Clinical Radiology (1992) 46, 311-317
Review of the Use of MRI in Soft Tissue Lesions S. J. A R M S T R O N G , C. J. W A K E L E Y , P. R. G O D D A R D and I. W A T T
Department of Clinical Radiology, United Bristol Healthcare NHS Trust, Bristol The MR images of 34 patients with soft tissue lesions were retrospectively evaluated to assess the accuracy of the technique in distinguishing benign from malignant lesions, and to assess the usefulness of various criteria in making this distinction. The overall sensitivity for the detection of malignancy was 75% with a specificity of 94%. Size of lesion was found to be a good criterion in predicting malignancy, lesion margin and signal intensity were less useful. The tissue type was determined in a few instances where signal characteristics were typical, notably lipomas and neural tumours, but this was not reliable and in most lesions the tissue of origin cannot be determined on MR imaging and biopsy is necessary.
Armstrong, S.J., Wakeley, C.J., Goddard, P.R. & Watt, I. (1992). Clinical Radiology 46, 311-317. Review of the Use of MRI in Soft Tissue Lesions
Accepted for Publication 19 July 1992
J
Magnetic resonance imaging (MRI) has been shown to be superior to CT in demonstrating soft tissue lesions, owing to the better contrast resolution and the ability to image in any plane [1]. MRI permits reliable identification of the spatial boundaries of tumour masses, encasement and displacement of major neurovascular bundles and the extent of joint involvement [2]. Also it can depict intralesional necrosis and haemorrhage [3]. The distinction between benign and malignant soft tissue lesions is assessed by applying generally accepted criteria (Table 1) but a difference of opinion exists over the usefulness and accuracy of these features [4,5]. Tissue characterization by M R I is limited, with a few notable exceptions. Measurement of relaxation times is technically difficult and has not proved to be of value in differentiating turnout types [2]. Thus the precise role of MRI in the investigation of soft tissue tumours remains unclear. In an attempt to clarify the position, patients referred to the Bristol MRI Centre for this indication were reviewed retrospectively to determine how reliably MRI could distinguish: 1 Benign from malignant lesions, 2 Benign/malignant turnouts from non-tumour lesions (e.g. inflammatory/post-traumatic), 3 Particular turnout types.
P A T I E N T S AND M E T H O D S The magnetic resonance (MR) images of 34 patients (19 males and 15 females) with a suspected soft tissue lesion referred to the Bristol MRI Centre between 1987 and 1991 were reviewed retrospectively. The age range was from 3 to 82 years (average age 40). All scans were obtained with a 0.5T superconducting magnet. Typical scan sequences included Tl-weighted spin echo (SE 500/ 26), and short tau inversion recovery (STIR) sequences, (1500/100/30). Twenty out of 34 patients also had T2weighted (SE 1500/80) images performed. Gadolinium (Gd-DTPA) enhancement was not used in any of the patients in this series. Standard sagittal, coronal and transverse scan planes were used. Scans had been reported previously by four radiologists, one of whom (IW) subsequently reviewed all the images without prior knowledge of the history, the diagnosis made or the results of any other imaging studies. Histological diagnosis was available in all 34 patients, obtained by biopsy or excision. Twenty-three patients had benign lesions including both benign neoplasms and inflammatory and tumour-like masses. Eleven had malignant lesions, nine primary, including two with NonHodgkin's lymphoma (being the initial presentation in both), and two secondary malignancies. Specific diagnoses are listed in Table 2. Presentation was with a localized (24) or diffuse (10) swelling, usually in a limb.
Table 1 -Distinctive MRI features in benign and malignant soft tissue lesions*
Benign
Malignant
Size Margin Signal intensity
< 3 cm Well-defined Homogeneous
Distribution
Confined to soft tissues
> 3 cm Ill-defined/irregular Inhomogeneous/intensity changes in surrounding tissues Infiltration of adjacent muscle or bone
* See Berquist et al. [4]. Correspondence to: Dr S. J. Armstrong, Department of Clinical Radiology, United Bristol Healthcare N H S Trust, Bristol BS2 8HW.
RESULTS Using the criteria in Table 1, 24 of these 34 lesions were judged by MRI to be benign and 10 to be malignant. Of the 24 lesions thought to be benign three were subsequently shown to be malignant lesions pathologically: one was a myxoid liposarcoma and two were metastatic deposits. Two of the 10 tumours suspected of being malignant were subsequently shown to be benign: a benign haemangioma was suspected on M R to be a rhabdomyosarcoma (Fig. 1) and a ganglion cyst of the
312
CLINICAL RADIOLOGY
knee was t h o u g h t on M R to be infiltrative and therefore possibly malignant in nature (Fig. 2). The sensitivity for the detection o f malignancy on M R I was 75% with a specificity of 94%.
T I S S U E C H A R A C T E R I Z A T I O N (Table 2) (a) Benign Lesions The diagnosis or differential diagnosis made by M R I was consistent with the histological result in all except five cases. One lesion t h o u g h t to be an atypical lipoma was f o u n d to be pigmented villonodular synovitis (PVNS), one suspected vascular m a l f o r m a t i o n an area o f scar tissue, and one suspected bursa a collection o f fibrotic lymph nodes. A n area o f fat necrosis in one patient and n o d u l a r fasciitis in another were correctly diagnosed as being benign lesions, although the precise nature of the lesions could not be determined from the M R images. (b) Malignant Lesions
(a)
O f the 11 pathologically confirmed malignancies the M R images correctly suggested a malignant cause in eight cases and incorrectly suggested a benign cause in three cases: one myxoid liposarcoma had M R appearances t h o u g h t to be typical o f a benign s c h w a n n o m a (Fig. 3), and two metastases were incorrectly interpreted as being due to h a e m a t o m a (Fig. 4). In one patient with N o n H o d g k i n ' s l y m p h o m a and another with malignant melan o m a both a turnout mass and involved lymph nodes were demonstrated (Table 3). All o f the wrongly diagnosed lesions had T l W , T 2 W and S T I R sequences performed, with the single exception o f the patient whose biopsy yielded scar tissue who had T 1 W and S T I R sequences only. (b) Fig. 1 A 14-year-old boy with pain and swelling in the thigh has an illdefined lesion of intermediate heterogeneous signal intensity in vastus lateralis on Tl-weighted image (a). On (b), a proton density-weighted image, the lesion is of high signal, again with ill-defined margins. Diagnosis: benign haemangioma.
