal tumours (33%) changed pattern (three homogeneous to intensity in four. 21 of the 24 exhibited high signal on heterogeneous and five vice versa). Sixteen of the 24 ZJ-WI. The three benign tumours that were not intense benign tumours (67%) did not change, with eight on ZJ-WI included two pigmented villonadular synovitis remaining homogeneous and eight heterogeneous on (PVNS), which were isointense on both Tr and 7J-WI, both sequences. and one desmoid tumour which was of low signal Combining the two groups, 26 of the 49 masses intensity on both. Overall, only seven of 24 (29%) changed pattern, and eighteen of the 26 (69%) were changed signal intensity. Combining the results, 18 malignant. More significantly, 19 of these 26 changed tumours were isointense on 7^-WI and bright on 7^-WI, from a homogeneous pattern on TJ'-WI to a hetero- of which 14 (78%) were malignant. geneous one on 7^-Wl, of which 16 (84%) were malignant. The remaining seven of 26 tumours changed from Margin definition heterogeneous to homogeneous, of which only two 10 malignant lesions were well defined, 10 were (29%) were malignant and five (71%) were benign. partially defined and one poorly defined on both Tr and In contrast, 23 of the 49 masses did not change 7^-WI. In addition, two changed from partially to well Table II. Malignancies (n = 25) Tumour
MFH MFH MFH MFH MFH MFH MFH MFH MFH MFH MFH MFH Synovial sarcoma Synovial sarcoma Synovial sarcoma Malignant Schwannoma Malignant Schwannoma Extraosseous Ewing's sarcoma Extraosseous Ewing's sarcoma Spindle cell sarcoma Liposarcoma Extraosseous osteosarcoma Lymphoma Haemangiopericytoma Alveolar soft-part sarcoma
7;-wi
Septation
T2-WI
Pattern
SI
Definition
Pattern
SI
Definition
Ht Hm Hm Hm Hm Hm Hm Hm Hm Hm Ht Ht Hm Hm Hm Ht Hm Hm Hm Ht Ht Ht Hm Hm Ht
=
+
+++ +++ +++ ++ +++ +++ ++ ++ +++ ++ +++ ++ +++ ++ +++ ++ ++ +++ ++ +++ ++
Hm Ht Ht Ht Ht Ht Ht Ht Ht Ht Hm Ht Ht Ht Ht Ht Ht Ht Ht Ht Ht Ht Ht Hm Ht
+ + + + + + + + + + + + + + +
+
= = =
+ + = + = = + + + — = + = = + + = = = +
+
++ ++
+ + + + + + + + + +
+++ +++ +++ ++ +++ +++ +++ ++ ++ +++ +++ ++ +++ ++ +++ ++ ++ +++ ++ +++ ++ +++ ++ ' ++
±
+
+
±
+ + + +
± — +
±
+ + — +
± — + + + +
±
MFH = Malignant fibrous histiocytoma. See Table I for key to symbols.
16
The British Journal of Radiology, January 1992
MRI of soft-tissue tumours
istic appearances of certain benign masses further aids differentiation. BENIGN MALIGNANT Although the shape of the tumour is non-specific, the presence of low-signal septation strongly suggests malignancy. This feature has been mentioned in other reports of soft-tissue sarcomas but its relationship to malignancy has not been emphasized. A septated appearance was described by London et al (1989) in their series on liposarcomas. Kransdorf et al (1989a) in their paper on soft-tissue tumours showed similar septation in illustrations of myxoid liposarcoma and neurofibrosarcoma but did not discuss this as a feature of malignancy. Similarly, Demas et al (1988) illustrated a soft-tissue sarcoma with low-signal septation. In our series, this septated appearance was apparent in 20 out of 25 of the malignant tumours (80%), in contrast with only two out of 24 of the benign tumours (8%). While it is not surprising that 18 out of 20 septated malignant tumours were spindle cell types since they are of mesenchymal origin, it is noteworthy that none of the benign spindle cell tumours in our series showed this septated morphology. Malignant fibrous histiocytoma (MFH) was the most common tumour in our series. 10 of the 12 cases of MFH had a lobulated appearance both on Tr and 7J-WI, consisting of irregular nodular masses separated by fine low-signal-intensity bands. This feature was also seen in our three synovial sarcomas. Septation in liposarcomas has previously been described by London et al (1989), who believed that septation represented a more differentiated fatty component in the tumour. Figure 1. Distribution of primary soft-tissue tumours (24 Alternatively these low-signal septa may represent benign, 25 malignant). streaks of fibrous tissue or regions of relative acellularity. One can speculate that non-uniform stages in the growth of tumours may be separated by fibrous tissue defined, one from poorly to well defined and one from and show a pattern of septation. Although this appearance is highly suggestive of well defined to partially defined. In the benign category, 