Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
1043
Detection Tumors:
of Malignant MR Imaging
Bone
vs Scintigraphy
:.
.
Joseph A. Frank12 Alexander Nicholas J. Patronas1’2 Jorge A. Carrasquillo3 Klara rv’2 Ann M. Hickey1 Andrew J. Dwyer1
2
One hundred six patients with a known or suspected diagnosis of bone cancer (11 patients with biopsy-proved primary tumors, 95 patients with metastatic disease) were evaluated with scintigraphy and MR imaging to determine the relative sensitivity of each technique in the detection of bone disease. MR imaging was performed at 0.5 T as part of the entry evaluation into Intramural Research Board protocols (30%), for evaluation of cord compression, or because of an equivocal scintigram. MR was performed with Ti-weighted (e.g., 300-500/10-20 [TRITE]), T2-weighted (e.g., 2000/80) spin-echo (SE), and a short-TI inversion recovery (STIR) pulse sequence Scintigrams were performed with “Tc-methylene diphosphonate. A retrospective analysis showed that in 30 (28%) of 106
patients,
MR
imaging
performed
over
a limited
region
of interest
revealed
a focal
consistent with tumor that was not observed on scintigraphy. Only one patient had an abnormality on scintigraphy, caused by a metastasis, that was not found on MR images. In 73 (69%)of the i06 patients, the results of MR imaging and scintigraphy were equivalent; in 41 cases results of both techniques were normal. A McNemar analysis of the discordant cases showed MR imaging to be more sensitive than scintigraphy was (p < .001). abnormality
Our
results
suggest
that
although
MR
imaging
has
a greater
sensitivity
in detecting
focal disease, scintigraphy is still the most useful screening test for evaluating the entire skeleton. MR imaging should be reserved for clarification of scintigraphic findings when suspicion is high for tumor. AJR
155:1043-1048,
November
1990
Accurate evaluation of the presence and extent of malignant bone tumors, both primary and metastatic, is crucial to proper staging and treatment of these diseases. Various imaging techniques are available for the evaluation of bone metastases, including plain film radiographs, Cl, scintigraphy, and MR imaging [1 2]. Scintigraphy is the most commonly used imaging technique for the staging and evaluation of bone metastases [3, 4]. MR imaging has recently been shown to be a sensitive technique for the detection of tumor involvement within the bone; however, only a few studies have compared the findings on scintigraphy with those on MR imaging [5-9]. The purpose of this study was to determine the relative sensitivities of MR imaging and scintigraphy for detecting primary malignant bone tumors and bone metastases. ,
Received March 5, 1990; accepted after revision June4, 1990. 1 Department of Diagnostic Radiology, Bldg. 10, Rm. 1 C660, National Institutes of Health, Bethesda, MD 20892. Address reprint requests to J. A. Frank. 2 Department of Radiology, Georgetown University Medical
Center,
Washington,
3 Department of Nuclear stitutes of Health, Bethesda,
0361-803X/90/1
© American
DC 20007.
Medicine, National MD 20892.
555-1043 Roentgen Ray Society
In-
Materials
and
Methods
We retrospectively reviewed MR images and scintigrams obtained from 1 986 to 1988 in 1 06 patients who were being considered for study or as part of entry criteria onto National Cancer
Institute
clinically
indicated
findings
on
Intramural
follow-up
scintigraphy
Research
examinations of the
primary
Board-approved
protocols
(32
for spinal cord compression, or
metastatic
disease
included in this study only if MR imaging and scintigraphy
(74
patients)
or who
had
focal pain, or equivocal patients).
were performed
Patients
were
within 36 days of
1044
FRANK
each other. raphy were 1 06 patients, 13% of the on the same
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
this
study
In more than 80% of the cases, MR imaging and scintigperformed within 20 days of each other. In 64 (60%) of MR imaging was performed after scintigraphy, and in patients, MR imaging and scintigraphy were performed day. Patients and lesions were considered evaluable for only
to
were examined
the
extent
that
with scintigraphy
comparable
regions
of
anatomy
and MR imaging (i.e., if the lumbar
spine was examined for cord compression by was compared only with the corresponding scintigram). In 91 (86%) of the 1 06 patients, two techniques involved regions of the spine (14%) of the patients’ examinations limited to long bones in the upper and lower extremities. The demographic data and distribution of per tumor type are listed below. Forty-one
MR imaging, the finding lumbar region on the the comparison of the and pelvis, with the rest examination of only the
the number of patients males and 65 females ranged in age from 4 to 71 years (mean 41 ± 1 4 years). Most (90%)
patients
studied
had proved
or suspected
metastatic
tumor
to the
bone; the rest had a primary malignant tumor of bone or bone marrow. Twenty-three patients had breast cancer; 1 6 patients had lymphoma (both Hodgkin and non-Hodgkin); 15 patients had hypernephroma; 1 5 patients had various sarcomas; 1 3 patients had Ewing sarcoma; six
patients
had
melanoma; esophageal carcinoma
adenocarcinoma
of
the
colon;
four
patients
had
three patients had prostatic cancer; two patients had cancer; and one patient each had rectal cancer, adenowith
an unknown
primary
tumor,
germ cell tumor, adrenal cell carcinoma, non-small-cell
lung
MR images
cancer,
and
were obtained
acute
small-cell
hepatoma,
lung
cancer,
testicular
cancer,
leukemia.
on a 0.5-T Picker
International
Vista HP
MR unit (Highland Heights, OH) by using a Ti-weighted spin-echo (SE), 300-500/i 0-26 (TR/TE) sequence (SE 300/i 0), a T2-weighted sequence
(SE
and an inversion
2000/80),
recovery
(STIR)
sequence
(IR i 500/i 00) [i 0]. MR images were obtained in up to three planes, as dictated by the evaluation protocol. MR imaging was performed with
5- to 10-mm contiguous cm and i 28 phase-encoded
slices with fields of view from 20 to 50 steps, resulting in a i -4 mm2 pixel. MR
images were displayed on a Bone scans were obtained injection of Tc-methylene i 8 years old received 0.3
256 x 256 image matrix. by using a gamma camera 2-3 hr after diphosphonate. Patients younger than mCi/kg (1 i .i MBq/kg), and all adults
received
25 mCi
(925
MBq).