Table 2 - Final diagnosis (34 patients) Benign lesions
Bursa Fibromatosis Fat necrosis Fibrotic lymph nodes Ganglion cyst Haemangioma Lipoma Lymphangioma Meniscal trauma Neuroma Nodular fasciitis PVNS Scar tissue Total
2 2 1 1 1 2 5 1 1 4 1 1 1 23
Malignant lesions
Alveolar cell sarcoma Fibrosarcoma Liposarcoma MFH Malignant melanoma NHL Rhabdomyosarcoma Total
1 1 2 2 2 2 1 11
Size (Table 4) Of the benign lesions 43% were less than 3 cm in diameter and 78% less than 5 cm. The benign lesions greater than 5 cm in diameter were either lipomas or fibromatosis. This c o m p a r e d with 55% o f malignant lesions that were greater than 5 cm and 82% over 3 cm. Only two (18%) o f the malignant lesions were less than 3 cm and these were b o t h metastatic malignant m e l a n o m a (Fig. 5). The malignant lesions greater than .5 cm in diameter were all sarcomas, including the m y x o i d liposarcoma. Margin and Signal Characteristics Benign lipomas tended to be well-defined and wellencapsulated. All but two were superficial, one o f the deep lesions was infiltrating between muscle fibres. All had a high signal on T 1 W images and low signal on S T I R images (Fig. 6).
MRI IN SOFT TISSUE LESIONS
313
on S T I R images. Neoplastic lesions tended to be inhomogeneous with a mixed intermediate or low signal intensity on T1W images and a high (but heterogeneous) signal on STIR. Six of the 11 malignant lesions had ill-defined margins, the margins of the other five being well-defined. Most but not all had heterogeneous signal intensity, three lesions had homogeneous signal: the myxoid liposarcoma (q.v.) and both cases of N o n - H o d g k i n ' s lymphoma.
DISCUSSION
(a)
M R I has been shown to be good at detecting soft tissue lesions, defining their extent and their relationships to surrounding structures. The distinction of benign from malignant has been proposed from consideration of the following:
(a) Size Berquist et al. [4] found that the size o f a tumour was a useful discriminator, with a specificity and accuracy of 90%. In our series almost a quarter of benign lesions and no primary malignancies were less than 3 cm in diameter, and 87% of malignancies were greater than 5 cm in diameter. Thus, size proved a useful but not reliable discriminator, in particular the two metastases in our group were less than 3 cm in diameter.
(b) Margin (b) Fig. 2 A 62-year-old man with a lump in the knee which is shown to be cystic and multiloculated with ill-defined margins laterally on proton density-weighted images (a). The margins are again ill-defined on STIR image (b). Diagnosis: benign ganglion cyst.
Benign neural lesions such as schwannomas, neurofibromas and neurilemmomas tended to be well-defined lesions and in close relationship to enlarged nerves. They all had a signal intensity similar to that of muscle on TI W images and a high signal on T2W and STIR images (Fig. 7). Two of the four neuromas in the present series had a low signal margin and two had a whorled or laminated appearance with low signal areas centrally. The myxoid liposarcoma mistakenly diagnosed on M R I as a benign schwannoma had the former appearance. Vascular lesions such as haemangiomas, lymphangiomas and haematomas tended to be more diffuse, illdefined lesions of heterogeneous signal intensity. They were either isointense or of slightly higher signal than muscle on T1W images and of slightly increased (but not the very bright signal seen with neural lesions) on S T I R images. In one of the two haemangiomas in the present series large feeding blood vessels could be identified (seen as areas of signal void). Joint-related lesions tended not to have any characteristic features but the diagnosis was aided by the typical site in close relationship to a joint. Cystic lesions such as bursae had a low signal on T1W images and high signal
In the series reported by Berquist et al. [4], benign lesions typically had well-defined or slightly irregular margins whereas this was the case with only 15% of malignancies. However, others have found this criterion unreliable [5] as some benign lesions are surrounded by oedema making them ill-defined. Aggressive fibromatosis is an example of a benign but locally invasive tumour that can involve bone, a feature usually thought to indicate malignancy [4]. In the present series, lesion margin was an unreliable sign as only half of the malignant lesions had ill-defined margins, the other half being well-defined.
