1049

Detection of Multiple Myeloma

Involving the Spine: Efficacy of Fat-
Suppression and Contrast-Enhanced MR
Imaging

Alain Rahmouni OBJECTIVE. To determine the efficacy of fat-suppressed sequences and contrast-
Marine Divine2 enhanced MR imaging for the detection of focal spinal lesions caused by multiple
myeloma, we obtained MR images in 32 patients with newly diagnosed myeloma who
Didier 1
had back pain.
Mondher GolIi1
SUBJECTS AND METHODS. All patients had biopsy-proved myeloma and had MR
Thu Ha Dao1 imaging at the painful level of the spine. Spin-echo Ti-weighted, T2-weighted, and
Nidal 1 short TI inversion-recovery (STIR) images; dynamic ultrafast low-angle shot (turbo-
Marie Christine Anglade1 FLASH) images after IV injection of a bolus of paramagnetic contrast material; and
Felix Reyes2 contrast-enhanced Ti-weighted images were obtained. We qualitatively compared
Norbert Vasile1 the signal intensities and contrast enhancement of focal lesions with those of the
surrounding vertebral bodies.
RESULTS. Multiple lesions were detected in all but two of the 32 patients. On T2-
weighted and STIR images, all lesions had homogeneously high signal intensity. On
Ti-weighted images, the lesions were visible as hypointense areas compared with
surrounding bone in all except four patients, in whom the lesions were isointense or
hyperintense. All tumor nodules enhanced on turbo-FLASH images obtained in the
arterial phase. No additional lesions were seen on STIR or contrast-enhanced
images. MR findings resulted in a change in the staging of the disease in one patient
and led to prompt treatment in five patients with epidural involvement.
CONCLUSION. MR imaging appears to be helpful in detecting spinal involvement
in patients with multiple myeloma. The diagnosis of spinal lesions is best achieved
by using either fat-suppressed or T2-weighted images. Although myeloma lesions
enhanced in all patients, contrast material appears to be of no value for the detection
of additional lesions.

AJR i993;i60:i049-i052

The presence and extent of osteobytic bone lesions in patients with multiple
myeboma are important factors in the staging of the disease [1]. The most com-
mon site of involvement is the spine, and epidural involvement is a well-known
complication of spinal myeloma [1 , 2]. Two studies [3, 4] have shown that MR
imaging is more sensitive than conventional radiography for the detection of spi-
nab myeboma. In those reports, the MR appearance of myeloma lesions on Ti -
and T2-weighted spin-echo (SE) images only was described. The value of fat-sup-
pnession sequences and contrast-enhanced MR imaging for the detection of spi-
nab myeloma lesions has not yet been studied.
Received August 25, 1992; accepted after revi-
sion December 3, 1992. We prospectively used MR imaging in 32 patients with newly diagnosed
1 Department of Radiology, H#{244}pitalHenri Mon- myeboma who had back pain. We evaluated MR features of spinal lesions on
don, 51 Avenue du Mar#{233}chaldo Lattre do Tassigny, unenhanced Ti- and T2-weighted SE images, unenhanced short TI inversion-
9401 0 Cr#{233}toil,France. Address correspondence to
A. Rahmouni.
recovery (STIR) images, contrast-enhanced dynamic ultrafast low-angle shot
(turbo-FLASH) images [5], and contrast-enhanced Ti-weighted SE images. The
2Department of Clinical Hematology, H#{244}pital
Henri Mondor, 94010 Cr#{233}teil,France. purpose of the study was twofold: to describe the MR appearance of myeboma
0361-803X/93/1605-1 049
lesions of the spine on unenhanced and enhanced MR images and to assess the
American Roentgen Ray Society value of MR contrast material in the detection of myeboma lesions.
1050 RAHMOUNI ET AL. AJR:160, May1993