16 tumours were well defined, three partially defined malignancy, it is not infallible. Two benign tumours, an and two poorly defined on both Tr and 7J-WI. Two abscess and an atypical lipoma, showed this appearance changed from partially to well defined, and one from and were mistakenly diagnosed pre-operatively as well to partially defined. In the benign category, 16 malignant. In a recent report, an intramuscular myxoma tumours were well defined, three partially defined and was mistakenly but reasonably diagnosed pre-operatwo poorly defined on both Tx- and 7J-WI. Two changed tively as a liposarcoma because of similar septation from partially to well defined, and one from well to (Kransdorf et al, 1989a). In the absence of septation we found no significant partially defined. patterns that differentiate benign nerve sheath tumours from their malignant counterparts. In all instances they Discussion It is generally accepted that soft-tissue tumours are were well demarcated from the neighbouring structures. One benign Schwannoma and two malignant better demonstrated with MRI than with CT or ultrasound (Petasnick et al, 1986; Totty et al, 1986; Chang et Schwannomas appeared homogeneous on 7J-WI and al, 1987; Demas et al, 1988; Sundaram et al, 1988; heterogeneous on ZJ-WI. None of the benign Kransdorf et al, 1989a; Berquist et al, 1990). The major Schwannomas showed evidence of septation. Thus, advantages of MRI are its ability to display contrast while the absence of septations does not exclude a between tissues and its capacity for multiplanar malignancy, their presence is strongly against a benign imaging. Recent investigations attempting to use MRI process. The second criterion we found useful was whether the to distinguish benign tumours from malignancies have been inconclusive (Pettersson et al, 1987; Sundaram et tumour was homogeneous or heterogeneous and if there al, 1988; Kransdorf et al, 1989a). Our study suggests was any change in pattern that occurred from TJ'-WI to that tumour septation and changes in homogeneity may j;-WI. The pattern of 18 of the 25 malignancies (72%) be useful in predicting malignancy, and that character- changed between 7J-WI and JJ-WI, with 16 of the 18 Vol. 65, No. 769
17
G. Hermann, I. F. Abdelwahab, T. T. Miller, M. J. Klein and M. M. Lewis
Figure 2. Malignant fibrous histiocytoma in a 72-year-old woman, (a) Sagittal image shows a slightly higher signal intensity mass behind the distal femur on 7J-WI (SE 600/30). (b) On ZJ-WI (SE 1800/100) the mass shows high signal, is heterogeneous and well defined. Note the low-signal-intensity septa (arrows) causing lobulation on both Tr and 7|-WI. (c) Malignant fibrous histiocytoma. Low power shows adjacent myxoid areas with cellular fibrous septa between them corresponding to an area of decreased signal intensity on 7J-WI (centre) ( x 11).
homogeneous on 7^-WI and heterogeneous on 7^-WI. Only three of 24 benign tumours (12.5%) behaved similarly. Thus it appears that the change from a homogeneous pattern to a heterogeneous one is of prognostic significance. The majority of benign tumours (67%) maintained their original pattern of homogeneity on both Tr and JJ-WI sequences. Moreover, of all the tumours in our series, nine were homogeneous on both sequences, of which eight (89%) were benign. If one examines only T2-Wl, our data agree with that of Petasnick et al (1986) who concluded that on IJ-WI, benign lesions tend to be more homogeneous than 18
malignant tumours. Indeed, of our 49 tumours, 18 were homogeneous on 7^-WI, of which 13 (72%) were benign. However, we emphasize that it is the change of the tumour from the 7^-WI pattern to the 7^-WI pattern that is important and not the isolated appearance of one without the other. Therefore, in each case both sequences should be carefully evaluated. The criteria of signal intensity and margin definition were generally not helpful. Almost all soft-tissue tumours are of high signal on 7^-WI, and lesions that demonstrated high signals in both sequences were almost equally divided between benign and malignant. The British Journal of Radiology, January 1992
MRI of soft-tissue tumours
Figure 3. Liposarcoma of the popliteal region in an 18-year-old male, (a) Sagittal 7J-WI (600/30) shows septation. (b) Coronal 7^-WI (2100/80) appears heterogeneous with exterior lowsignal-intensity septa that divide the mass into smaller lobules (arrows).