(c) Signal Intensity and Homogeneity It is unusual for a malignant lesion to show homogeneous signal intensity [4] but heterogeneous signal intensity is not specific for malignancy and can be seen in m a n y benign lesions [5]. I f uniform signal intensity is used as a criterion of benignity then false positive diagnoses of malignancy will occur with necrotic tumours and haemangiomas. Increased signal intensity in skeletal muscle adjacent to a soft tissue mass on T2W images is an important indicator of either oedema or tumour invasion [6] but some benign conditions such as infection, myositis or haematoma, e.g. post-surgery, can produce similar M R I appearances. Most soft tissue masses have a low signal intensity, equal to or lower than muscle, on T1W images and a high signal intensity on T2W images [1]. Notable exceptions are lesions containing fat or haemorrhage which have a
314
CLINICALRADIOLOGY
(a)
(b)
(e) Fig. 3 - A 47-year-oldwoman with swellingin the left calf shows a uniform ovoid lesion isointensewith muscleon Tl-weighted image (a) (open arrow). The lesion has high signal on a T2-weightedimage (b) and is homogeneousand well-definedusing STIR sequence (c). Diagnosis: myxoid liposarcoma.
high signal on T1W images, e.g. lipomas, liposarcomas, haemangiomas, non-acute haematomas and haemorrhage into a pre-existing mass [7]. The signal strength of haemorrhage depends on its age, acute haemorrhage (within 24 tO 36 h) having a low/medium signal on T1W, and subacute haemorrhage a high signal due to shortening o f T 1 by the paramagnetic effect of methaemoglobin. Benign lipomas are probably the most characteristic soft tissue lesion on MRI, having a high signal on T1W and T2W and very low signal on S T I R (which suppresses fat): Soft tissue haemangiomas can be further distinguished by a high signal striated/separated configuration on certain spin-echo sequences [8]. This is thought to correlate histologically with vascular channels of varying sizes separated by fibrofatty septa. Tubular areas of signal void in relation to a soft tissue abnormality are due to
dilated tortuous blood vessels and are characteristic of arteriovenous malformations [9]. The appearance is similar on both T1W and T2W images, the flow void effect being unaltered by the T R values, presumably due to the high velocity of blood flow within arteriovenous malformations. This appearance was noted in a superficial haemangioma in the present series. Peripheral nerve sheath tumours usually have an ovoid or fusiform shape and are of intermediate to moderately high signal intensity on T1W images and high signal on T2W images [10]. They tend to exhibit variable inhomogeneity of signal on both sequences, whether benign or malignant, which is thought to be d tt¢ to coexistent areas of hypocellularity and hypercellularity, and differentiating benign from malignant turnouts is unreliable. In the present series peripheral nerve sheath tumours had a
MRI IN SOFT TISSUE LESIONS
315
Fig. 4 - A previously fit 77-year-old woman with a lump in the right calf which was isointense with muscle on T 1 weighting (not illustrated) and high signal on a STIR image. The peripheral site and close relationship to dilated veins suggested an area of haematoma. Diagnosis: vascular amelanotic malignant melanoma. Fig. 5 - A 67-year-old woman with known malignant melanoma and a lump in the thigh showing a small well-defined lesion with a high signal on a sagittal T2-weighted image. Diagnosis: metastatic melanoma within a thrombosed vein. Table 3 - M R I diagnosis vs pathological findings (34 patients)
MRI benign MRI malignant
Path benign
Path malignant
21 2
3 8
Table 4 - M R I estimation o f size (diameter)
Small ( < 3 cm)
Intermediate (3 5 cm)
Large ( > 5 cm)
Bursa x 2 Alveolar cell sarcoma Fibrotic lymph nodes Ganglion cyst Haemangioma Haemangioma Lipoma Lipoma x 2 Meniscal trauma Lymphangioma Metastatic MM × 2 MFH Neuroma x 3 Neuroma Nodular fasciitis NHL PVNS Sear tissue
Fibromatosis x 2 Fat necrosis Fibrosarcoma Lipoma × 2 Liposarcoma x 2 MFH NHL Rhabdomyosarcoma
Total = 13 (11B, 2M)
Total = 11 (5B, 6M)
Total = 10 (7B, 3M)
B, Benign; M, malignant.
signal intensity equal to t h a t o f muscle on T I W i m a g e s and a high signal on S T I R sequences ( a n d T 2 W images). Relative acellularity is also said to a c c o u n t for a n atypical low signal on T 2 W images seen in s o m e cases o f aggressive f i b r o m a t o s i s [11,12], which m o r e t y p i c a l l y
have a h e t e r o g e n e o u s high signal on T 2 W images [13], as o b s e r v e d in the p r e s e n t series. I n t r a m u s c u l a r m y x o m a s are characteristically well circumscribed, o f u n i f o r m low signal a n d high signal on T 1 W a n d T 2 W images respectively. These i m a g i n g characteristics m a y also be seen in o t h e r benign soft tissue lesions, e.g. n e u r o m a s . M o s t m y x o i d - c o n t a i n i n g malignancies follow the general rules o f h a v i n g ill-defined m a r g i n s a n d being o f i n h o m o g e n e o u s signal b u t they m a y be p r e d o m i n a n t l y h o m o g e n e o u s a n d circumscribed. M y x o i d l i p o s a r c o m a s b e h a v e differently to o t h e r liposarc o m a s on M R I , h a v i n g a h o m o g e n e o u s l y l o w / m e d i u m signal on T 1W images [7,14]. T h e m y x o i d l i p o s a r c o m a in the present series h a d a signal intensity equal to t h a t o f muscle on T 1 W images, high signal on T 2 W i m a g e s a n d h o m o g e n e o u s l y high signal o n S T I R . It also h a d very well-defined m a r g i n s a n d simulated a benign neural tumour. T h e S T I R sequence has been shown to have a d v a n t a g e s over s t a n d a r d spin echo sequences in detecting e x t r e m i t y t u m o u r s as b o t h T1 recovery a n d T2 d e c a y c o n t r i b u t e to increased tissue signal [15]. M o s t a b n o r m a l tissue has a very high signal intensity on S T I R i m a g e s a n d is conspicuous (Fig. 8). H o w e v e r b o t h n e o p l a s t i c a n d nonn e o p l a s t i c disease (such as o e d e m a a n d i n f l a m m a t i o n ) a p p e a r o f equally high signal, a n d S T I R also d e m o n strates areas o f p e r i t u m o u r a l high signal (as described a b o v e with T 2 W sequences) which m a y represent oedema.
