| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Musculoskeletal Imaging |
1 From the Departments of Radiology (M.R.S., J.H., C.A.B., M.Z.) and Orthopedic Surgery (P.V.), Orthopedic University Hospital Balgrist, Forchstrasse 340, 8008 Zurich, Switzerland. From the 2000 RSNA scientific assembly. Received July 23, 2001; revision requested September 12; revision received November 15; accepted January 9, 2002. Address correspondence to M.R.S. (e-mail: mariusschmid@hotmail.com).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Fifty-one consecutive patients with bone marrow abnormalities depicted on turbo STIR images were examined with additional T1-weighted contrast-enhanced (0.1 mmol/kg gadopentetate dimeglumine) MR imaging with fat suppression. Volume and signal difference–to-noise ratio (SDNR) were measured. An additional qualitative analysis was performed by two experienced musculoskeletal radiologists to correlate the presence or absence of ill-defined edema-like zones, well-defined zones, and cystlike zones. Diagnoses determined with MR findings with each sequence were compared with the results of a review panel. Correlation coefficients (r2) and paired t tests were calculated for all measurements. Agreement percentages and 
values were calculated for inter- and intraobserver reproducibility.
RESULTS: Regarding volume of bone marrow abnormalities, a high correlation (r2 = 0.98) of both sequences was found. SDNR was substantially higher on T1-weighted contrast-enhanced images than on STIR images (mean, 125.9 vs 95.4; P < .001). The qualitative analysis demonstrated identical imaging patterns with both sequences in 96% (79 of 82, = 0.38) of ill-defined zones, in 88% (72 of 82, = 0.76) of well-defined zones, and in 98% (80 of 82, = 0.84) of cystlike zones. Interobserver reproducibility of the three imaging patterns was similar for both sequences. The values for these three zones with STIR sequence were 0.55, 0.68, and 0.69, and those for the T1-weighted contrast-enhanced MR sequence were 0.49, 0.73, and 0.58, respectively. Diagnoses determined with MR findings were equal with both sequences in 94% (80 of 85) of involved bones.
CONCLUSION: STIR images and T1-weighted contrast-enhanced fat-suppressed MR images demonstrate almost identical imaging patterns, and diagnoses determined with these findings show little difference.
© RSNA, 2002
Index terms: Ankle, abnormalities, 46.219, 46.833 • Ankle, MR, 46.121413, 46.121415, 46.12143 • Bone marrow, abnormalities, 46.219, 46.833 • Foot, abnormalities, 46.219, 46.833 • Magnetic resonance (MR), contrast enhancement, 46.121415, 46.12143 • Magnetic resonance (MR), inversion recovery, 46.121413
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
However, in some cases the cause of bone marrow abnormalities remains unknown. To improve the specificity of the MR examination, radiologists may perform additional T1-weighted contrast material–enhanced sequences with fat suppression. During clinical routine work, we observed cases in which contrast-enhanced fat-suppressed images did not differ from STIR images in depiction of bone marrow abnormalities of the foot.
Although the reasons for signal intensity alterations are not identical for the two types of sequences, the use of contrast-enhanced MR images may not add relevant information and, therefore, may be less valuable than commonly expected.
To our knowledge, the value of T1-weighted contrast-enhanced fat-suppressed SE MR images in comparison with STIR images for depiction of bone marrow abnormalities in the foot and ankle has not been formally evaluated. The purpose of this investigation was to compare STIR and T1-weighted contrast-enhanced fat-suppressed SE sequences for depiction of bone marrow abnormalities of the foot and ankle.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
All examinations were performed with a 1.0-T MR system (Expert; Siemens Medical Systems, Erlangen, Germany). A transmit-receive extremity coil was used. Examinations of the forefoot (n = 7) were performed with the patient in the prone position with plantar flexion of the foot (12). The ankle and hindfoot (n = 44) were examined with the patient in the supine position with dorsiflexion of the foot. The imaging protocol was chosen in a standardized fashion on the basis of the referring physician’s question regarding diagnosis of the patient’s problem and of findings of previously performed radiologic examinations. In general, T1-weighted SE (repetition time msec/effective echo time msec, 500–720/15–20) and T2-weighted turbo SE (repetition time msec/echo time msec, 4,000–4,500/96) imaging was performed in two planes.
In addition, a turbo SE STIR sequence (repetition time msec/echo time msec/inversion time msec, 4,800/30[effective]/150; echo train length, seven; field of view, 180 x 180–220 x 220 mm; matrix size, 256 x 256; section thickness, 3–4 mm; and intersection gap, 0.3–0.8 mm) was performed in all patients. With two acquisitions, acquisition time for the turbo STIR sequence varied between 4 minutes 53 seconds and 5 minutes 51 seconds. For the ankle and hindfoot, images were obtained in the sagittal plane, and for the mid- and forefoot, they were obtained in the transverse plane. A T1-weighted SE sequence (500-800/20 [effective]) with fat suppression was performed after intravenous administration of 0.1 mmol per kilogram of body weight of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany). The imaging plane, section thickness (3–4 mm), intersection gap (0.3–0.8 mm), field of view (180 x 180–220 x 220 mm), and image matrix (256 x 256) were identical to those used with the turbo STIR sequence. With three acquisitions, acquisition time for the fat-suppressed T1-weighted sequence varied between 4 minutes 25 seconds and 4 minutes 51 seconds.