Subjects and Methods The signal intensities of all lesions were qualitatively compared
with the signal intensities of the surrounding vertebral bodies on
Thirty-two patients with newly diagnosed, biopsy-proved multiple STIR and Ti- and T2-weighted SE images. The hypenintense foci
myeboma underwent MR imaging of the spine before treatment. All on Ti-weighted SE images corresponding to hypointense foci on
patients had intense back pain with limited motion (grade 2 or more, STIR images were considered to show fatty tissue replacement.
according to the World Health Organization grading system), and 15 The signal intensity of the largest lesion was compared with that of
had radicubar pain. The patients were 38-78 years old (median age, surrounding bone on dynamic enhanced turbo-FLASH and delayed
58 years); 17 were women and 15 were men. Seventeen patients contrast-enhanced Ti-weighted SE images. With dynamic en-
had immunogbobulin G myeloma, nine had immunoglobulin A hanced turbo-FLASH images, the delay before enhancement of the
myeboma, four had Bence Jones proteinunia, and two were non- lesion was also assessed.
secretors of immunogbobulin. Four patients had stage II and 28 had
stage III disease, according to the Dune and Salmon staging system
[1 6], which is based on hemoglobin levels, serum levels of calcium,
, Results
rate of production of monoclonal component, and bone lesions
shown on radiographs. In two patients with stage II disease, no vertebral abnor-
Imaging was done with a 1 .5-T magnet (Magnetom SP 63; Sie- malities could be detected on MR images. In the remaining
mens, Erlangen, Germany) and a surface coil. The following MR 30 patients, three types of bony involvement were detected
images were obtained in succession: Ti-weighted (500/15 [TR/ (Figs. 1 and 2): heterogeneous involvement, with small
TE]) SE images; T2-weighted (2000/45,90) SE images; and STIR lesions no more than 1 cm in diameter (two patients);
(2200/i40/15 [TR/Tb/TE]) images, with the TI of 140 msec chosen involvement, with lange lesions more than 1 cm in diameter
in order to suppress signal from fatty tissue. Imaging parameters (14 patients); and mixed involvement, with associated large
for these SE images were identical: 12 sagittal interleaved contigu-
and small lesions (14 patients). Except in one patient, MR
ous slices of 5-mm thickness, 45-cm field of view, 300 x 51 2 matrix,
findings did not change the staging of the disease. The dis-
two excitations with half-Fourier transform, and vertical phase
encoding or use of a coronal presaturating band i 0 cm thick ante- ease in this patient was initially graded as stage II. However,
nor to the spine to decrease flow and motion artifacts. In order to a stage Ill classification was considered in view of the
perform a dynamic study of the contrast enhancement of myeloma numerous foci shown on MR images.
lesions, the turbo-FLASH technique was combined with administra- All lesions had similar, homogeneous signals on MR
tion of a contrast agent. Short data acquisition times are possible images obtained by using the same parameters. On T2-
with this gradient-echo sequence because short TRs and small flip weighted SE images, all the lesions were hypenintense com-
angles are used. In order to generate Ti contrast, a i800 pulse is pared with surrounding bone tissue (Figs. iA and 2A). On
used to invert the spins magnetization before one short image Ti-weighted SE images, the lesions were isointense or
acquisition. In order to obtain heavily Ti-weighted turbo-FLASH
hypenintense compared with surrounding vertebrae in four of
images, the TI (defined as the time between the 180 inversion
the 30 patients with vertebral abnormalities shown on MR
pulse and the beginning of the sequence) was set at 400 msec [7].
One single-shot axial turbo-FLASH image (7/400/3 [TRITIITE], 25- images (Fig. 1 B) and hypointense in the remaining 26
cm field of view, 128 x 128 matrix, and 10-mm slice thickness) was patients (Fig. 2B). All the lesions were markedly hypenin-
acquired in the plane of the largest bone lesion seen on SE tense on STIR images; the lesions were the most intense
images. Because the small available matrix size resulted in poor areas on the image (Fig. 1C). No additional lesions were
spatial resolution for large fields of view, no sagittal turbo-FLASH detected on STIR images.
images were acquired. On axial turbo-FLASH images, all the lesions were mark-
A bolus injection of 0.2 mI/kg (0.1 mmol/kg) of gadolinium tetra- edly and homogeneously enhanced in the arterial phase,
azacycbododecane tetraacetic acid (Gd-DOTA; Laboratoire Guerbet, 1 0-20 sec after injection of a bolus of contrast material
Aulnay-sous-Bois, France) in a concentration of 0.5 mob/l was
(Figs. i D and 1 E). On delayed turbo-FLASH images, 110-
injected through an antecubital vein; this was followed by a rapid
130 sec after bobus injection, and on contrast-enhanced Ti-
infusion of saline. With the same parameters, axial turbo-FLASH
imaging was started in the plane of the largest bone lesion seen on weighted SE images, a persistence of paramagnetic con-
T2-weighted SE images. Forty successive axial turbo-FLASH trast agent was always visible within the lesions when com-
images separated by 2-sec delays were then acquired during 130 pared with the appearance on unenhanced images (Figs. 1F
sec, allowing a dynamic study of the lesions hemokinetics. and 2C). However, no lesions were seen on contrast-
Four minutes after the bolus injection, sagittal Ti -weighted SE enhanced Ti -weighted SE images that had not been seen
(500/15) images were acquired. When epidural or soft-tissue on T2-weighted SE images (Fig. 2C).
involvement was present, additional axial
Ti-weighted SE images Narrowing of the spinal canal was caused by compression
(540/15, 30-cm field of view, 128 x 256 matrix, 1 8 contiguous inter-
fracture of a vertebral body in three cases and by epidural
leaved 5-mm-thick slices) were used to determine the exact lateral
involvement in five cases (Fig. 2). Soft-tissue involvement
port for radiation therapy.
Because most of the patients had innumerable tumor nodules was seen in six patients, five of whom had epidunal involve-
and because the MR slice thickness was 5 mm, the only lesions ment. Extension to the surrounding soft tissues was always
used for evaluations of signal intensity were those that were more associated with large bone lesions nearby. The contrast
than 1 cm in diameter. Images were analyzed by two radiologists, enhancement of soft-tissue masses was identical to that of
with final conclusions reached by consensus, as follows. the bone lesion.
AJR:160, May 1993 MR DETECTION OF MULTIPLE MYELOMA iOSi