316
CLINICAL RADIOLOGY
(a)
(a)
(b)
(b) Fig. 6 A large well-defined lesion of homogeneous high signal (equal to that of fat) on (a) a sagittal TI -weighted image and suppressed signal on (b) a STIR image (curved arrow). The signal from subcutaneous fat has also been suppressed (open arrow). Diagnosis: benign lipoma.
CONCLUSION M R i m a g i n g has been s h o w n to be g o o d at detecting soft tissue lesions a n d d e m o n s t r a t i n g their size a n d extent [4]. This study has c o n f i r m e d the usefulness o f M R I in the
Fig. 7 - A well-defined lesion in the popliteal fossa ofintenaaediate signal intensity with a low signal central area on proton density-weighted image (a), and high signal on a STIR image (b). Diagnosis: neurilemmoma.
detection a n d d e l i n e a t i o n o f soft tissue lesions, the c o n s p i c u i t y o f which is further increased by use o f the S T I R sequence. A s s e s s m e n t o f the definition o f lesion m a r g i n s a n d the presence o r absence o f i n v a s i o n into a d j a c e n t tissues give s o m e idea o f the aggressive n a t u r e o f the lesion b u t these signs are n o t reliable in distinguishing benign from m a l i g n a n t lesions. Lesion size is a m o r e useful criterion with s a r c o m a s tending to be 5 c m o r m o r e
MRI IN SOFT TISSUE LESIONS
317
typical signal characteristics) and aggressive fibrQmatosis (which merits treatment as a malignant lesion because of its high local recurrence rate). Acknowledgements. We would like to thank the trustees and staff of the Bristol MRI Centre, in particular Miss Ann Case.
REFERENCES
(a)
(b) Fig. 8 A 41-year-old woman with a left thigh lump showing a small lesion on the deep fascia isosignal with muscle on T 1-weighted image (a). The lesion has high signal on a STIR image (b). Diagnosis: nodular fasciitis.
in d i a m e t e r , t h e b e n i g n l e s i o n s in this c a t e g o r y t e n d i n g t o be either b e n i g n l i p o m a s (readily recognizable by their
1 Sundaram M, McGuire MH, Herbold DR. Magnetic resonance imaging of soft tissue masses: an evaluation of fifty-three histologically proven tumours. Magnetic" Resonance Imaging 1988;6:237 248. 2 Ehman RL, Berquist TH, McLeod RA. MR imaging of the musCuloskeletal system: a 5 year appraisal. Radiology 1988;t66: 313-320. 3 Pettersson H, Gillespy T, Hamlin D J, Enneking WF, Springfield DS, Andrew ER et al. Primary musculoskeletal tumors: examination with MR imaging compared with conventional modalities. Radiology 1987;164:237 241. 4 Berquist TH, Ehman RL, King BF, Hodgman CG, Ilstrup DM. Value of MR imaging in differentiating benign from malignant soft tissue masses: study of 95 lesions. American Journal of Roentgenology 1990;155:1251-1255. 5 Kransdorf M J, Jelinek JS, Moser RP, Utz JA, Brower AC, Hudson TM, Berrey BH. Soft-tissue masses: diagnosis using MR imaging. American Journal of Radiology 1989;153:541-547. 6 Beltran J, Simon DC, Katz W, Weis, LD. Increased MR signal intensity in skeletal muscle adjacent to malignant tumours: pathologic correlation and clinical relevance. Radiology 1987;162:251255. 7 Sundaram M, McGuire MH, Herbold DR, Beshany SE, Fletcher JW. High intensity soft tissue masses on TI weighted pulsing sequences. Skeletal Radiology 1987; 16:30-36. 8 Cohen EK, Kressel HY, Perosio T, Burk DL, Dalinka MK, Kanal E et al. MR imaging of soft-tissue hemangiomas: correlation with pathologic findings. American Journal of Roentgenology 1988; 150:1079 1081. 9 Cohen JM, Weinreb JC, Redman HC. Arteriovenous malformations of the extremities: M R imaging. Radiology 1986; 158:475 479. l0 Stull MA, Moser RP, Kransdorf M J, Bogumill GP, Nelson MC. Magnetic resonance imaging of peripheral nerve sheath tumours. Skeletal Radiology 1991;20:9 14. 11 Aisen AM, Martel W, Braunstein EM, McMillin KI, Phillips WA, Kling TF. MRI and CT evaluation of primary bone and soft-tissue tumours. American Journal of Roentgenology 1985;146:749 756. 12 Sundaram M, McGuire MH, Schajowicz F. Soft-tissue masses: histologic basis for decreased signal (short T2) on T2-weighted MR images. American Journal of Roentgenology 1987; 148:1247-1250. 13 O'Keefe F, Kin: EE, Wallace S. Magnetic resonance imaging in aggressive fibromatosis. Clinical Radiology 1990;42:170-173 14 Peterson KK, Renfrew DL, Feddersen RM, Buckwalter JA, E1Khoury GY. Magnetic resonance imaging of myxoid containing tumours. Skeletal Radiology 1991;20:245-250. 15 Shuman WP, Patten RM, Baron RL, Liddell RM, Conrad EU, Richardson ML. Comparison of STIR and spin-echo MR imaging at 1.5T in 45 suspected extremity tumors: lesion conspicuity and extent. Radiology 1991;179:247 252.