MR images and scintigrams were reviewed separately by two to three radiologists who were blinded to findings of related imaging tests, then jointly reviewed and classified according to relative performance of corresponding images. Focal or diffuse signal intensity on T2-weighted or STIR images intensity
on Ti -weighted
images
relative
to signal
areas with high and low signal
intensity
of normal
bone or bone marrow were considered abnormal when shown in two different scanning planes or two distinct pulse sequences. On scintigrams, areas of excess radiotracer uptake and photopenic areas were considered abnormal. Because many studies identified multiple lesions, these studies were considered positive only if they showed all abnormal foci identified by either MR imaging or scintigraphy. Scintigraphically corresponding
detected lesions outside MR images were excluded.
scans were confirmed or documented up bone scans were
performed
compared imaging
in this study. and/or
biopsies,
of view of the shown on the
imaging studies,
clinical progression of response to therapy on followor MR images. Most of the follow-up examinations within 6 months of the MR imaging or scintigraphy
Confirmation
scintigraphy
scans were discordant, histologic
with bone marrow
the field Lesions
confirmation,
was
of lesions
depicted
by MR
sought
only for patients whose 33 (31 %) of i06 patients. Follow-up studies, or both were available for 24 (73%) of 33
patients. The presumption made was that all focal lesions on scintigrams or MR images were considered tumor unless proved otherwise.
El
AL.
AJR:155,
Patients
whose
MR
imaging
studies
either the brain or liver were excluded was not performed holding techniques
comprised
November
only
1990
images
of
from this study. MR imaging
in the appropriate scanning plane or with breathto evaluate the ribs for tumor involvement. MR
images of the head also were not included in the study. Statistical
The McNemar
analysis
was
performed
test is a useful
by using
method
studies (i.e., the relative sensitivity in the detection
[1 1]. By using serve
of diffuse
or focal
the McNemar
a McNemar
for comparing
test
paired
[i 1].
data
or
of MR imaging and scintigraphy) involvement
test,
of the bone
we can assume
with
tumor
that the patients
as their own control subjects.
Results The appearance of malignant lesions on MR images deponds on the pulse sequence used. On 11 -weighted images, tumors appear hypointense (i.e., dark) compared with the normal marrow (Figs. 1 and 2). On 12-weighted images, metastatic disease can appear isointense (i.e., similar in signal intensity) to normal marrow [1 2]. Focal lesions on the STIR images, because of the suppression of the fat signal, appeared bright and were easily differentiated from the red and yellow marrow [1 2-1 4] (Fig. 2 and 3). In 41 (39%) of the 1 06 patients, no tumor involvement of the bone was detected in the regions of interest covered by both techniques. In 32 (30%) patients, both techniques had positive findings and showed similar anatomic distribution of the tumor involvement in the bone. MR imaging showed areas of presumed metastatic tumor involvement in the bone not observed on scintigraphy in 30 (28%) of the 1 06 patients. The distribution of the number of patients per tumor type is as follows: eight patients with breast cancer, six patients with various sarcomas, six patients with lymphoma, four patients with renal cell carcinoma, four patients with Ewing sarcoma, and one each with a germ cell tumor and leukemia. In 16 (53%) of these 30 cases in which MR imaging findings were positive and scintigraphic findings were negative, MR imaging was performed 1 -35 days after scintigraphy either to resolve a focal clinical problem (i.e., pain, equivocal findings) or for the evaluation of cord compression. Twenty-one (7O%) of the 30 patients in whom MR imaging findings were positive and scintigraphic findings were negative had follow-up imaging studies as a means of confirmation 0-6 months after the comparative studies (Fig. 1). Four of the 21 patients had biopsy-proved malignant tumors. In nine of these 30 patients, findings on scintigraphy were normal, whereas MR images showed focal areas of abnormal signal intensity that were presumed to be tumors; pathologic confirmation was not available. In almost all 30 cases, MR imaging was performed in the coronal or sagittal plane, which efficiently encompassed the greatest extent of bone at risk on a single image. Scintigraphic findings were positive in two cases (Ewing sarcoma, colonic carcinoma) and indeterminate in one case (lymphoma) in which MR imaging findings were negative. Only one of these three cases (colonic carcinoma) was subsequently proved to be a metastatic lesion (Fig. 4) on the basis of follow-up imaging studies. The other two cases were considered clinically and radiologically compatible with stress or compression fractures.
AJR:155,
November
MR
1990
AND
SCINTIGRAPHY
OF
BONE
TUMORS
104
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
-
.
..
..
.
II A
Fig. 1.-l 1-year-old boy initially presenting with a primary Ewing sarcoma of fibula. MR imaging showed another focus of tumor that was observed on scintigram obtained 3 months after initial examination. A, Scintigram shows a solitary metastasis to right femoral diaphysis. B, Ti-weighted SE MR image, 300/26, obtained I 1 days after scintigraphy clearly shows a smaller satellite lesion (arrow) and tumor in contralateral femur. C, Scintigram obtained 3 months after initial scintigraphy shows second focus of tumor. Patient later died of diffuse metastatic Ewing sarcoma.
Fig. 2.-46-year-old
man with renal
cell carcinoma. MR imaging has greater sensitivity for metastases than scintigraphy does. A and B, MR imaging performed as part of staging evaluation reveals vertebral lesion (arrow) on STIR (IR 1500/ 100) (A) and Ti-weighted SE 433/26 (B) MR images. C, Posterior scintigram obtained 1 day after MR imaging is normal.