However, those tumours that were isointense on 7^-WJ and of high signal on 7J-WI were predominantly malignant. In addition, certain characteristic benign patterns have been described with which the findings in our cases concur; for example, PVNS is isointense on 7J-WI and 7^-WI (Kottal et al, 1987; Jelinek et al, 1989) and ganglia (Totty et al, 1986; Petasnick et al, 1986) and myxomas (Kransdorf et al, 1989) are low on TrWl and high on 7^-WI. In our series, desmoids gave a low signal on both 7/-WI and 7^-WI. These characteristic appearances are important because they aid in the recognition of these processes as benign. For example, despite the infiltrative, locally aggressive nature of desmoid tumours mentioned by Berquist et al (1990), the persistent low signal intensity on both 7J-WI and 7J-WI should strongly suggest the diagnosis. The underlying pathological explanation for these appearances is the presence of fluid in ganglia and myxomas, dense fibrotic tissue in desmoids, and intra and extracellular haemosiderin in PVNS. Similarly, decreased signal owing to haemosiderin deposition has been observed in cases of repeated transfusion for thalassaemia (Brasch et al, 1984). In one case of MFH in which myxomatous tissue was predominant, the tumour was isointense on TJ'-WI and of homogeneously high signal on 7J-WI (Fig. 4). Sundaram et al (1987) found that tumours which are relatively acellular and contain a large amount of collagen show relatively low signal intensity on 7J-WI, and concluded Vol. 65, No. 769
that the cellularity of a tumour rather than the histological diagnosis influences the MRI signal on 7J-WI. Petasnick et al (1986) and Berquist et al (1990) suggest that benign masses show well defined borders. Our data generally agree with this finding, although in our series it was not reliable enough to be useful in distinguishing benign from malignant lesions. In summary, our findings suggest that tumours which change in homogeneity from 7J-WI to T2-Wl, and especially those that alter from a homogeneous pattern to a heterogeneous pattern, are more likely to be malignant. Lesions that do not change, and especially those that remain homogeneous, are most likely benign. In addition, tumours with septation are most likely malignant, and probably of spindle cell origin. Where these two criteria conflict (i.e. a lobulated tumour that does not change pattern), we believe that the presence of lowsignal septation is a stronger predictor of malignancy than a change in pattern. References BERQUIST, T. H., EHMAN, R. L., KING, B. F., HODGMAN, C.
G.
& ILSTRUP, D. M., 1990. Value of MR imaging in differentiating benign from malignant soft-tissue masses: study of 95 lesions. American Journal of Roentgenology, 155, 1251-1255. BRASCH, R. C , WESBEY, G. E., GOODING, C. A. & KOERPER,
M. A., 1984. Magnetic resonance imaging of transfusional hemosiderosis complicating thalassemia major. Radiology, 150, 767-77'1.
19
G. Hermann, I. F. Abdelwahab, T. T. Miller, M. J. Klein and M. M. Lewis
I
(b)
Figure 4. 52-year-old woman. Mass on the dorsal aspect of the foot in sagittal plane showing (a) homogeneous low signal intensity on 7J-WI (600/20). (b) High signal on 7^-WI (2500/80). (c) Low-grade myxoid malignant fibrous histiocytoma. High magnification reveals stellate and spindle-shaped cells with long cytoplasmic processes and atypical nuclei in a myxoid stroma (x88). myxoma: MR features. Journal of Computer Tomography, 13, 836-839.
Assisted
CHANG, A. E., MATORY, Y. L., DWYER, A. J., HILL, S. C , GIRTON, M. E., STEINBERG, S. M., KNOP, R. H., FRANK, J. A., HYAMS, D., DOPPMAN, J. L. & ROSENBERG, S. A., 1987.
LONDON, J., KIM, E. E., WALLACE, S., SHIRKHODA, A., COAN, J.
Magnetic resonance imaging versus computed tomography in the evaluation of soft tissue tumors of the extremities. Annals of Surgery, 205, 340-348.
& EVANS, H., 1989. MR imaging of liposarcomas: correlation of MR features and histology. Journal of Computer Assisted Tomography, 12, 832-835.
DEMAS, B. E., HEELAN, R. T., LANE, J., MARCOVE, R., HAJDU,
PETASNICK, J. P., TURNER, D. A., CHARTERS, J. R., GITELIS, S.
S. & BRENNAN, M. F., 1988. Soft tissue sarcomas of the extremities: comparison of MR and CT in determining the extent of disease. American Journal of Roentgenology, 150, 615-620. JELINEK, J. S., KRANSDORF, M. J., U T Z , J. A., BERREY, B. H., THOMSON, J. D., HEEKIN, R. D. & RADOWICH, M. S., 1989.
Imaging of pigmented villonodular synovitis with emphasis on MR imaging. American Journal of Roentgenology, 152, 337-342. KOTTAL, R. A., VOGLER, J. B., Ill, MATAMOROS, A., ALEXANDER, A. H. & COCKSON, J. L., 1987. Pigmented
villonodular synovitis: a report of MR imaging in two cases. Radiology, 163, 551-553. KRANSDORF, M. J., JELINEK, J. S., MOSER, R. P., JR, U T Z , J. A., BROWER, A. C , HUDSON, T. M. & BERREY, B. H., 1989a.
Soft tissue masses: diagnosis using MR imaging. American Journal of Roentgenology, 153, 541-547. KRANSDORF, M. F., MOSER, R. P., JR, JELINEK, J. S., WEISS,
& ZACHARIAS, C , 1986. Soft-tissue masses of the locomotor system: comparison of MR imaging with CT. Radiology, 160, 125-133. PETTERSSON, H., GILEPSY, T., HOUNLIN, D. J., ENNEKING, W. F., SPRINGFIELD, D. S., ANDREW, R. E., SPANIER, S. &
STONE, R., 1987. Primary musculoskeletal tumors: examination with MR imaging compared with conventional modalities. Radiology, 164, 237-241. SUNDARAM,
M., MCGUIRE,
M. H. & HERBOLD,
D. R., 1988.
Magnetic resonance imaging of soft tissue masses: an evaluation of fifty-three histologically proven tumors. Magnetic Resonance Imaging, 6, 237-248. SUNDARAM, M., MCGUIRE, M. H. & SCHAJOWICZ, F., 1987.