Review of the Use of MRI in Soft Tissue Lesions S. J. A R M S T R O N G , C. J. W A K E L E Y , P. R. G O D D A R D and I. W A T T
Department of Clinical Radiology, United Bristol Healthcare NHS Trust, Bristol The MR images of 34 patients with soft tissue lesions were retrospectively evaluated to assess the accuracy of the technique in distinguishing benign from malignant lesions, and to assess the usefulness of various criteria in making this distinction. The overall sensitivity for the detection of malignancy was 75% with a specificity of 94%. Size of lesion was found to be a good criterion in predicting malignancy, lesion margin and signal intensity were less useful. The tissue type was determined in a few instances where signal characteristics were typical, notably lipomas and neural tumours, but this was not reliable and in most lesions the tissue of origin cannot be determined on MR imaging and biopsy is necessary.
Armstrong, S.J., Wakeley, C.J., Goddard, P.R. & Watt, I. (1992). Clinical Radiology 46, 311-317. Review of the Use of MRI in Soft Tissue Lesions
Accepted for Publication 19 July 1992
J
Magnetic resonance imaging (MRI) has been shown to be superior to CT in demonstrating soft tissue lesions, owing to the better contrast resolution and the ability to image in any plane [1]. MRI permits reliable identification of the spatial boundaries of tumour masses, encasement and displacement of major neurovascular bundles and the extent of joint involvement [2]. Also it can depict intralesional necrosis and haemorrhage [3]. The distinction between benign and malignant soft tissue lesions is assessed by applying generally accepted criteria (Table 1) but a difference of opinion exists over the usefulness and accuracy of these features [4,5]. Tissue characterization by M R I is limited, with a few notable exceptions. Measurement of relaxation times is technically difficult and has not proved to be of value in differentiating turnout types [2]. Thus the precise role of MRI in the investigation of soft tissue tumours remains unclear. In an attempt to clarify the position, patients referred to the Bristol MRI Centre for this indication were reviewed retrospectively to determine how reliably MRI could distinguish: 1 Benign from malignant lesions, 2 Benign/malignant turnouts from non-tumour lesions (e.g. inflammatory/post-traumatic), 3 Particular turnout types.
P A T I E N T S AND M E T H O D S The magnetic resonance (MR) images of 34 patients (19 males and 15 females) with a suspected soft tissue lesion referred to the Bristol MRI Centre between 1987 and 1991 were reviewed retrospectively. The age range was from 3 to 82 years (average age 40). All scans were obtained with a 0.5T superconducting magnet. Typical scan sequences included Tl-weighted spin echo (SE 500/ 26), and short tau inversion recovery (STIR) sequences, (1500/100/30). Twenty out of 34 patients also had T2weighted (SE 1500/80) images performed. Gadolinium (Gd-DTPA) enhancement was not used in any of the patients in this series. Standard sagittal, coronal and transverse scan planes were used. Scans had been reported previously by four radiologists, one of whom (IW) subsequently reviewed all the images without prior knowledge of the history, the diagnosis made or the results of any other imaging studies. Histological diagnosis was available in all 34 patients, obtained by biopsy or excision. Twenty-three patients had benign lesions including both benign neoplasms and inflammatory and tumour-like masses. Eleven had malignant lesions, nine primary, including two with NonHodgkin's lymphoma (being the initial presentation in both), and two secondary malignancies. Specific diagnoses are listed in Table 2. Presentation was with a localized (24) or diffuse (10) swelling, usually in a limb.
Table 1 -Distinctive MRI features in benign and malignant soft tissue lesions*
Benign
Malignant
Size Margin Signal intensity
< 3 cm Well-defined Homogeneous
Distribution
Confined to soft tissues
> 3 cm Ill-defined/irregular Inhomogeneous/intensity changes in surrounding tissues Infiltration of adjacent muscle or bone
* See Berquist et al. [4]. Correspondence to: Dr S. J. Armstrong, Department of Clinical Radiology, United Bristol Healthcare N H S Trust, Bristol BS2 8HW.
RESULTS Using the criteria in Table 1, 24 of these 34 lesions were judged by MRI to be benign and 10 to be malignant. Of the 24 lesions thought to be benign three were subsequently shown to be malignant lesions pathologically: one was a myxoid liposarcoma and two were metastatic deposits. Two of the 10 tumours suspected of being malignant were subsequently shown to be benign: a benign haemangioma was suspected on M R to be a rhabdomyosarcoma (Fig. 1) and a ganglion cyst of the
312
CLINICAL RADIOLOGY
knee was t h o u g h t on M R to be infiltrative and therefore possibly malignant in nature (Fig. 2). The sensitivity for the detection o f malignancy on M R I was 75% with a specificity of 94%.
T I S S U E C H A R A C T E R I Z A T I O N (Table 2) (a) Benign Lesions The diagnosis or differential diagnosis made by M R I was consistent with the histological result in all except five cases. One lesion t h o u g h t to be an atypical lipoma was f o u n d to be pigmented villonodular synovitis (PVNS), one suspected vascular m a l f o r m a t i o n an area o f scar tissue, and one suspected bursa a collection o f fibrotic lymph nodes. A n area o f fat necrosis in one patient and n o d u l a r fasciitis in another were correctly diagnosed as being benign lesions, although the precise nature of the lesions could not be determined from the M R images. (b) Malignant Lesions
(a)
O f the 11 pathologically confirmed malignancies the M R images correctly suggested a malignant cause in eight cases and incorrectly suggested a benign cause in three cases: one myxoid liposarcoma had M R appearances t h o u g h t to be typical o f a benign s c h w a n n o m a (Fig. 3), and two metastases were incorrectly interpreted as being due to h a e m a t o m a (Fig. 4). In one patient with N o n H o d g k i n ' s l y m p h o m a and another with malignant melan o m a both a turnout mass and involved lymph nodes were demonstrated (Table 3). All o f the wrongly diagnosed lesions had T l W , T 2 W and S T I R sequences performed, with the single exception o f the patient whose biopsy yielded scar tissue who had T 1 W and S T I R sequences only. (b) Fig. 1 A 14-year-old boy with pain and swelling in the thigh has an illdefined lesion of intermediate heterogeneous signal intensity in vastus lateralis on Tl-weighted image (a). On (b), a proton density-weighted image, the lesion is of high signal, again with ill-defined margins. Diagnosis: benign haemangioma.