The two-sided McNemar test demonstrated that the overall positive rates for a case-by-case analysis of MR imaging differed from scintigraphy at p < .001 . The advantage of the McNemar test in the analysis of matched comparisons is its sole dependence on discordant cases (i.e., MR imaging findings positive, scintigraphic findings negative or MR imaging findings negative, scintigraphic findings positive). If we disre-
gard the nine (30%) of 30 unconfirmed (either by biopsy or follow-up examination) cases in which MR imaging findings were positive and scintigraphic findings were negative, the relative sensitivity of MR imaging differed from that of scintigraphy at p < .003. As a worst case, MR imaging was shown to have a higher sensitivity than scintigraphy did for all bony lesions in patients at risk of metastatic disease.
1046
FRANK
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
Discussion MR imaging has been shown to be sensitive for the detection of bone marrow metastases [5-1 0, 1 2, 1 3-1 6]. Normal red and yellow marrow can be distinguished on an MR image because of differences in signal intensity or morphology [16,
-
Fig. 3.-58-year-old man with nodular poorly differentiated lymphoma. MR imaging was performed as a basis for monitoring the patient’s response to therapy. A, Coronal STIR MR image obtained for evaluation of retroperitoneal adenopathy clearly shows multiple focal thoracolumbar vertebral lesions (arrows). B, Posterior scintigram of spine obtained 2 days later is unremarkable. Rib lesion, not seen on MR, is evident.
El
AL.
AJR:155,
November
1990
17]. A significant variety of diseased tissues (e.g., tumor, inflammation, and infection) have relaxation times that differ from normal marrow to a similar degree, resulting in similar appearances on MR images [1 4, 1 8-20] and a lack of histologic specificity. Scintigraphy is a sensitive but not specific imaging technique for the detection of malignant tumors because benign conditions such as degenerative joint disease, trauma, and infection can result in positive test results. The lack of histologic specificity stems from the mechanisms of isotope uptake (“hot spot”) or lack of tracer uptake (photopenic) areas on the bone scan. Hot spots predominately reflect the pattern of bone response to a local insult, and focal decreases in skeletal tracer uptake result from bone destruction without repair [21]. We found MR imaging to be more sensitive than scintigraphy was for the detection and delineation of bone tumor. A possible explanation for MR imaging’s superiority is that hematogenously seeded intramedullary metastasis produces detectable lesions by replacement of marrow. This occurs before either intrinsic or reactive metabolic changes in cancellous and cortical bone can be detected scintigraphically or radiologically. Usually, destruction of bony trabeculae and mobilization of osteoclastic reparative mechanisms are prerequisites for positive findings on a scintigram [21]. In a preliminary study examining the extent of bone marrow involvement in patients with small-cell lung carcinoma, 10 (42%) of 24 patients who were considered to have limited disease by standard staging criteria (including scintigraphy) had their stage of disease changed to extensive disease after MR imaging [9]. In these cases, focal or diffuse abnormalities in signal intensity were noted on Ti -weighted images of the pelvis and proximal femur not observed on bone scans. In a study of 50 patients with malignant tumors and suspected skeletal involvement, MR imaging was shown to have a lower false-positive rate than scintigraphy and 1 00% sensitivity and specificity in patients with multiple myeloma with marrow
Fig. 4.-44-year-old man with metastatic adenocarcinorna of colon. This is only before MR imaging. A, STIR MR image fails to reveal posterior acetabulum lesion. B, Scintigram made I month before MR image shows lesion (arrow). C, MR image obtained 4 months after scintigraphy shows lesion (arrow).
case
in this
series
in which
scintigraphy
showed
a malignant
lesion
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
AJR:155,
November
1990
MR
AND
SCINTIGRAPHY
involvement [6]. There are also reports of patients with lymphoma in whom MR imaging showed marrow involvement despite a normal iliac crest bone marrow biopsy [7, 8]. These studies support our results. It could be argued that during the 36-day (maximum) interval between scintigraphy and MR imaging, a new malignant lesion could develop, which would result in positive findings for one scanning technique while the other showed negative findings. In addition, delay between the two studies might introduce a selection bias. In only nine of the 31 malignant cases in which the MR imaging findings differed from scintigraphy was there a delay of more than 20 days between the two examinations. Deleting these patients from the statistical analysis would not change the overall results of this study (p < .003). If an MR imaging examination was ordered or not obtained on the basis of results of a scintigram, then it is possible that the number of cases of positive scintigraphic/ negative MR imaging findings reported may underestimate the true incidence offalse-negative MR imaging examinations. In 16 (53%) of the 30 cases with negative scintigraphic findings and positive MR imaging findings, MR imaging was performed after scintigraphy. If these studies were excluded from the analysis, the sensitivity of MR imaging would differ from that of bone scanning at p < .01. Under ideal circumstances, a comparison study would relate results of two imaging techniques to the true condition of a patient at every lesion described by either test. Unfortunately, although histology is a universally accepted estimate of “truth,” percutaneous biopsy of even a single site, much less at every one of multiple sites, has not been possible in our population of patients. Noninvasive assessment through follow-up imaging studies, although frequently available in our series, is burdened by ambiguity-if a marrow lesion disappears during therapy, did it represent a metastasis or a focal site of “reconverted” red marrow responding to cytotoxic drugs or radiation? Observing progression toward a subsequently unambiguous state without treatment is universally feasible only in an animal study. It is important to note that red marrow can have an appearance on MR images similar to focal or diffuse tumor involvement depending on lesion location and age of the patient. In patients with aplastic anemia, focal islands of hematopoietic marrow can have signal-intensity pattern and distribution mimicking metastatic disease [22]. We encountered insufficient numbers of patients with predominantly red marrow to establish whether our overall results would apply to this group [1 7]. Distinction between hyperplastic red marrow, particularly when focally deposited, and neoplasm may be difficult with MR imaging. Although only additional study of such patients can resolve this issue, behavior similar to that exhibited by metastases in response to therapy, that is, either replacement with fatty marrow [23] or overpopulation with hematopoietic elements (i.e., stress) after chemotherapy [24], also may complicate both experimental design and clinical diagnosis. The high sensitivity of MR imaging in relation to that of scintigraphy by itself does not alter scintigraphy’s role as the most useful screening examination for primary and metastatic disease. Scintigraphy examines 1 00% of the skeleton, is
OF
BONE
TUMORS
1047
widely available, and at present is less expensive than MR imaging. The use of MR imaging to evaluate skeletal metastases is limited by its expense, lesser availability, and regional purview. Lesions of the ribs and skull were excluded from this study principally because so few images included them. However, rib metastases might prove difficult to show on MR images because of respiratory motion, paucity of at-risk tissue (red marrow), or lack of a convenient imaging plane (Fig. 3). Metastasis to the skull would normally be included on a separate MR imaging examination of the brain; these were not reviewed as part of this study. Although MR imaging as currently practiced is not an efficient technique for screening the entire skeleton, we find it extremely valuable in clarifying equivocal or (suspected) falsely positive or negative bone scans. Because of MR imaging’s inherent anatomic detail, we also may find it useful as a localization technique for directed biopsies of suspected lesions [5-8]. Future improvement in MR imaging techniques (e.g., development of faster scanning techniques or surface coils) may make MR imaging an efficient screening technique for detecting and staging bone marrow metastasis.