Soft-tissue masses: histological basis for decreased signal (short 7J) on ^-weighted MR images. American Journal of Roentgenology, 148, 1247-1250. TOTTY, W. G., MURPHY, W. A. & LEE, J. K. T., 1986. Soft
tissue tumors: MR imaging. Radiology, 160, 135-141.
S. W., BEUTOW, P. C. & BERRY, B. H., 1989b. Intramuscular
20
The British Journal of Radiology, January 1992
Tumour and tumour-like conditions of the soft tissue: magnetic resonance imaging features differentiating benign from malignant masses By George Hermann, MD, Ibrahim Fikry Abdelwahab, MD, Theodore T. Miller, MD, "Michael J. Klein, MD and tMichael M. Lewis, MD Departments of Radiology, *Pathology and tOrthopedics of The Mount Sinai Medical Center, CUNY, New York, USA {Received 5 March 1991 and in revised form 21 June 1991, accepted 23 July 1991)
Keywords: Soft-tissue tumour, MRI, Benign, Malignant
Abstract. 49 primary soft-tissue masses were studied by magnetic resonance imaging (MRI) in an effort to differentiate benign from malignant lesions. There were 24 benign and 25 malignant tumours. Images were evaluated for the presence of low-signal septation, changes in pattern of homogeneity, signal intensity, and margin definition and shape. 20 of the 25 malignant tumours (80%) demonstrated low-signal septation, and two of the 24 benign tumours (8%) also demonstrated this feature. Malignant tumours tended to change pattern from being homogeneous on Trweighted images to being heterogeneous on ^-weighted images (72%). Benign tumours maintained their pattern on both sequences (67%). Only three of the 24 benign tumours (12.5%) changed from being homogeneous to heterogeneous. Our findings suggest that tumours which change pattern, from homogeneous on ^-weighted images to heterogeneous on 7^-weighted images, and tumours which have low-signal septations are likely to be malignant. Moreover benign lesions tend to have well defined margins, and some benign masses have characteristic appearances that aid in their differentiation from malignant processes.
Since the introduction of magnetic resonance imaging (MRI), its diagnostic efficacy has been compared with computed tomography (CT). While the superiority of MRI for visualizing the skeleton and surrounding soft tissues has been documented (Petasnick et al, 1986; Totty et al, 1986; Demas et al, 1988; Sundaram et al, 1988; Kransdorf et al, 1989a; Berquist et al, 1990), its ability to provide a histological diagnosis has not been established. There are no established features with which to predict or distinguish malignant from benign lesions, although tumour signal intensity, margination, homogeneity and the presence of calcification have been used by some authors (Totty et al, 1986; Sundaram et al, 1988; Kransdorf et al, 1989a; Berquist et al, 1990). By modifying these criteria to include tumour morphology, presence of septation, changes in pattern of homogeneity, signal intensity and margin definition, we have attempted to identify certain trends that may be useful in distinguishing benign from malignant tumours. Materials and methods
MR images of 49 primary soft-tissue masses in 49 patients were reviewed. The patients, 26 males and 23 females, ranged in age from 6 to 75 years. The overall Address correspondence to George Hermann, MD, Box 1234, Department of Radiology, Mount Sinai Medical Center, One Gustave L. Levy Place, New York, NY 10029-6574, USA. 14
average age of the patients was 42 years in the benign and 38 years in the malignant groups. In most cases, the MRI studies were performed in our institution, but in several instances the patients were referred to us for second opinions or for surgery. All patients underwent surgery at our institution. The MRI scans were obtained on an Elscint 5000 0.5 T, Siemens Magnetom 1.0 T or a Diasonic 0.3 T system, but we did not evaluate images for possible differences in field strength. Multisection and multiplanar techniques were applied in the majority of cases. Spin-echo Tr and ^-weighted pulse sequences were 450-700/20-30 TR/TE (repetition time/echo time) and 1800-2200/80 TR/TE, respectively. The images were obtained in a combination of coronal, sagittal and/or axial planes. Section thickness varied from 4 to 10 mm. Surface coils were used appropriate to the site being imaged. In all cases diagnosis was confirmed with the pathological specimen except in three cases in which a radiological diagnosis could be made with certainty (e.g. two small lipomas and one popliteal cyst), obviating the need for operation. The MR images were reviewed by two radiologists together (G.H. and T.T.M.) and were evaluated with regard to: (1) shape, (2) morphology, (3) pattern, (4) signal intensity, and (5) definition of margins. The shape of the tumour was considered either round-oval or irregular. Morphology refers to the presence of lowsignal septations which traversed the lesion and divided The British Journal of Radiology, January 1992
MRI of soft-tissue tumours
it into lobules. Tumours could demonstrate low-signal septations while maintaining an overall shape of roundoval, but an external contour of lobulation was not possible without septation, and such a shape by definition was irregular. The pattern was described either as homogeneous if all sections had uniform signal, or heterogeneous if the lesion contained varied signal. Any change in pattern between ^-weighted images (T^-WI) and %-weighted images (7J-WI) was noted. The evaluation of signal intensity was based on comparison with adjacent muscles in both 7J-WI and 7^-WI and was graded as low, equal or high. Tumour margins were evaluated as ill defined, partially or well defined according to their relationship to the surrounding tissue: ill defined when the borders were confluent with the surrounding tissue; partially defined if at least part of the border of the mass was confluent with the surrounding tissue; and well defined when the entire tumour was sharply demarcated from the adjacent tissues. Results
Of the 49 soft-tissue masses, 24 were benign and 25 malignant (Tables I, II).