Table 2 - Final diagnosis (34 patients) Benign lesions
Bursa Fibromatosis Fat necrosis Fibrotic lymph nodes Ganglion cyst Haemangioma Lipoma Lymphangioma Meniscal trauma Neuroma Nodular fasciitis PVNS Scar tissue Total
2 2 1 1 1 2 5 1 1 4 1 1 1 23
Malignant lesions
Alveolar cell sarcoma Fibrosarcoma Liposarcoma MFH Malignant melanoma NHL Rhabdomyosarcoma Total
1 1 2 2 2 2 1 11
Size (Table 4) Of the benign lesions 43% were less than 3 cm in diameter and 78% less than 5 cm. The benign lesions greater than 5 cm in diameter were either lipomas or fibromatosis. This c o m p a r e d with 55% o f malignant lesions that were greater than 5 cm and 82% over 3 cm. Only two (18%) o f the malignant lesions were less than 3 cm and these were b o t h metastatic malignant m e l a n o m a (Fig. 5). The malignant lesions greater than .5 cm in diameter were all sarcomas, including the m y x o i d liposarcoma. Margin and Signal Characteristics Benign lipomas tended to be well-defined and wellencapsulated. All but two were superficial, one o f the deep lesions was infiltrating between muscle fibres. All had a high signal on T 1 W images and low signal on S T I R images (Fig. 6).
MRI IN SOFT TISSUE LESIONS
313
on S T I R images. Neoplastic lesions tended to be inhomogeneous with a mixed intermediate or low signal intensity on T1W images and a high (but heterogeneous) signal on STIR. Six of the 11 malignant lesions had ill-defined margins, the margins of the other five being well-defined. Most but not all had heterogeneous signal intensity, three lesions had homogeneous signal: the myxoid liposarcoma (q.v.) and both cases of N o n - H o d g k i n ' s lymphoma.
DISCUSSION
(a)
M R I has been shown to be good at detecting soft tissue lesions, defining their extent and their relationships to surrounding structures. The distinction of benign from malignant has been proposed from consideration of the following:
(a) Size Berquist et al. [4] found that the size o f a tumour was a useful discriminator, with a specificity and accuracy of 90%. In our series almost a quarter of benign lesions and no primary malignancies were less than 3 cm in diameter, and 87% of malignancies were greater than 5 cm in diameter. Thus, size proved a useful but not reliable discriminator, in particular the two metastases in our group were less than 3 cm in diameter.
(b) Margin (b) Fig. 2 A 62-year-old man with a lump in the knee which is shown to be cystic and multiloculated with ill-defined margins laterally on proton density-weighted images (a). The margins are again ill-defined on STIR image (b). Diagnosis: benign ganglion cyst.
Benign neural lesions such as schwannomas, neurofibromas and neurilemmomas tended to be well-defined lesions and in close relationship to enlarged nerves. They all had a signal intensity similar to that of muscle on TI W images and a high signal on T2W and STIR images (Fig. 7). Two of the four neuromas in the present series had a low signal margin and two had a whorled or laminated appearance with low signal areas centrally. The myxoid liposarcoma mistakenly diagnosed on M R I as a benign schwannoma had the former appearance. Vascular lesions such as haemangiomas, lymphangiomas and haematomas tended to be more diffuse, illdefined lesions of heterogeneous signal intensity. They were either isointense or of slightly higher signal than muscle on T1W images and of slightly increased (but not the very bright signal seen with neural lesions) on S T I R images. In one of the two haemangiomas in the present series large feeding blood vessels could be identified (seen as areas of signal void). Joint-related lesions tended not to have any characteristic features but the diagnosis was aided by the typical site in close relationship to a joint. Cystic lesions such as bursae had a low signal on T1W images and high signal
In the series reported by Berquist et al. [4], benign lesions typically had well-defined or slightly irregular margins whereas this was the case with only 15% of malignancies. However, others have found this criterion unreliable [5] as some benign lesions are surrounded by oedema making them ill-defined. Aggressive fibromatosis is an example of a benign but locally invasive tumour that can involve bone, a feature usually thought to indicate malignancy [4]. In the present series, lesion margin was an unreliable sign as only half of the malignant lesions had ill-defined margins, the other half being well-defined.