ACKNOWLEDGMENTS The authors thank Anna Scheib for her preparation of the manuscript. In addition, the authors acknowledge the work of the MR
technologists
who performed
the studies.
REFERENCES 1 . Volger JB, Murphy WA. Bone marrow imaging. Radiology 679-693 2. Kagan AR, Bassett LW, Gold RH. Radiologic contributions AJR 1986;147:305-312 3. McNeil BJ. Value of bone scanning 1984;14:277-286
i988;168:
to cancer
management.
4. Krishnamurthy metastatic
bone
in neoplastic
disease. Semin Nucl Med
GT, Tubis M, Hiss J, Bland WH. Distribution pattern of disease: a need for total body skeletal image. JAMA
1977;237:2504-2506
5. Khurana JS, Rosenthal Dl, Rosenberg AE, Mankia HJ. Skeletal metastases in liposarcoma detectable only by magnetic resonance imaging. Clin Orthop 1989;243:204-207 6. Daffner R, Lupetin A, Dash N, Beeb Z, Sefczek R, Shapiro A. MRI in the detection of malignant infiltration of bone marrow. AiR 1986;146:353-358 7. Dohner H, Guckel F, Knauf W, et al. Magnetic resonance imaging of bone marrow in lymphoproliferative disorders: correlation with bone marrow biopsy. Br J Haematol 1989;73: 12-17 8. Linden A, Zankovich R, Theissen P. Diehl V. Schicha H. Malignant lymphoma: bone marrow imaging versus biopsy. Radiology i989;173:
335-339 9. Carney DN, Redmond
0, Harford P, Stark J, Ennis J. Bone marrow involvement (BMI) by small cell lung cancer (SCLC) using magnetic resonance imaging. Proc Am Soc Clin Oncol 1989;8:228 10. Dwyer AJ, Frank JA, Sank VJ, et al. Short TI inversion recovery pulse sequence: analysis and initial experience in cancer imaging. Radiology 1988;168:827-836 1 1 . Conover WJ. Practical nonparametric statistics, 2nd ed. New York: Wiley, 1980:130 12. Porter BA, Shields AF, Olson DO. Magnetic resonance marrow disorders. Radiol Clin North Am 1986;24:269-289
imaging of bone
13. Shields AF, Porter BA, Churchley 5, et al. The detection of bone marrow involvement by lymphoma using magnetic resonance imaging. J Clin Oncol 1987;5:225-230
14. Frank JA, Dwyer AJ, Doppman JL. Magnetic resonance imaging in oncology. In: DeVita VT, Hellman 5, Rosenberg SA, eds. important advances in
FRANK
1048
oncology
1987.
Philadelphia:
Lippincott,
i987:
133-1
Invest Radiol 1988;23: 193-1 99 Smoker WRK, Godersky JC, Knutzon RK, Keyes WD, Norman
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
W. The role of MR imaging in evaluating
metastatic
D, Bergman spinal disease. AJR
i987;149:1241-1248
17. Kricun ME. Red-yellow solitary
bone lesions.
18. Unger E, Moldofsky osteomyelitis
marrow
its eftect on the location of Skeletal Radiol 1985;1 4:10-19 P, Gatenby R, Hartz W, Broden G. Diagnosis of
by MR imaging.
conversion:
Radiology
1987;1 64:449-454
19. Mitchell DG, Rao VM, Dalinku MK, et al. Femoral head avascular necrosis: correlation
of MR imaging,
radiographic
staging,
AL.
radionuclide
imaging
and
AJR:155, November 1990
20.
clinical finding. Radiology 1987;162:709-715 Medic MT. Masaryk TJ, Paushter D. Magnetic spine. Radiol Clin North Am 1986;24:229-245
21.
Charkes
74
15. Fruehwald FJ, Tscholakoff D, Schwaighofer B, et al. Magnetic resonance imaging of the lower vertebral column in patients with multiple myeloma. 1 6.
El
ND.
Mechanisms
of
skeletal
resonance
tracer
uptake.
imaging J
Nucl
of the Med
1979;20:794-795 22. Kaplan PA, Asleson RJ, Klassen LW, Duggan MJ. Bone marrow pattems in aplastic anemia: 1987;1 64:441 -444
observations
with
1 ST
MR
imaging.
Radiology
23. Ramsey RG, Zacharias CE. MR imaging of the spine after radiation therapy: easily
recognizable
effects.
AJR 1985;144:
1131-1135
24. McKinstry CS, Steiner RE, Young AT, Jones L, Swirsky D, Aber V. Bone marrow in leukemia and aplastic anemia: MR imaging before, during, and after treatment. Radiology i987;162:701-707
1043
Detection Tumors:
of Malignant MR Imaging
Bone
vs Scintigraphy
:.