In the benign group, eight occurred in the thigh, four in the pelvis, three in the shoulder region, three in the calf, two in the forearm, two in the popliteal fossa, one in the ankle and one in the foot. In the malignant group, 10 occurred in the thigh, four in the popliteal fossa, three in the calf, three in the back, two in the foot, two in the elbow, and one in the hip (Fig. I). The size of the tumours ranged from 2 to 20 cm, with an average of 8 cm in greatest diameter. Shape and morphology The overall configurations of the mass in both benign and malignant lesions were equally divided. 27 were round-oval (14 malignant, 13 benign) and 22 irregular (11 malignant, 11 benign). Low-signal septation was present in 20 of 25 malignant lesions (80%) (Fig. 2) and in only two of 24 benign masses (8%). 18 of the 20 (90%) malignant septated masses can be broadly classified histologically as spindle cell tumours: 10 malignant fibrous histiocytomas, three synovial sarcomas, one malignant Schwannoma, one undifferentiated spindle cell sarcoma, one liposarcoma, one haemangiopericytoma and one extraosseous osteosarcoma. The remaining two septated malignancies were extraosseous
Table I. Benign conditions (n = 24) Tumour Pattern Lipoma Lipoma Lipoma Lipoma Lipoma Atypical lipoma Neurofibroma Neurofibroma Neurofibroma PVNS PVNS Ganglion Ganglion Desmoid Desmoid Haematoma Haematoma Abscess Abscess Infection Popliteal cyst Myxoma Haemangioma Benign Schwannoma
Septation
T2-WI
Hm Ht Hm Hm Hm Hm Ht Hm Ht Hm
Ht Ht Hm Ht Hm Ht Ht Ht Ht Ht Ht Hm Ht Hm
SI
Definition
Pattern
SI
Definition
Hm Ht Hm Hm Hm Hm Ht Hm Hm Ht Ht Hm Ht Ht Hm Ht Ht Hm Hm Hm Ht Hm Ht Ht
Pattern: Hm, homogeneous; Ht, heterogeneous. SI: (—) SI Jess than adjacent muscle, (=) SI equa) to adjacent muscle, (-h) SI greater than adjacent musde. Definition: (+) poorly defined margins, (+ +) partially well defined, (+ + +) well defined. Septation: (—) none, ( + ) seen on T2-WI only, ( + ) present on Tr and r 2 -WI. Vol. 65, No. 769
15
G. Hermann, I. F. Abdelwahab, T. T. Miller, M. J. Klein and M. M. Lewis
Ewing's sarcoma and an alveolar soft-part sarcoma. The two benign septated masses consisted of an abscess and an atypical lipoma. None of the benign spindle cell tumours (e.g. neurofibroma, benign Schwannoma, desmoid) showed any evidence of septation.
pattern. Of these 23 masses, 16 (70%) were benign. Nine tumours remained homogeneous, of which eight (89%) were benign. 14 tumours remained heterogeneous: eight benign (57%) and six malignant (43%).
