709
Figure 1
4,
‘H
to correspond to MR images in either the We focused on the contribution of MR radiologist’s bias for an aesthetically
coronal or the sagittal plane. Extraosseous imaging and CT in enabling detection of pleasing image compared with a less
extent and longitudinal extent in the me- the lesion, prediction of its malignancy or pleasing image containing better informa-
dullary canal (including epiphyseal in- benignancy, and delineation of its anatomic tion.
volvement) were specifically noted by the extent. MR images and CT hard-copy films
consulting pathologist. All imaging studies were reviewed. The anatomic extent on MR
were independently reviewed by three of images was compared with that on CT scans RESULTS
us (W.D.Z., T.H.B., and RAM.). Three re- by noting the following for each modality
Of 52 patients, 32 had primary neo-
viewers were used in an attempt to balance in each patient: cortical destruction; tumor
plasms of bone (25 malignant and
the potential bias for MR imaging of a extent in the medullary canal; size of mass
skeletal radiologist active in MR imaging in extraosseous soft tissue; and neurovas- seven benign). The primary malignant
with the potential bias against MR imaging cular, fat, muscle, and joint involvement. neoplasms included ten osteosarcomas,
of the other two reviewers who were more The extent of a lesion was subdivided into six chondrosarcomas, four lymphomas,
accustomed to seeing lesions on CT scans. these categories in an attempt to control the three Ewing sarcomas, a fibrosarcoma,
not cause significant artifact on CT the many patients who could not tol- normal marrow signal. Judging from
.4
scans. In the remaining four patients, erate additional time in the magnet. tumor signal alone, we found no sig-
the CT scan was significantly degraded Data obtained with spectroscopically nificant difference between benign
because of metal artifacts. In one of evaluated samples revealed that Ti and malignant tumors. Both gave H
these patients, the recurrent tumor was values were accurate above 100 msec lower signal than normal marrow on
missed using CT because of artifact and that T2 values were accurate from inversion-recovery sequences and on
from an internal-fixation device (Fig. 25 to 200 msec, with an increase in the partial-saturation sequences (spin-echo
3). Another patient with suspected re- standard deviation above this level (J. with short repeat times). Both the be-
current neoplasm and a metallic pros- E. Gray, unpublished data). Ti values nign and the malignant tumors gave
thesis underwent MR imaging only; for normal marrow ranged from 195 to increased signal on spin-echo Se-
CT was not used because of the obvious 307 msec, and T2 values from 80 to 116 quences with long repeat times. Le-
superiority of MR imaging in this set- msec (Table 2). Relaxation times for sions containing calcium or causing 4,
ting. An earlier CT scan was rendered small bones were difficult to assess ac- sclerosis may show decreased signal on
indeterminate by artifact. The MR curately because of the small area any spin-echo or inversion-recovery
image was free of significant artifact available for measurement. In a patient sequence compared with normal mar-
and accurately demonstrated recurrent with osteomyelitis of the distal tibia, row signal.
neoplasm (Fig. 4). the uninvolved marrow had a higher Many of the criteria for differen-
Ti than expected, possibly due to hy- tiating benign from malignant lesions
peremia. Ti values for malignancies with the use of plain films and CT
ranged from 404 to 887 msec; for be- scans are applicable to MR images (8).
nign lesions, the range was 375-628 All seven benign neoplasms were
712 Radiology
#{149} June 1985
Figure 4
.1,
C.
4.
5’
intensity than normal marrow, and its signal is inhomogeneous. Predominantly low signal corresponds to extent of calcification seen on
plain films and CT scan. Scattered areas of higher and lower signal are typical of malignancy and could be due to hemorrhage, necrosis, V
or calcification. Tumor is not separated from the normal marrow by a rim of low density, as is true of many benign lesions (see Fig. 6). Cortex
is somewhat indistinct at several locations around periphery, suggesting cortical invasion. Posteriorly, there is a thin rim of high signal
of unknown cause. Popliteal artery (arrow) is clearly not invaded. Note normal black appearance of patellar tendon (arrowhead).
c. SE sequence (TE 40 msec, TR 500 msec). Sagittal image shows tumor abutting but not crossing the physis. Distal extent is also well
4
shown.
d. Specimen from above-knee amputation 2 days after MR imaging shows tumor abutting but not crossing the epiphyseal plate. Tumor invaded
the cortex and elevated the periosteum. There was no soft-tissue invasion.