.
Joseph A. Frank12 Alexander Nicholas J. Patronas1’2 Jorge A. Carrasquillo3 Klara rv’2 Ann M. Hickey1 Andrew J. Dwyer1
2
One hundred six patients with a known or suspected diagnosis of bone cancer (11 patients with biopsy-proved primary tumors, 95 patients with metastatic disease) were evaluated with scintigraphy and MR imaging to determine the relative sensitivity of each technique in the detection of bone disease. MR imaging was performed at 0.5 T as part of the entry evaluation into Intramural Research Board protocols (30%), for evaluation of cord compression, or because of an equivocal scintigram. MR was performed with Ti-weighted (e.g., 300-500/10-20 [TRITE]), T2-weighted (e.g., 2000/80) spin-echo (SE), and a short-TI inversion recovery (STIR) pulse sequence Scintigrams were performed with “Tc-methylene diphosphonate. A retrospective analysis showed that in 30 (28%) of 106
patients,
MR
imaging
performed
over
a limited
region
of interest
revealed
a focal
consistent with tumor that was not observed on scintigraphy. Only one patient had an abnormality on scintigraphy, caused by a metastasis, that was not found on MR images. In 73 (69%)of the i06 patients, the results of MR imaging and scintigraphy were equivalent; in 41 cases results of both techniques were normal. A McNemar analysis of the discordant cases showed MR imaging to be more sensitive than scintigraphy was (p < .001). abnormality
Our
results
suggest
that
although
MR
imaging
has
a greater
sensitivity
in detecting
focal disease, scintigraphy is still the most useful screening test for evaluating the entire skeleton. MR imaging should be reserved for clarification of scintigraphic findings when suspicion is high for tumor. AJR
155:1043-1048,
November
1990
Accurate evaluation of the presence and extent of malignant bone tumors, both primary and metastatic, is crucial to proper staging and treatment of these diseases. Various imaging techniques are available for the evaluation of bone metastases, including plain film radiographs, Cl, scintigraphy, and MR imaging [1 2]. Scintigraphy is the most commonly used imaging technique for the staging and evaluation of bone metastases [3, 4]. MR imaging has recently been shown to be a sensitive technique for the detection of tumor involvement within the bone; however, only a few studies have compared the findings on scintigraphy with those on MR imaging [5-9]. The purpose of this study was to determine the relative sensitivities of MR imaging and scintigraphy for detecting primary malignant bone tumors and bone metastases. ,
Received March 5, 1990; accepted after revision June4, 1990. 1 Department of Diagnostic Radiology, Bldg. 10, Rm. 1 C660, National Institutes of Health, Bethesda, MD 20892. Address reprint requests to J. A. Frank. 2 Department of Radiology, Georgetown University Medical
Center,
Washington,
3 Department of Nuclear stitutes of Health, Bethesda,
0361-803X/90/1
© American
DC 20007.
Medicine, National MD 20892.
555-1043 Roentgen Ray Society
In-
Materials
and
Methods
We retrospectively reviewed MR images and scintigrams obtained from 1 986 to 1988 in 1 06 patients who were being considered for study or as part of entry criteria onto National Cancer
Institute
clinically
indicated
findings
on
Intramural
follow-up
scintigraphy
Research
examinations of the
primary
Board-approved
protocols
(32
for spinal cord compression, or
metastatic
disease
included in this study only if MR imaging and scintigraphy
(74
patients)
or who
had
focal pain, or equivocal patients).
were performed
Patients
were
within 36 days of
1044
FRANK
each other. raphy were 1 06 patients, 13% of the on the same
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
this
study
In more than 80% of the cases, MR imaging and scintigperformed within 20 days of each other. In 64 (60%) of MR imaging was performed after scintigraphy, and in patients, MR imaging and scintigraphy were performed day. Patients and lesions were considered evaluable for only
to
were examined
the
extent
that
with scintigraphy
comparable
regions
of
anatomy
and MR imaging (i.e., if the lumbar
spine was examined for cord compression by was compared only with the corresponding scintigram). In 91 (86%) of the 1 06 patients, two techniques involved regions of the spine (14%) of the patients’ examinations limited to long bones in the upper and lower extremities. The demographic data and distribution of per tumor type are listed below. Forty-one
MR imaging, the finding lumbar region on the the comparison of the and pelvis, with the rest examination of only the
the number of patients males and 65 females ranged in age from 4 to 71 years (mean 41 ± 1 4 years). Most (90%)
patients
studied
had proved
or suspected
metastatic
tumor
to the
bone; the rest had a primary malignant tumor of bone or bone marrow. Twenty-three patients had breast cancer; 1 6 patients had lymphoma (both Hodgkin and non-Hodgkin); 15 patients had hypernephroma; 1 5 patients had various sarcomas; 1 3 patients had Ewing sarcoma; six
patients
had
melanoma; esophageal carcinoma
adenocarcinoma
of
the
colon;
four
patients
had
three patients had prostatic cancer; two patients had cancer; and one patient each had rectal cancer, adenowith
an unknown
primary
tumor,
germ cell tumor, adrenal cell carcinoma, non-small-cell
lung
MR images
cancer,
and
were obtained
acute
small-cell
hepatoma,
lung
cancer,
testicular
cancer,
leukemia.
on a 0.5-T Picker
International
Vista HP
MR unit (Highland Heights, OH) by using a Ti-weighted spin-echo (SE), 300-500/i 0-26 (TR/TE) sequence (SE 300/i 0), a T2-weighted sequence
(SE
and an inversion
2000/80),
recovery
(STIR)
sequence
(IR i 500/i 00) [i 0]. MR images were obtained in up to three planes, as dictated by the evaluation protocol. MR imaging was performed with
5- to 10-mm contiguous cm and i 28 phase-encoded
slices with fields of view from 20 to 50 steps, resulting in a i -4 mm2 pixel. MR
images were displayed on a Bone scans were obtained injection of Tc-methylene i 8 years old received 0.3
256 x 256 image matrix. by using a gamma camera 2-3 hr after diphosphonate. Patients younger than mCi/kg (1 i .i MBq/kg), and all adults
received
25 mCi
(925
MBq).