Signal intensity Pattern Compared with adjacent muscle on 7^-WI, 10 of 25 In the malignant group, 18 of the 25 tumours (72%) malignant tumours showed a high signal, one a low changed pattern between Tx- and T^-WI: 16 changed signal and 14 of 25 as isointense signal. However, all from homogeneous to heterogeneous (Fig. 3) and two these tumours showed high signal on TrWI. Overall, 15 vice versa. Seven of 25 (28%) did not change pattern: of 25 (60%) changed signal intensity. one remained homogeneous and six remained heteroThe benign tumours had high signal intensity in 14 of geneous. In contrast, only eight of the 24 benign 24 lesions on TJ'-WI, isointensity in six and low signal tumours (33%) changed pattern (three homogeneous to intensity in four. 21 of the 24 exhibited high signal on heterogeneous and five vice versa). Sixteen of the 24 ZJ-WI. The three benign tumours that were not intense benign tumours (67%) did not change, with eight on ZJ-WI included two pigmented villonadular synovitis remaining homogeneous and eight heterogeneous on (PVNS), which were isointense on both Tr and 7J-WI, both sequences. and one desmoid tumour which was of low signal Combining the two groups, 26 of the 49 masses intensity on both. Overall, only seven of 24 (29%) changed pattern, and eighteen of the 26 (69%) were changed signal intensity. Combining the results, 18 malignant. More significantly, 19 of these 26 changed tumours were isointense on 7^-WI and bright on 7^-WI, from a homogeneous pattern on TJ'-WI to a hetero- of which 14 (78%) were malignant. geneous one on 7^-Wl, of which 16 (84%) were malignant. The remaining seven of 26 tumours changed from Margin definition heterogeneous to homogeneous, of which only two 10 malignant lesions were well defined, 10 were (29%) were malignant and five (71%) were benign. partially defined and one poorly defined on both Tr and In contrast, 23 of the 49 masses did not change 7^-WI. In addition, two changed from partially to well Table II. Malignancies (n = 25) Tumour
MFH MFH MFH MFH MFH MFH MFH MFH MFH MFH MFH MFH Synovial sarcoma Synovial sarcoma Synovial sarcoma Malignant Schwannoma Malignant Schwannoma Extraosseous Ewing's sarcoma Extraosseous Ewing's sarcoma Spindle cell sarcoma Liposarcoma Extraosseous osteosarcoma Lymphoma Haemangiopericytoma Alveolar soft-part sarcoma
7;-wi
Septation
T2-WI
Pattern
SI
Definition
Pattern
SI
Definition
Ht Hm Hm Hm Hm Hm Hm Hm Hm Hm Ht Ht Hm Hm Hm Ht Hm Hm Hm Ht Ht Ht Hm Hm Ht
=
+
+++ +++ +++ ++ +++ +++ ++ ++ +++ ++ +++ ++ +++ ++ +++ ++ ++ +++ ++ +++ ++
Hm Ht Ht Ht Ht Ht Ht Ht Ht Ht Hm Ht Ht Ht Ht Ht Ht Ht Ht Ht Ht Ht Ht Hm Ht
+ + + + + + + + + + + + + + +
+
= = =
+ + = + = = + + + — = + = = + + = = = +
+
++ ++
+ + + + + + + + + +
+++ +++ +++ ++ +++ +++ +++ ++ ++ +++ +++ ++ +++ ++ +++ ++ ++ +++ ++ +++ ++ +++ ++ ' ++
±
+
+
±
+ + + +
± — +
±
+ + — +
± — + + + +
±
MFH = Malignant fibrous histiocytoma. See Table I for key to symbols.
16
The British Journal of Radiology, January 1992
MRI of soft-tissue tumours
istic appearances of certain benign masses further aids differentiation. BENIGN MALIGNANT Although the shape of the tumour is non-specific, the presence of low-signal septation strongly suggests malignancy. This feature has been mentioned in other reports of soft-tissue sarcomas but its relationship to malignancy has not been emphasized. A septated appearance was described by London et al (1989) in their series on liposarcomas. Kransdorf et al (1989a) in their paper on soft-tissue tumours showed similar septation in illustrations of myxoid liposarcoma and neurofibrosarcoma but did not discuss this as a feature of malignancy. Similarly, Demas et al (1988) illustrated a soft-tissue sarcoma with low-signal septation. In our series, this septated appearance was apparent in 20 out of 25 of the malignant tumours (80%), in contrast with only two out of 24 of the benign tumours (8%). While it is not surprising that 18 out of 20 septated malignant tumours were spindle cell types since they are of mesenchymal origin, it is noteworthy that none of the benign spindle cell tumours in our series showed this septated morphology. Malignant fibrous histiocytoma (MFH) was the most common tumour in our series. 10 of the 12 cases of MFH had a lobulated appearance both on Tr and 7J-WI, consisting of irregular nodular masses separated by fine low-signal-intensity bands. This feature was also seen in our three synovial sarcomas. Septation in liposarcomas has previously been described by London et al (1989), who believed that septation represented a more differentiated fatty component in the tumour. Figure 1. Distribution of primary soft-tissue tumours (24 Alternatively these low-signal septa may represent benign, 25 malignant). streaks of fibrous tissue or regions of relative acellularity. One can speculate that non-uniform stages in the growth of tumours may be separated by fibrous tissue defined, one from poorly to well defined and one from and show a pattern of septation. Although this appearance is highly suggestive of well defined to partially defined. In the benign category, 16 tumours were well defined, three partially defined malignancy, it is not infallible. Two benign tumours, an and two poorly defined on both Tr and 7J-WI. Two abscess and an atypical lipoma, showed this appearance changed from partially to well defined, and one from and were mistakenly diagnosed pre-operatively as well to partially defined. In the benign category, 16 malignant. In a recent report, an intramuscular myxoma tumours were well defined, three partially defined and was mistakenly but reasonably diagnosed pre-operatwo poorly defined on both Tx- and 7J-WI. Two changed tively as a liposarcoma because of similar septation from partially to well defined, and one from well to (Kransdorf et al, 1989a). In the absence of septation we found no significant partially defined. patterns that differentiate benign nerve sheath tumours from their malignant counterparts. In all instances they Discussion It is generally accepted that soft-tissue