(c) Signal Intensity and Homogeneity It is unusual for a malignant lesion to show homogeneous signal intensity [4] but heterogeneous signal intensity is not specific for malignancy and can be seen in m a n y benign lesions [5]. I f uniform signal intensity is used as a criterion of benignity then false positive diagnoses of malignancy will occur with necrotic tumours and haemangiomas. Increased signal intensity in skeletal muscle adjacent to a soft tissue mass on T2W images is an important indicator of either oedema or tumour invasion [6] but some benign conditions such as infection, myositis or haematoma, e.g. post-surgery, can produce similar M R I appearances. Most soft tissue masses have a low signal intensity, equal to or lower than muscle, on T1W images and a high signal intensity on T2W images [1]. Notable exceptions are lesions containing fat or haemorrhage which have a
314
CLINICALRADIOLOGY
(a)
(b)
(e) Fig. 3 - A 47-year-oldwoman with swellingin the left calf shows a uniform ovoid lesion isointensewith muscleon Tl-weighted image (a) (open arrow). The lesion has high signal on a T2-weightedimage (b) and is homogeneousand well-definedusing STIR sequence (c). Diagnosis: myxoid liposarcoma.
high signal on T1W images, e.g. lipomas, liposarcomas, haemangiomas, non-acute haematomas and haemorrhage into a pre-existing mass [7]. The signal strength of haemorrhage depends on its age, acute haemorrhage (within 24 tO 36 h) having a low/medium signal on T1W, and subacute haemorrhage a high signal due to shortening o f T 1 by the paramagnetic effect of methaemoglobin. Benign lipomas are probably the most characteristic soft tissue lesion on MRI, having a high signal on T1W and T2W and very low signal on S T I R (which suppresses fat): Soft tissue haemangiomas can be further distinguished by a high signal striated/separated configuration on certain spin-echo sequences [8]. This is thought to correlate histologically with vascular channels of varying sizes separated by fibrofatty septa. Tubular areas of signal void in relation to a soft tissue abnormality are due to
dilated tortuous blood vessels and are characteristic of arteriovenous malformations [9]. The appearance is similar on both T1W and T2W images, the flow void effect being unaltered by the T R values, presumably due to the high velocity of blood flow within arteriovenous malformations. This appearance was noted in a superficial haemangioma in the present series. Peripheral nerve sheath tumours usually have an ovoid or fusiform shape and are of intermediate to moderately high signal intensity on T1W images and high signal on T2W images [10]. They tend to exhibit variable inhomogeneity of signal on both sequences, whether benign or malignant, which is thought to be d tt¢ to coexistent areas of hypocellularity and hypercellularity, and differentiating benign from malignant turnouts is unreliable. In the present series peripheral nerve sheath tumours had a
MRI IN SOFT TISSUE LESIONS
315
Fig. 4 - A previously fit 77-year-old woman with a lump in the right calf which was isointense with muscle on T 1 weighting (not illustrated) and high signal on a STIR image. The peripheral site and close relationship to dilated veins suggested an area of haematoma. Diagnosis: vascular amelanotic malignant melanoma. Fig. 5 - A 67-year-old woman with known malignant melanoma and a lump in the thigh showing a small well-defined lesion with a high signal on a sagittal T2-weighted image. Diagnosis: metastatic melanoma within a thrombosed vein. Table 3 - M R I diagnosis vs pathological findings (34 patients)
MRI benign MRI malignant
Path benign
Path malignant
21 2
3 8
Table 4 - M R I estimation o f size (diameter)
Small ( < 3 cm)
Intermediate (3 5 cm)
Large ( > 5 cm)
Bursa x 2 Alveolar cell sarcoma Fibrotic lymph nodes Ganglion cyst Haemangioma Haemangioma Lipoma Lipoma x 2 Meniscal trauma Lymphangioma Metastatic MM × 2 MFH Neuroma x 3 Neuroma Nodular fasciitis NHL PVNS Sear tissue
Fibromatosis x 2 Fat necrosis Fibrosarcoma Lipoma × 2 Liposarcoma x 2 MFH NHL Rhabdomyosarcoma
Total = 13 (11B, 2M)
Total = 11 (5B, 6M)
Total = 10 (7B, 3M)
B, Benign; M, malignant.
signal intensity equal to t h a t o f muscle on T I W i m a g e s and a high signal on S T I R sequences ( a n d T 2 W images). Relative acellularity is also said to a c c o u n t for a n atypical low signal on T 2 W images seen in s o m e cases o f aggressive f i b r o m a t o s i s [11,12], which m o r e t y p i c a l l y
have a h e t e r o g e n e o u s high signal on T 2 W images [13], as o b s e r v e d in the p r e s e n t series. I n t r a m u s c u l a r m y x o m a s are characteristically well circumscribed, o f u n i f o r m low signal a n d high signal on T 1 W a n d T 2 W images respectively. These i m a g i n g characteristics m a y also be seen in o t h e r benign soft tissue lesions, e.g. n e u r o m a s . M o s t m y x o i d - c o n t a i n i n g malignancies follow the general rules o f h a v i n g ill-defined m a r g i n s a n d being o f i n h o m o g e n e o u s signal b u t they m a y be p r e d o m i n a n t l y h o m o g e n e o u s a n d circumscribed. M y x o i d l i p o s a r c o m a s b e h a v e differently to o t h e r liposarc o m a s on M R I , h a v i n g a h o m o g e n e o u s l y l o w / m e d i u m signal on T 1W images [7,14]. T h e m y x o i d l i p o s a r c o m a in the present series h a d a signal intensity equal to t h a t o f muscle on T 1 W images, high signal on T 2 W i m a g e s a n d h o m o g e n e o u s l y high signal o n S T I R . It also h a d very well-defined m a r g i n s a n d simulated a benign neural tumour. T h e S T I R sequence has been shown to have a d v a n t a g e s over s t a n d a r d spin echo sequences in detecting e x t r e m i t y t u m o u r s as b o t h T1 recovery a n d T2 d e c a y c o n t r i b u t e to increased tissue signal [15]. M o s t a b n o r m a l tissue has a very high signal intensity on S T I R i m a g e s a n d is conspicuous (Fig. 8). H o w e v e r b o t h n e o p l a s t i c a n d nonn e o p l a s t i c disease (such as o e d e m a a n d i n f l a m m a t i o n ) a p p e a r o f equally high signal, a n d S T I R also d e m o n strates areas o f p e r i t u m o u r a l high signal (as described a b o v e with T 2 W sequences) which m a y represent oedema.