MR images and scintigrams were reviewed separately by two to three radiologists who were blinded to findings of related imaging tests, then jointly reviewed and classified according to relative performance of corresponding images. Focal or diffuse signal intensity on T2-weighted or STIR images intensity
on Ti -weighted
images
relative
to signal
areas with high and low signal
intensity
of normal
bone or bone marrow were considered abnormal when shown in two different scanning planes or two distinct pulse sequences. On scintigrams, areas of excess radiotracer uptake and photopenic areas were considered abnormal. Because many studies identified multiple lesions, these studies were considered positive only if they showed all abnormal foci identified by either MR imaging or scintigraphy. Scintigraphically corresponding
detected lesions outside MR images were excluded.
scans were confirmed or documented up bone scans were
performed
compared imaging
in this study. and/or
biopsies,
of view of the shown on the
imaging studies,
clinical progression of response to therapy on followor MR images. Most of the follow-up examinations within 6 months of the MR imaging or scintigraphy
Confirmation
scintigraphy
scans were discordant, histologic
with bone marrow
the field Lesions
confirmation,
was
of lesions
depicted
by MR
sought
only for patients whose 33 (31 %) of i06 patients. Follow-up studies, or both were available for 24 (73%) of 33
patients. The presumption made was that all focal lesions on scintigrams or MR images were considered tumor unless proved otherwise.
El
AL.
AJR:155,
Patients
whose
MR
imaging
studies
either the brain or liver were excluded was not performed holding techniques
comprised
November
only
1990
images
of
from this study. MR imaging
in the appropriate scanning plane or with breathto evaluate the ribs for tumor involvement. MR
images of the head also were not included in the study. Statistical
The McNemar
analysis
was
performed
test is a useful
by using
method
studies (i.e., the relative sensitivity in the detection
[1 1]. By using serve
of diffuse
or focal
the McNemar
a McNemar
for comparing
test
paired
[i 1].
data
or
of MR imaging and scintigraphy) involvement
test,
of the bone
we can assume
with
tumor
that the patients
as their own control subjects.
Results The appearance of malignant lesions on MR images deponds on the pulse sequence used. On 11 -weighted images, tumors appear hypointense (i.e., dark) compared with the normal marrow (Figs. 1 and 2). On 12-weighted images, metastatic disease can appear isointense (i.e., similar in signal intensity) to normal marrow [1 2]. Focal lesions on the STIR images, because of the suppression of the fat signal, appeared bright and were easily differentiated from the red and yellow marrow [1 2-1 4] (Fig. 2 and 3). In 41 (39%) of the 1 06 patients, no tumor involvement of the bone was detected in the regions of interest covered by both techniques. In 32 (30%) patients, both techniques had positive findings and showed similar anatomic distribution of the tumor involvement in the bone. MR imaging showed areas of presumed metastatic tumor involvement in the bone not observed on scintigraphy in 30 (28%) of the 1 06 patients. The distribution of the number of patients per tumor type is as follows: eight patients with breast cancer, six patients with various sarcomas, six patients with lymphoma, four patients with renal cell carcinoma, four patients with Ewing sarcoma, and one each with a germ cell tumor and leukemia. In 16 (53%) of these 30 cases in which MR imaging findings were positive and scintigraphic findings were negative, MR imaging was performed 1 -35 days after scintigraphy either to resolve a focal clinical problem (i.e., pain, equivocal findings) or for the evaluation of cord compression. Twenty-one (7O%) of the 30 patients in whom MR imaging findings were positive and scintigraphic findings were negative had follow-up imaging studies as a means of confirmation 0-6 months after the comparative studies (Fig. 1). Four of the 21 patients had biopsy-proved malignant tumors. In nine of these 30 patients, findings on scintigraphy were normal, whereas MR images showed focal areas of abnormal signal intensity that were presumed to be tumors; pathologic confirmation was not available. In almost all 30 cases, MR imaging was performed in the coronal or sagittal plane, which efficiently encompassed the greatest extent of bone at risk on a single image. Scintigraphic findings were positive in two cases (Ewing sarcoma, colonic carcinoma) and indeterminate in one case (lymphoma) in which MR imaging findings were negative. Only one of these three cases (colonic carcinoma) was subsequently proved to be a metastatic lesion (Fig. 4) on the basis of follow-up imaging studies. The other two cases were considered clinically and radiologically compatible with stress or compression fractures.
AJR:155,
November
MR
1990
AND
SCINTIGRAPHY
OF
BONE
TUMORS
104
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
-
.
..
..
.
II A
Fig. 1.-l 1-year-old boy initially presenting with a primary Ewing sarcoma of fibula. MR imaging showed another focus of tumor that was observed on scintigram obtained 3 months after initial examination. A, Scintigram shows a solitary metastasis to right femoral diaphysis. B, Ti-weighted SE MR image, 300/26, obtained I 1 days after scintigraphy clearly shows a smaller satellite lesion (arrow) and tumor in contralateral femur. C, Scintigram obtained 3 months after initial scintigraphy shows second focus of tumor. Patient later died of diffuse metastatic Ewing sarcoma.
Fig. 2.-46-year-old
man with renal
cell carcinoma. MR imaging has greater sensitivity for metastases than scintigraphy does. A and B, MR imaging performed as part of staging evaluation reveals vertebral lesion (arrow) on STIR (IR 1500/ 100) (A) and Ti-weighted SE 433/26 (B) MR images. C, Posterior scintigram obtained 1 day after MR imaging is normal.