tumours are were well demarcated from the neighbouring structures. One benign Schwannoma and two malignant better demonstrated with MRI than with CT or ultrasound (Petasnick et al, 1986; Totty et al, 1986; Chang et Schwannomas appeared homogeneous on 7J-WI and al, 1987; Demas et al, 1988; Sundaram et al, 1988; heterogeneous on ZJ-WI. None of the benign Kransdorf et al, 1989a; Berquist et al, 1990). The major Schwannomas showed evidence of septation. Thus, advantages of MRI are its ability to display contrast while the absence of septations does not exclude a between tissues and its capacity for multiplanar malignancy, their presence is strongly against a benign imaging. Recent investigations attempting to use MRI process. The second criterion we found useful was whether the to distinguish benign tumours from malignancies have been inconclusive (Pettersson et al, 1987; Sundaram et tumour was homogeneous or heterogeneous and if there al, 1988; Kransdorf et al, 1989a). Our study suggests was any change in pattern that occurred from TJ'-WI to that tumour septation and changes in homogeneity may j;-WI. The pattern of 18 of the 25 malignancies (72%) be useful in predicting malignancy, and that character- changed between 7J-WI and JJ-WI, with 16 of the 18 Vol. 65, No. 769
17
G. Hermann, I. F. Abdelwahab, T. T. Miller, M. J. Klein and M. M. Lewis
Figure 2. Malignant fibrous histiocytoma in a 72-year-old woman, (a) Sagittal image shows a slightly higher signal intensity mass behind the distal femur on 7J-WI (SE 600/30). (b) On ZJ-WI (SE 1800/100) the mass shows high signal, is heterogeneous and well defined. Note the low-signal-intensity septa (arrows) causing lobulation on both Tr and 7|-WI. (c) Malignant fibrous histiocytoma. Low power shows adjacent myxoid areas with cellular fibrous septa between them corresponding to an area of decreased signal intensity on 7J-WI (centre) ( x 11).
homogeneous on 7^-WI and heterogeneous on 7^-WI. Only three of 24 benign tumours (12.5%) behaved similarly. Thus it appears that the change from a homogeneous pattern to a heterogeneous one is of prognostic significance. The majority of benign tumours (67%) maintained their original pattern of homogeneity on both Tr and JJ-WI sequences. Moreover, of all the tumours in our series, nine were homogeneous on both sequences, of which eight (89%) were benign. If one examines only T2-Wl, our data agree with that of Petasnick et al (1986) who concluded that on IJ-WI, benign lesions tend to be more homogeneous than 18
malignant tumours. Indeed, of our 49 tumours, 18 were homogeneous on 7^-WI, of which 13 (72%) were benign. However, we emphasize that it is the change of the tumour from the 7^-WI pattern to the 7^-WI pattern that is important and not the isolated appearance of one without the other. Therefore, in each case both sequences should be carefully evaluated. The criteria of signal intensity and margin definition were generally not helpful. Almost all soft-tissue tumours are of high signal on 7^-WI, and lesions that demonstrated high signals in both sequences were almost equally divided between benign and malignant. The British Journal of Radiology, January 1992
MRI of soft-tissue tumours
Figure 3. Liposarcoma of the popliteal region in an 18-year-old male, (a) Sagittal 7J-WI (600/30) shows septation. (b) Coronal 7^-WI (2100/80) appears heterogeneous with exterior lowsignal-intensity septa that divide the mass into smaller lobules (arrows).
However, those tumours that were isointense on 7^-WJ and of high signal on 7J-WI were predominantly malignant. In addition, certain characteristic benign patterns have been described with which the findings in our cases concur; for example, PVNS is isointense on 7J-WI and 7^-WI (Kottal et al, 1987; Jelinek et al, 1989) and ganglia (Totty et al, 1986; Petasnick et al, 1986) and myxomas (Kransdorf et al, 1989) are low on TrWl and high on 7^-WI. In our series, desmoids gave a low signal on both 7/-WI and 7^-WI. These characteristic appearances are important because they aid in the recognition of these processes as benign. For example, despite the infiltrative, locally aggressive nature of desmoid tumours mentioned by Berquist et al (1990), the persistent low signal intensity on both 7J-WI and 7J-WI should strongly suggest the diagnosis. The underlying pathological explanation for these appearances is the presence of fluid in ganglia and myxomas, dense fibrotic tissue in desmoids, and intra and extracellular haemosiderin in PVNS. Similarly, decreased signal owing to haemosiderin deposition has been observed in cases of repeated transfusion for thalassaemia (Brasch et al, 1984). In one case of MFH in which myxomatous tissue was predominant, the tumour was isointense on TJ'-WI and of homogeneously high signal on 7J-WI (Fig. 4). Sundaram et al (1987) found that tumours which are relatively acellular and contain a large amount of collagen show relatively low signal intensity on 7J-WI, and concluded Vol. 65, No. 769
that the cellularity of a tumour rather than the histological diagnosis influences the MRI signal on 7J-WI. Petasnick et al (1986) and Berquist et al (1990) suggest that benign masses show well defined borders. Our data generally agree with this finding, although in our series it was not reliable enough to be useful in distinguishing benign from malignant lesions. In summary, our findings suggest that tumours which change in homogeneity from 7J-WI to T2-Wl, and especially those that alter from a homogeneous pattern to a heterogeneous pattern, are more likely to be malignant. Lesions that do not change, and especially those that remain homogeneous, are most likely benign. In addition, tumours with septation are most likely malignant, and probably of spindle cell origin. Where these two criteria conflict (i.e. a lobulated tumour that does not change pattern), we believe that the presence of lowsignal septation is a stronger predictor of malignancy than a change in pattern. References BERQUIST, T. H., EHMAN, R. L., KING, B. F., HODGMAN, C.