Image Analysis
The signal difference–to-noise ratio (SDNR) (13) of increased bone marrow signal was measured for both sequences as follows: A region of interest (ROI) was placed within the zone of increased bone marrow signal and in the nearest most normal-appearing part of the bone marrow. Area of ROI within the bone marrow varied between 0.1 and 0.2 cm2 for both sequences. Size of ROI within the area of increased bone marrow signal and size of ROI within normal bone marrow were identical. The difference between the mean signal intensities then was divided by the SD of the signal intensity measured outside the foot (noise, ROI = 1 cm2).
Volumes of the zones of increased bone marrow signal were determined by means of measurement of cross-sectional areas within every section and multiplication by the section thickness plus intersection gap. Both measurements were performed manually by one radiologist (M.R.S.) with the standard MR console.
An additional qualitative analysis was performed by two musculoskeletal radiologists (M.Z., J.H.) with 8 years of experience in this subspecialty. Both readers were blinded to the clinical data. They were to determine one of 13 possible diagnoses: arthritis, diabetic osteoarthropathy, plantar fasciitis, fracture, insufficiency fracture, intraosseous ganglion, osteochondral lesion, osteomyelitis, osteonecrosis, bone infarct, benign bone tumor, malignant bone tumor, and nonspecific bone marrow changes. Two readings were performed. First, both readers analyzed the standard SE sequences together with the STIR sequence. After a minimum of 3 weeks, a second session was performed. During this second session, the standard SE sequences were reviewed together with the T1-weighted contrast-enhanced fat-suppressed SE sequence. The standard of reference was provided by a review panel including one radiologist (M.R.S.) and an orthopedic surgeon (P.V.). They analyzed clinical data and findings of MR examinations. Surgery reports were available for all cases with the final diagnosis of osteomyelitis and bone tumors.
In addition to the final diagnosis, both readers identified one of the following image patterns (14) with both sets of sequences: (a) ill-defined edema-like zones, (b) well-defined zones, and (c) cystlike zones.
Statistical Analysis
Software (StatView; SAS Institute, Cary, NC) was used for calculation of P values (paired t test) and correlation coefficients (r2) of volume and SDNR of signal alterations. To analyze inter- and intraobserver reliability, values (15) and agreement percentages were calculated.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Mean SDNR of increased bone marrow signal intensity was significantly higher (P < .001) with the T1-weighted contrast-enhanced fat-suppressed sequence than it was with the STIR sequence (125.9 vs 95.4). With the T1-weighted contrast-enhanced fat-suppressed sequence, a higher SDNR was demonstrated in 43 bones; with the STIR sequence, a higher SDNR was demonstrated in 39 bones.
The mean volume of bone marrow signal intensity abnormalities (Table 1) was slightly greater with the STIR sequence (mean, 8.75 mL; SD, 10.7) than with the T1-weighted contrast-enhanced fat-suppressed sequence (mean, 8.29 mL; SD, 9.9). A high correlation (r2 = 0.98) was observed between the volumes of abnormal bone marrow with both sequences.
|
values for the comparison of the sequences are listed in Table 2. Interobserver reproducibility in the evaluation of ill- and well-defined lesions was almost identical for both sequences, with a slightly higher value with the STIR sequence ( = 0.69) compared with that for the T1-weighted contrast-enhanced fat-suppressed sequence ( = 0.58) in the evaluation of cystlike lesions (Table 3).
|
|
|
|
|
|
|
|
|
|
Three other diagnoses (ie, malignant bone tumor, plantar fasciitis, and fractures related to acute trauma) in our previously mentioned list of possible diagnoses were not determined in our study population. Intraobserver agreement with both sequences was 94% (80 of 85, reader 1) and 92% (78 of 85, reader 2). Compared with the findings of the review panel, agreement of the STIR and T1-weighted contrast-enhanced fat-suppressed sequences for reader 1 was 84% (71 of 85) and 87% (74 of 85), respectively. Agreement between findings of the review panel and reader 2 was 78% (66 of 85) for the STIR sequence and 79% (67 of 85) for the T1-weighted contrast-enhanced fat-suppressed sequence. Interobserver agreement of both readers was 79% (67 of 85) for the STIR sequence and 82% (70 of 85) for the T1-weighted contrast-enhanced fat-suppressed sequence.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Interobserver and intraobserver agreement in the depiction of the three image patterns were comparable for both sequences (high percentages for agreement and fair to high values), and this comparable agreement indicated that one sequence did not have an advantage over the other. Low values of both readers (0.38,reader 1; 0.31, reader 2) in the comparison of depiction of ill-defined bone marrow changes were not contrary to the high percentages for agreement, because in almost all cases ill-defined marrow abnormalities were depicted with both sequences, and in such unequally distributed cases, values are always less reliable.