The two-sided McNemar test demonstrated that the overall positive rates for a case-by-case analysis of MR imaging differed from scintigraphy at p < .001 . The advantage of the McNemar test in the analysis of matched comparisons is its sole dependence on discordant cases (i.e., MR imaging findings positive, scintigraphic findings negative or MR imaging findings negative, scintigraphic findings positive). If we disre-
gard the nine (30%) of 30 unconfirmed (either by biopsy or follow-up examination) cases in which MR imaging findings were positive and scintigraphic findings were negative, the relative sensitivity of MR imaging differed from that of scintigraphy at p < .003. As a worst case, MR imaging was shown to have a higher sensitivity than scintigraphy did for all bony lesions in patients at risk of metastatic disease.
1046
FRANK
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
Discussion MR imaging has been shown to be sensitive for the detection of bone marrow metastases [5-1 0, 1 2, 1 3-1 6]. Normal red and yellow marrow can be distinguished on an MR image because of differences in signal intensity or morphology [16,
-
Fig. 3.-58-year-old man with nodular poorly differentiated lymphoma. MR imaging was performed as a basis for monitoring the patient’s response to therapy. A, Coronal STIR MR image obtained for evaluation of retroperitoneal adenopathy clearly shows multiple focal thoracolumbar vertebral lesions (arrows). B, Posterior scintigram of spine obtained 2 days later is unremarkable. Rib lesion, not seen on MR, is evident.
El
AL.
AJR:155,
November
1990
17]. A significant variety of diseased tissues (e.g., tumor, inflammation, and infection) have relaxation times that differ from normal marrow to a similar degree, resulting in similar appearances on MR images [1 4, 1 8-20] and a lack of histologic specificity. Scintigraphy is a sensitive but not specific imaging technique for the detection of malignant tumors because benign conditions such as degenerative joint disease, trauma, and infection can result in positive test results. The lack of histologic specificity stems from the mechanisms of isotope uptake (“hot spot”) or lack of tracer uptake (photopenic) areas on the bone scan. Hot spots predominately reflect the pattern of bone response to a local insult, and focal decreases in skeletal tracer uptake result from bone destruction without repair [21]. We found MR imaging to be more sensitive than scintigraphy was for the detection and delineation of bone tumor. A possible explanation for MR imaging’s superiority is that hematogenously seeded intramedullary metastasis produces detectable lesions by replacement of marrow. This occurs before either intrinsic or reactive metabolic changes in cancellous and cortical bone can be detected scintigraphically or radiologically. Usually, destruction of bony trabeculae and mobilization of osteoclastic reparative mechanisms are prerequisites for positive findings on a scintigram [21]. In a preliminary study examining the extent of bone marrow involvement in patients with small-cell lung carcinoma, 10 (42%) of 24 patients who were considered to have limited disease by standard staging criteria (including scintigraphy) had their stage of disease changed to extensive disease after MR imaging [9]. In these cases, focal or diffuse abnormalities in signal intensity were noted on Ti -weighted images of the pelvis and proximal femur not observed on bone scans. In a study of 50 patients with malignant tumors and suspected skeletal involvement, MR imaging was shown to have a lower false-positive rate than scintigraphy and 1 00% sensitivity and specificity in patients with multiple myeloma with marrow
Fig. 4.-44-year-old man with metastatic adenocarcinorna of colon. This is only before MR imaging. A, STIR MR image fails to reveal posterior acetabulum lesion. B, Scintigram made I month before MR image shows lesion (arrow). C, MR image obtained 4 months after scintigraphy shows lesion (arrow).
case
in this
series
in which
scintigraphy
showed
a malignant
lesion
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
AJR:155,
November
1990
MR
AND
SCINTIGRAPHY
involvement [6]. There are also reports of patients with lymphoma in whom MR imaging showed marrow involvement despite a normal iliac crest bone marrow biopsy [7, 8]. These studies support our results. It could be argued that during the 36-day (maximum) interval between scintigraphy and MR imaging, a new malignant lesion could develop, which would result in positive findings for one scanning technique while the other showed negative findings. In addition, delay between the two studies might introduce a selection bias. In only nine of the 31 malignant cases in which the MR imaging findings differed from scintigraphy was there a delay of more than 20 days between the two examinations. Deleting these patients from the statistical analysis would not change the overall results of this study (p < .003). If an MR imaging examination was ordered or not obtained on the basis of results of a scintigram, then it is possible that the number of cases of positive scintigraphic/ negative MR imaging findings reported may underestimate the true incidence offalse-negative MR imaging examinations. In 16 (53%) of the 30 cases with negative scintigraphic findings and positive MR imaging findings, MR imaging was performed after scintigraphy. If these studies were excluded from the analysis, the sensitivity of MR imaging would differ from that of bone scanning at p < .01. Under ideal circumstances, a comparison study would relate results of two imaging techniques to the true condition of a patient at every lesion described by either test. Unfortunately, although histology is a universally accepted estimate of “truth,” percutaneous biopsy of even a single site, much less at every one of multiple sites, has not been possible in our population of patients. Noninvasive assessment through follow-up imaging studies, although frequently available in our series, is burdened by ambiguity-if a marrow lesion disappears during therapy, did it represent a metastasis or a focal site of “reconverted” red marrow responding to cytotoxic drugs or radiation? Observing progression toward a subsequently unambiguous state without treatment is universally feasible only in an animal study. It is important to note that red marrow can have an appearance on MR images similar to focal or diffuse tumor involvement depending on lesion location and age of the patient. In patients with aplastic anemia, focal islands of hematopoietic marrow can have signal-intensity pattern and distribution mimicking metastatic disease [22]. We encountered insufficient numbers of patients with predominantly red marrow to establish whether our overall results would apply to this group [1 7]. Distinction between hyperplastic red marrow, particularly when focally deposited, and neoplasm may be difficult with MR imaging. Although only additional study of such patients can resolve this issue, behavior similar to that exhibited by metastases in response to therapy, that is, either replacement with fatty marrow [23] or overpopulation with hematopoietic elements (i.e., stress) after chemotherapy [24], also may complicate both experimental design and clinical diagnosis. The high sensitivity of MR imaging in relation to that of scintigraphy by itself does not alter scintigraphy’s role as the most useful screening examination for primary and metastatic disease. Scintigraphy examines 1 00% of the skeleton, is
OF
BONE
TUMORS
1047
widely available, and at present is less expensive than MR imaging. The use of MR imaging to evaluate skeletal metastases is limited by its expense, lesser availability, and regional purview. Lesions of the ribs and skull were excluded from this study principally because so few images included them. However, rib metastases might prove difficult to show on MR images because of respiratory motion, paucity of at-risk tissue (red marrow), or lack of a convenient imaging plane (Fig. 3). Metastasis to the skull would normally be included on a separate MR imaging examination of the brain; these were not reviewed as part of this study. Although MR imaging as currently practiced is not an efficient technique for screening the entire skeleton, we find it extremely valuable in clarifying equivocal or (suspected) falsely positive or negative bone scans. Because of MR imaging’s inherent anatomic detail, we also may find it useful as a localization technique for directed biopsies of suspected lesions [5-8]. Future improvement in MR imaging techniques (e.g., development of faster scanning techniques or surface coils) may make MR imaging an efficient screening technique for detecting and staging bone marrow metastasis.