316
CLINICAL RADIOLOGY
(a)
(a)
(b)
(b) Fig. 6 A large well-defined lesion of homogeneous high signal (equal to that of fat) on (a) a sagittal TI -weighted image and suppressed signal on (b) a STIR image (curved arrow). The signal from subcutaneous fat has also been suppressed (open arrow). Diagnosis: benign lipoma.
CONCLUSION M R i m a g i n g has been s h o w n to be g o o d at detecting soft tissue lesions a n d d e m o n s t r a t i n g their size a n d extent [4]. This study has c o n f i r m e d the usefulness o f M R I in the
Fig. 7 - A well-defined lesion in the popliteal fossa ofintenaaediate signal intensity with a low signal central area on proton density-weighted image (a), and high signal on a STIR image (b). Diagnosis: neurilemmoma.
detection a n d d e l i n e a t i o n o f soft tissue lesions, the c o n s p i c u i t y o f which is further increased by use o f the S T I R sequence. A s s e s s m e n t o f the definition o f lesion m a r g i n s a n d the presence o r absence o f i n v a s i o n into a d j a c e n t tissues give s o m e idea o f the aggressive n a t u r e o f the lesion b u t these signs are n o t reliable in distinguishing benign from m a l i g n a n t lesions. Lesion size is a m o r e useful criterion with s a r c o m a s tending to be 5 c m o r m o r e
MRI IN SOFT TISSUE LESIONS
317
typical signal characteristics) and aggressive fibrQmatosis (which merits treatment as a malignant lesion because of its high local recurrence rate). Acknowledgements. We would like to thank the trustees and staff of the Bristol MRI Centre, in particular Miss Ann Case.
REFERENCES
(a)
(b) Fig. 8 A 41-year-old woman with a left thigh lump showing a small lesion on the deep fascia isosignal with muscle on T 1-weighted image (a). The lesion has high signal on a STIR image (b). Diagnosis: nodular fasciitis.
in d i a m e t e r , t h e b e n i g n l e s i o n s in this c a t e g o r y t e n d i n g t o be either b e n i g n l i p o m a s (readily recognizable by their
1 Sundaram M, McGuire MH, Herbold DR. Magnetic resonance imaging of soft tissue masses: an evaluation of fifty-three histologically proven tumours. Magnetic" Resonance Imaging 1988;6:237 248. 2 Ehman RL, Berquist TH, McLeod RA. MR imaging of the musCuloskeletal system: a 5 year appraisal. Radiology 1988;t66: 313-320. 3 Pettersson H, Gillespy T, Hamlin D J, Enneking WF, Springfield DS, Andrew ER et al. Primary musculoskeletal tumors: examination with MR imaging compared with conventional modalities. Radiology 1987;164:237 241. 4 Berquist TH, Ehman RL, King BF, Hodgman CG, Ilstrup DM. Value of MR imaging in differentiating benign from malignant soft tissue masses: study of 95 lesions. American Journal of Roentgenology 1990;155:1251-1255. 5 Kransdorf M J, Jelinek JS, Moser RP, Utz JA, Brower AC, Hudson TM, Berrey BH. Soft-tissue masses: diagnosis using MR imaging. American Journal of Radiology 1989;153:541-547. 6 Beltran J, Simon DC, Katz W, Weis, LD. Increased MR signal intensity in skeletal muscle adjacent to malignant tumours: pathologic correlation and clinical relevance. Radiology 1987;162:251255. 7 Sundaram M, McGuire MH, Herbold DR, Beshany SE, Fletcher JW. High intensity soft tissue masses on TI weighted pulsing sequences. Skeletal Radiology 1987; 16:30-36. 8 Cohen EK, Kressel HY, Perosio T, Burk DL, Dalinka MK, Kanal E et al. MR imaging of soft-tissue hemangiomas: correlation with pathologic findings. American Journal of Roentgenology 1988; 150:1079 1081. 9 Cohen JM, Weinreb JC, Redman HC. Arteriovenous malformations of the extremities: M R imaging. Radiology 1986; 158:475 479. l0 Stull MA, Moser RP, Kransdorf M J, Bogumill GP, Nelson MC. Magnetic resonance imaging of peripheral nerve sheath tumours. Skeletal Radiology 1991;20:9 14. 11 Aisen AM, Martel W, Braunstein EM, McMillin KI, Phillips WA, Kling TF. MRI and CT evaluation of primary bone and soft-tissue tumours. American Journal of Roentgenology 1985;146:749 756. 12 Sundaram M, McGuire MH, Schajowicz F. Soft-tissue masses: histologic basis for decreased signal (short T2) on T2-weighted MR images. American Journal of Roentgenology 1987; 148:1247-1250. 13 O'Keefe F, Kin: EE, Wallace S. Magnetic resonance imaging in aggressive fibromatosis. Clinical Radiology 1990;42:170-173 14 Peterson KK, Renfrew DL, Feddersen RM, Buckwalter JA, E1Khoury GY. Magnetic resonance imaging of myxoid containing tumours. Skeletal Radiology 1991;20:245-250. 15 Shuman WP, Patten RM, Baron RL, Liddell RM, Conrad EU, Richardson ML. Comparison of STIR and spin-echo MR imaging at 1.5T in 45 suspected extremity tumors: lesion conspicuity and extent. Radiology 1991;179:247 252.