Fig. 1.-Newly diagnosed myeloma

in a 43-year-old woman with back pain
at thoracic level.
A, T2-welghted (2000/90) SE MR im-
age shows hypenintense macronodu-
lar lesions of body and posterior arch
of TI 0 vertebra (arrows) and several
tiny disseminated mlcronodular foci.
B, Ti-weighted (500/15) SE MR im-
age shows that largest lesion is
isointense compared with surround-
ing vertebra.
C, STIR (2200/140/15 (TR/Tl/TE])
MR image shows that largest lesions
are markedly hypenintense, but no ad-
ditional lesions are visualized when
appearance is compared with that on
T2-weighted SE MR image (A).
D, Unenhanced axial turbo-FLASH
(7/400/3 [TR/11/TE]) MR image at level
of largest lesion does not show lesion.
E, Turbo-FLASH (7/400/3) MR image
obtained 1 0 sec after injection of con-
trast material clearly shows hypervas-
cularizatlon of two lesions as
homogeneously hypenintense areas
(arrows).
F Turbo-FLASH (7/400/3) MR image
obtained 120 sec after injection of con-
trast material shows persistence of
paramagnetic contrast agent within le-
slons.

Fig. 2.-Newly diagnosed myeloma

in a 62-year-old man with Intense back
pain associated with radicular pain.
Because of multiple threatening ma-
cronodular lesions and epidural in-
volvement, radiotherapy was started
Immediately.
A, T2-weighted (2000/90) SE MR Im-
age shows massive involvement of T4
and T6 (arrows) with collapsed verte-
bral bodies and epidural involvement.
All macronodular lesions are hyperin-
tense. Tiny hyperintense foci can be
seen In all vertebral bodies also, repre-
senting mixed Involvement with asso-
dated small and large lesions.
B, Unenhanced Ti-weighted (500/
15) SE MR image shows lesions as hy-
pointense.
C, Ti-weighted (500/15) SE MR im-
age acquired 4 mm after IV injection of
contrast material. When compared
with appearance on unenhanced Ti -
weighted SE image (B), lesions are
markedly enhanced. No additional le-
sions can be seen when appearance is
compared with that on T2-weighted SE
image (A).
1052 RAHMOUNI ET AL. AJR:160, May1993