ACKNOWLEDGMENTS The authors thank Anna Scheib for her preparation of the manuscript. In addition, the authors acknowledge the work of the MR
technologists
who performed
the studies.
REFERENCES 1 . Volger JB, Murphy WA. Bone marrow imaging. Radiology 679-693 2. Kagan AR, Bassett LW, Gold RH. Radiologic contributions AJR 1986;147:305-312 3. McNeil BJ. Value of bone scanning 1984;14:277-286
i988;168:
to cancer
management.
4. Krishnamurthy metastatic
bone
in neoplastic
disease. Semin Nucl Med
GT, Tubis M, Hiss J, Bland WH. Distribution pattern of disease: a need for total body skeletal image. JAMA
1977;237:2504-2506
5. Khurana JS, Rosenthal Dl, Rosenberg AE, Mankia HJ. Skeletal metastases in liposarcoma detectable only by magnetic resonance imaging. Clin Orthop 1989;243:204-207 6. Daffner R, Lupetin A, Dash N, Beeb Z, Sefczek R, Shapiro A. MRI in the detection of malignant infiltration of bone marrow. AiR 1986;146:353-358 7. Dohner H, Guckel F, Knauf W, et al. Magnetic resonance imaging of bone marrow in lymphoproliferative disorders: correlation with bone marrow biopsy. Br J Haematol 1989;73: 12-17 8. Linden A, Zankovich R, Theissen P. Diehl V. Schicha H. Malignant lymphoma: bone marrow imaging versus biopsy. Radiology i989;173:
335-339 9. Carney DN, Redmond
0, Harford P, Stark J, Ennis J. Bone marrow involvement (BMI) by small cell lung cancer (SCLC) using magnetic resonance imaging. Proc Am Soc Clin Oncol 1989;8:228 10. Dwyer AJ, Frank JA, Sank VJ, et al. Short TI inversion recovery pulse sequence: analysis and initial experience in cancer imaging. Radiology 1988;168:827-836 1 1 . Conover WJ. Practical nonparametric statistics, 2nd ed. New York: Wiley, 1980:130 12. Porter BA, Shields AF, Olson DO. Magnetic resonance marrow disorders. Radiol Clin North Am 1986;24:269-289
imaging of bone
13. Shields AF, Porter BA, Churchley 5, et al. The detection of bone marrow involvement by lymphoma using magnetic resonance imaging. J Clin Oncol 1987;5:225-230
14. Frank JA, Dwyer AJ, Doppman JL. Magnetic resonance imaging in oncology. In: DeVita VT, Hellman 5, Rosenberg SA, eds. important advances in
FRANK
1048
oncology
1987.
Philadelphia:
Lippincott,
i987:
133-1
Invest Radiol 1988;23: 193-1 99 Smoker WRK, Godersky JC, Knutzon RK, Keyes WD, Norman
Downloaded from www.ajronline.org by 117.253.106.10 on 11/08/15 from IP address 117.253.106.10. Copyright ARRS. For personal use only; all rights reserved
W. The role of MR imaging in evaluating
metastatic
D, Bergman spinal disease. AJR
i987;149:1241-1248
17. Kricun ME. Red-yellow solitary
bone lesions.
18. Unger E, Moldofsky osteomyelitis
marrow
its eftect on the location of Skeletal Radiol 1985;1 4:10-19 P, Gatenby R, Hartz W, Broden G. Diagnosis of
by MR imaging.
conversion:
Radiology
1987;1 64:449-454
19. Mitchell DG, Rao VM, Dalinku MK, et al. Femoral head avascular necrosis: correlation
of MR imaging,
radiographic
staging,
AL.
radionuclide
imaging
and
AJR:155, November 1990
20.
clinical finding. Radiology 1987;162:709-715 Medic MT. Masaryk TJ, Paushter D. Magnetic spine. Radiol Clin North Am 1986;24:229-245
21.
Charkes
74
15. Fruehwald FJ, Tscholakoff D, Schwaighofer B, et al. Magnetic resonance imaging of the lower vertebral column in patients with multiple myeloma. 1 6.
El
ND.
Mechanisms
of
skeletal
resonance
tracer
uptake.
imaging J
Nucl
of the Med
1979;20:794-795 22. Kaplan PA, Asleson RJ, Klassen LW, Duggan MJ. Bone marrow pattems in aplastic anemia: 1987;1 64:441 -444
observations
with
1 ST
MR
imaging.
Radiology
23. Ramsey RG, Zacharias CE. MR imaging of the spine after radiation therapy: easily
recognizable
effects.
AJR 1985;144:
1131-1135
24. McKinstry CS, Steiner RE, Young AT, Jones L, Swirsky D, Aber V. Bone marrow in leukemia and aplastic anemia: MR imaging before, during, and after treatment. Radiology i987;162:701-707