G.
& ILSTRUP, D. M., 1990. Value of MR imaging in differentiating benign from malignant soft-tissue masses: study of 95 lesions. American Journal of Roentgenology, 155, 1251-1255. BRASCH, R. C , WESBEY, G. E., GOODING, C. A. & KOERPER,
M. A., 1984. Magnetic resonance imaging of transfusional hemosiderosis complicating thalassemia major. Radiology, 150, 767-77'1.
19
G. Hermann, I. F. Abdelwahab, T. T. Miller, M. J. Klein and M. M. Lewis
I
(b)
Figure 4. 52-year-old woman. Mass on the dorsal aspect of the foot in sagittal plane showing (a) homogeneous low signal intensity on 7J-WI (600/20). (b) High signal on 7^-WI (2500/80). (c) Low-grade myxoid malignant fibrous histiocytoma. High magnification reveals stellate and spindle-shaped cells with long cytoplasmic processes and atypical nuclei in a myxoid stroma (x88). myxoma: MR features. Journal of Computer Tomography, 13, 836-839.
Assisted
CHANG, A. E., MATORY, Y. L., DWYER, A. J., HILL, S. C , GIRTON, M. E., STEINBERG, S. M., KNOP, R. H., FRANK, J. A., HYAMS, D., DOPPMAN, J. L. & ROSENBERG, S. A., 1987.
LONDON, J., KIM, E. E., WALLACE, S., SHIRKHODA, A., COAN, J.
Magnetic resonance imaging versus computed tomography in the evaluation of soft tissue tumors of the extremities. Annals of Surgery, 205, 340-348.
& EVANS, H., 1989. MR imaging of liposarcomas: correlation of MR features and histology. Journal of Computer Assisted Tomography, 12, 832-835.
DEMAS, B. E., HEELAN, R. T., LANE, J., MARCOVE, R., HAJDU,
PETASNICK, J. P., TURNER, D. A., CHARTERS, J. R., GITELIS, S.
S. & BRENNAN, M. F., 1988. Soft tissue sarcomas of the extremities: comparison of MR and CT in determining the extent of disease. American Journal of Roentgenology, 150, 615-620. JELINEK, J. S., KRANSDORF, M. J., U T Z , J. A., BERREY, B. H., THOMSON, J. D., HEEKIN, R. D. & RADOWICH, M. S., 1989.
Imaging of pigmented villonodular synovitis with emphasis on MR imaging. American Journal of Roentgenology, 152, 337-342. KOTTAL, R. A., VOGLER, J. B., Ill, MATAMOROS, A., ALEXANDER, A. H. & COCKSON, J. L., 1987. Pigmented
villonodular synovitis: a report of MR imaging in two cases. Radiology, 163, 551-553. KRANSDORF, M. J., JELINEK, J. S., MOSER, R. P., JR, U T Z , J. A., BROWER, A. C , HUDSON, T. M. & BERREY, B. H., 1989a.
Soft tissue masses: diagnosis using MR imaging. American Journal of Roentgenology, 153, 541-547. KRANSDORF, M. F., MOSER, R. P., JR, JELINEK, J. S., WEISS,
& ZACHARIAS, C , 1986. Soft-tissue masses of the locomotor system: comparison of MR imaging with CT. Radiology, 160, 125-133. PETTERSSON, H., GILEPSY, T., HOUNLIN, D. J., ENNEKING, W. F., SPRINGFIELD, D. S., ANDREW, R. E., SPANIER, S. &
STONE, R., 1987. Primary musculoskeletal tumors: examination with MR imaging compared with conventional modalities. Radiology, 164, 237-241. SUNDARAM,
M., MCGUIRE,
M. H. & HERBOLD,
D. R., 1988.
Magnetic resonance imaging of soft tissue masses: an evaluation of fifty-three histologically proven tumors. Magnetic Resonance Imaging, 6, 237-248. SUNDARAM, M., MCGUIRE, M. H. & SCHAJOWICZ, F., 1987.
Soft-tissue masses: histological basis for decreased signal (short 7J) on ^-weighted MR images. American Journal of Roentgenology, 148, 1247-1250. TOTTY, W. G., MURPHY, W. A. & LEE, J. K. T., 1986. Soft
tissue tumors: MR imaging. Radiology, 160, 135-141.
S. W., BEUTOW, P. C. & BERRY, B. H., 1989b. Intramuscular
20
The British Journal of Radiology, January 1992