Discussion contrast enhancement of myeboma lesions. The uptake of

The spine is the most common site of myeloma lesions contrast material by all lesions occurred 10-20 sec after the
bolus injection, in the arterial phase (Fig. 1 E). This finding is
[1]. MR imaging can be considered the gold standard for the
examination of patients with suspected epidunal involvement consistent with the angiognaphic appearance of myeboma
lesions [ii]. The persistence of contrast agent was always
[3]. In our study, use of MR imaging for diagnosis and debin-
eation of epidunal involvement bed to prompt planning of evident on contrast-enhanced Ti-weighted SE images (Fig.
2C). This persistent enhancement could be rebated to vascu-
treatment in five patients. Local radiotherapy was started in
Ian abnormalities or to newly formed extravascular spaces.
three patients because paraplegia was imminent (Fig. 2). A
regimen including anthnacycline was started in two patients In conclusion, MR imaging is useful in detecting spinal
myeboma, particularly epidunab involvement. Our results con-
less than 55 years old so that they later could have total
firm that the detection of myeboma foci within the spine is
body irradiation and intensified chemotherapy if needed.
The Dune and Salmon staging system highlights the need best achieved by using STIR and T2-weighted images. Con-
trast-enhanced MR images show enhancement of all the
to detect foci of myeboma. In two studies [2, 4], myeboma
lesions but are useless for the detection of additional
tumor nodules were visualized as hypenintense foci within
myeboma lesions. Signal intensities and contrast enhance-
the vertebrae on T2-weighted SE images. Some authors [8]
ment of all myeloma foci of the spine appear to be similar in
have described a different appearance of myeloma lesions
all patients. However, treatment may affect the signal inten-
on T2-weighted SE images. However, no data were avail-
able about the treatments given to those patients with krwn sity on contrast enhancement of the lesions.
myeloma. This variable appearance could have been due to
signal changes associated with treatment. We found that REFERENCES
untreated myeboma lesions were always hypenintense on
1 . Salmon SE, Cassady JR. Plasma cell neoplasms. In: Do vita VT, Hellman
T2-weighted SE images. Libshitz et ab. [4] have differentiated 5, Rosenberg SA, eds. Cancer: principles and practice in oncology, 3rd
tumor nodules from diffuse infiltration caused by myeloma. od. Philadelphia: Lippincott, 1989:1853-1895
In cases of diffuse infiltration, these authors suggest that 2. Siegal T, Siogal T. Spinal opidural involvement in haematological tumors:
absence of fatty replacement on a generalized decrease in clinical features and therapeutic options. Leuk Lymphoma 1991 5:
101-110
signal intensity on Ti -weighted SE images can also suggest
3. Fnuehwald FX, Tscholakoff D, Schaighofor F, et al. Magnetic resonance
multiple myeboma. The fatty tissue replacement occurring imaging of the lower vertebral column in patients with multiple myoloma.
within the vertebrae in olden patients is a reason to use fat- Invest Radiol i988;23:1 93-1 99
suppressed MR sequences for detecting myeloma lesions 4. Libshitz HI, Malthouse SR, Cunningham D, MacVicar AD, Husband JE.
Multiple myoloma: appearance at MR imaging. Radiology i992;182:
[9, 1 0]. Our results showed that no additional focal lesions
833-837
were detected on STIR images (Fig. 1C). However, the 5. Haase A. Snapshot FLASH MRI: applications to Ti , T2 and chemical shift
lesions seen on T2-weighted SE images were always seen imaging. Magn Reson Med 1990;13:77-89
on STIR images and appeared as the most intense areas of 6. Dune BGM, Salmon SE. A clinical staging system for multiple myoloma
the image. These results confirm the reports of increased correlation of measured cell mass with presenting clinical features,
response to treatment and survival. Cancer 1975;36:842-852
conspicuity of these lesions on STIR images [10]. Our study 7. Mathieu D, Rahmouni A, Anglade MC, of al. Focal nodular hyperplasia of
also shows the limited value of Ti -weighted SE images, on the liven: assessment with contrast enhanced turbo-FLASH MR imaging.
which myeboma lesions have a variable appearance and Radiology 1991180:25-30
sometimes cannot be detected (Fig. 1 B). 8. Daffner RH, Lupetin AR, Dash N, Doob ZL, Sefczok RJ, Schapiro RL. MRI
in the detection of malignant infiltration of bone marrow. AJR
The role of contrast agents in the detection of myeboma
1986;146:353-358
lesions on MR images has not been previously investigated. 9. Dwyen AJ, Frank SL, Sank VJ, Reinig JW, Hickoy AM, Doppman SL.
Turbo-FLASH images are suitable for studying dynamic con- Short TI inversion recovery pulse sequence: analysis and initial oxpori-
tnast enhancement of normal and abnormal tissues because once in cancer imaging. Radio/ogy 1988;168:827-836
the short TEs reduce T2-weighted effects and the acquisition 10. Ricci C, Cova M, Kang YS, of al. Normal age-related patterns of cellular
and fatty bone marrow distribution in the axial skeleton: MR imaging
time is short. We have combined this technique (by using a
study. Radiology 1990;1 77:83-88
1 80#{176}
prepulse and a TI of 400 msec, which yields heavily Ti -
11. Launin 5, Akonman M, Kindblom LG, Guntenbeng B. Angiognaphy in
weighted images) with the injection of Gd-DOTA to study the myeloma (plasmocytoma). Skeletal Radiol i979;4:8-i 8

The readers attention is directed to the companion article that appears on pages 1053-1057.