‘1
C1.4
Extent of Tumors on MR Images resolution was inferior to that of CT derwent both MR and CT examina-
and CT Scans scans. However, even in this group of tions, CT was more accurate than MR
CT and MR were compared in their patients, MR images were sometimes in showing cortex destruction. CT was
ability to depict the extent of bone tu- more accurate, because of superior superior to MR in demonstrating ma-
mors as determined by surgical pa- contrast capabilities, than CT scans trix mineralization and pathologic
thology (Tables 3 and 4). In MR images were in showing extent of tumor. fractures. CT was also better at imaging
of poor or only fair quality, spatial For 18% of the 45 patients who un- normal cortical bone. MR imaging was
714 Radiology
#{149} June 1985
Figure 7
a. c.
Osteomyelitis (Staphylococcus aureus) involving both distal tibias.
a. Lytic lesion crosses epiphyseal plate of distal right tibia.
b. SE sequence (TE 40 msec, TR = 500 msec). Sagittal image shows lesion (arrow), whose signal is equal to that of
normal marrow on this sequence. Lesion crossing the physis can be detected because of the low-intensity interface
with normal marrow.
c. SE sequence (TE = 40 msec, TR 2,000 msec). Lesions are seen in both distal tibial metaphyses. Lesions are of higher
signal than normal tibial marrow. Low-intensity shell separates lesions from normal marrow. Appearance is typical
of benign lesion. Multiplicity suggests infection rather than primary neoplasm. Achilles tendons appear black.
p Muscles appear as an intermediate gray.
#1
#{149}0
4
Grade 4 osteosarcoma in distal left femur. Patient had above-knee amputation on the right 3 years earlier for grade 4 osteosarcoma.
a. Sclerotic lesion in diaphysis of femur, with a biopsy defect laterally.
b. SE sequence (TE = 40 msec, TR = 600 msec). Coronal image shows proximal extent of sarcoma and biopsy site.
c. SE sequence (TE = 40 msec, TR = 600 msec). Coronal image inferior to the site in b shows three distal skip lesions (arrows) that were not .4
716 Radiology
#{149} June 1985
msec, TR = 600 msec) are that it can be 33% (4 of 12) of the benign lesions. In Skip lesions localized to the marrow
performed more quickly and it is rela- one patient, MR images allowed us to can be seen on MR images, even when
tively Ti weighted. This is useful when predict preoperatively that a lesion was they are not evident on plain films or
, I time is an important factor, as in an likely benign (aneurysmal bone cyst) CT scans. MR images can show the
anxious or claustrophobic patient. when findings on CT scans and plain presence or absence of joint and epi-
‘ Greater T2 weighting is possible using films suggested Ewing sarcoma. physeal involvement, which is not well
longer TE and TR values. The poten- CT is better than MR imaging for depicted on CT scans because of the
.4 tially better sensitivity with greater T2 showing normal cortex, while MR axial format of CT. In our opinion, co-
CIZ weighting should be considered in imaging is better than CT for showing ronal or sagittal images are essential to
light of the increased examination time normal marrow because normal mar- define longitudinal involvement of the
needed for longer TR values. Our ex- row emits high signal when MR is medullary cavity and soft tissues. Co-
perience with pulse sequences with used. ronal and sagittal images also aid in
longer TE and TR values is too limited Because of its superior spatial reso- clarifying the anatomic relationship
.1 to enable strong conclusions regarding lution and ability to demonstrate between a soft-tissue mass and a nor-
their use in the imaging of bone tu- mineral as an area of increased signal, mal anatomic structure (vessel, nerve,
1 mors. CT is better than MR imaging in muscle; Fig. i), but this type of infor-
Lesions in bone marrow yield a de- showing pathologic fractures and mation can usually be obtained on axial
* creased signal on inversion-recovery mineralization. The relatively poor MR images.
and partial-saturation sequences (TE spatial resolution of MR imaging will MR imaging has potential in exami-
‘ 40 msec, TR = 500 msec). Uncalcified improve as hardware and software nations of the tumor response tora-
k marrow
with long
lesions
repeat
on spin-echo
times (TE
sequences
60 msec,
improve.
In general, MR is better than CT in
diotherapy or chemotherapy. In our
series, only one patient underwent MR
1 TR = 2,000 msec) give increased signal. showing the extent of tumor in marrow imaging both before and after preop-
These results are due to an increase of and soft tissues. Marrow lesions are erative chemotherapy. In this case, the
‘4 both Ti and T2 in bone tumors corn- often not visible on plain films and are previously high-contrast lesion was
pared with normal marrow. On spin- subtle on CT scans. The general supe- difficult to detect on the postche-
. echo sequences, calcified or sclerotic riority of MR is due to two main factors. motherapeutic study.
4 regions (including osteoblastic metas- First, MR is able to show high-contrast In summary, MR imaging is occa-
tases and lesions that produce osteoid differences between soft-tissue struc- sionally helpful in the diagnosis of
V or cartilage) give decreased signal rel- tures that are of about equal density on tumor type. Plain films remain supe-
C’ ative to normal marrow, depending on CT. This is because, on a CT scan, den- nor for enabling prediction of the
the amount of calcification in the le- sities of adjacent soft-tissue structures histologic diagnosis, with CT and MR
may differ by only a fraction of a per- imaging helpful in rare instances. The
Cortical lesions on both spin-echo centage while, for an MR image, in- signal intensity of MR has not been
, and inversion-recovery sequences tensities of the same adjacent soft-tissue useful in enabling prediction of ma-
show a signal where there should be structures may differ by several hun- lignancy or benignancy. Further work
none. Cortical involvement appears as dred milliseconds, owing to differ- is necessary to determine if Ti and T2
a shade of gray rather than the normal ences in nuclear spin density, Ti, and are of any use for prediction of the
uniform black rim surrounding the T2 (6). This contrast superiority of MR histologic diagnosis. On an MR image,
. rnedullary canal. over CT is most commonly used in poorly defined irregular inhomo-
MR imaging does not allow predic- distinguishing a soft-tissue mass from geneities other than calcium within a
Jr tion of the malignancy or benignancy skeletal muscle, which are of about tumor are highly suggestive of malig-
of a tumor by its signal intensity alone. equal density on CT images not en- nancy. A well-defined, smooth, low-
V We were able to obtain Ti and T2 in hanced by contrast material (Fig. 2). intensity border between the tumor
only 13 patients. On the basis of these MR is also superior to CT in distin- and normal marrow suggests benig-
limited data, benign and malignant guishing tumor from unopacified nancy.
‘. tumors cannot be differentiated by loops of bowel. Another advantage of MR imaging is superior to CT in
their relaxation times. MR is that the presence or absence of demonstrating the extent of tumor
Occasionally, the intraosseous por- vascular involvement is shown on within the medullary canal and soft
8 tion of a tumor was of different (usu- images without the risk of intravenous tissues. MR imaging becomes more
. ally lower) signal intensity than ex- contrast material. Blood vessels may valuable, relative to CT, as one pro-
trasseous soft-tissue mass. This differ- appear black, white or target-like (9), ceeds distally from the axial skeleton.
ence may be due to the extraosseous depending on the flow velocity and MR imaging should be useful in eval-
1 portion being composed of purely pulse sequence. Figure 2b shows that uations for metastases in a patient with
tumor cells, while the intraosseous MR imaging depicts vascular structures a known primary lesion when the bone
. portion is composed of tumor mixed better than CT does. In our opinion, the scan is positive and the plain film
* with residual bone of low signal in- degree to which soft-tissue mass in- findings are normal. Such metastases
tensity. volves adjacent structures such as involving exclusively the marrow
The presence of poorly defined in- muscles, nerves, and blood vessels is cavity are difficult to identify on CT
homogeneities other than calcium usually adequately shown on axial MR scans. MR imaging is occasionally
‘ within a primary bone tumor suggests images. useful in demonstrating that a demin-
malignancy. This MR finding was Second, MR permits direct coronal eralized area is the result of disuse os-
#{188} present in more than one-fourth (ii of and sagittal imaging. This is particu- teoporosis and not tumor invasion
40) of the malignant lesions in our larly valuable in delineating the lon- from an adjacent bone. It is valuable
study. When present, it enabled a reli- gitudinal extent of a primary malig- when the optimal surgical approach
able prediction of malignancy. A sign nancy in the medullary canal (Figs. 1 demands precise knowledge of tumor
of benignancy is a well-defined low- and 5). This longitudinal extent is of extent, as in limb-salvage operations.
signal rim between the lesion and the critical importance in planning the Skip lesions in the marrow which are
adjacent marrow: this was noted in 5% level for amputation or the limits of missed on plain films and CT scans can
(2 of 40) of the malignant lesions but in resection for a limb-salvage operation. be detected with the use of MR imag-
C’
.*
718 Radiology
#{149} June 1985