In our patient group, no additional lesion in the affected bones or in another bone was depicted with the T1-weighted contrast-enhanced fat-suppressed sequence compared with the STIR sequence. Almost identical diagnostic agreement between the findings of the review panel and the findings with both sequences was observed for both readers. In the few cases of osteomyelitis (n = 6) and diabetic osteoarthropathy (n = 9), identical diagnoses were determined by both readers with both sequences. We must admit, however, that none of these cases was accompanied by abscess formation. Both malignant primary bone tumors and metastases in the foot are rare, and this rarity is the reason that we did not determine such diagnoses. Three benign bone tumors (two cases of osteoid osteoma and one of enchondroma) and no malignant bone tumors were observed in our study group.
Therefore, our conclusions regarding limited efficacy of contrast-enhanced imaging cannot be extended to this group of abnormalities. Findings in the investigation of May et al (16) indicate advantages of additional contrast-enhanced images in cases of malignant bone tumors because of optimization of biopsy planning and better detection of recurrent malignancy. However, in this investigation, specificity was influenced only to a minor degree. Advantages of obtaining T1-weighted fat-suppressed images after intravenous contrast material administration were observed by Gronemeyer et al (17) and Iwasawa et al (18) in cases of malignant bone tumors with regard to better delineation of neurovascular bundle infiltration (17), identification of tumor necrosis (17), and detectability of microscopic intraosseous tumor invasion (18). Therefore, we believe that T1-weighted contrast-enhanced images must be obtained in cases of malignant tumors in the foot and ankle.
In many patients with bone marrow abnormalities, the final diagnosis can be determined regardless of the precise sequence used. Osteochondral lesions are characterized by chondral defects related to a subchondral bone marrow abnormality. Cystic lesions within the talus often occur in cases of osteochondral defects (19) but also are observed in cases of osteoarthritis and osteonecrosis and as isolated bone cysts. In insufficiency fractures, a hypointense fracture line may be observed within the abnormality (5). Osteonecrosis can occur in the foot and ankle and in other locations, such as the femoral head (7), the femoral condyles (20), or the wrist (21). In characteristic cases, there is a demarcated zone of signal abnormalities within the more diffuse edema-like zones, with or without subchondral fractures. In cases of osteoarthritis, ill-defined bone marrow abnormalities occur adjacent to the affected joint (14). In such cases with well-defined signal alterations (eg, a fracture line) or in cases with cystic bone marrow changes, it is much easier to determine a distinct diagnosis.
In many cases, however, the diagnosis cannot be determined by using findings of examinations, including findings of T1- and T2-weighted SE and STIR imaging. Some of these cases include early phases of avascular necrosis and insufficiency fractures. If only one side of a joint is affected in osteoarthritis, it might be difficult to differentiate this degenerative cause of bone marrow abnormality from other conditions, such as early insufficiency fracture (before a distinct fracture line occurs), overuse (10), or bone marrow abnormalities caused by altered biomechanics (9). In such equivocal cases, many radiologists attempt to improve the value of the examination with additional contrast-enhanced sequences, although little published evidence supports this concept.
The term "bone marrow edema" is widely used (4,9,10,22,23) to describe morphologic changes with ill-defined hypointense zones on T1-weighted SE and hyperintense zones on STIR or T2-weighted fat-suppressed images. The fact that STIR images and T1-weighted contrast-enhanced fat-suppressed images show an almost identical image pattern indicates that increased signal within bone marrow cannot completely be explained by the presence of bone marrow edema alone. Zanetti et al (14) demonstrated that bone marrow signal intensity alterations associated with osteoarthritis of the knee are caused not only by edema but also more prominently by necrosis, fibrosis, hemorrhage, and trabecular abnormalities. In their study, marrow edema was rarely an isolated finding. Similar histologic findings with necrosis of fat cells and hematopoietic bone marrow were observed by Plenk et al (24).
Because of the prospective nature of the study, with inclusion of consecutive cases, there is a mixture of many different diagnoses, some of which are represented by only one case each. This limitation is most relevant with regard to both benign and malignant bone tumors and infections, and in diagnosis of these cases, intravenous administration of gadolinium-based contrast agents may be important because enhancement helps to differentiate necrosis or abscess from vascularized tissue. Findings in this study mainly refer to nonspecific bone marrow abnormalities (probably caused by overuse or early cases of osteonecrosis or insufficiency fractures), and for these cases, radiologists tend to perform additional contrast-enhanced imaging.
Findings in this study demonstrated that, for depiction of many bone marrow abnormalities in the foot and ankle, STIR and T1-weighted contrast-enhanced fat-suppressed MR sequences produce an almost identical image pattern. We believe that only one of the two sequences must be performed in cases such as those included in this investigation. Because the cost is lower and no intravenous injection is required, we prefer the use of the STIR sequence.
|
FOOTNOTES |
|---|
Abbreviations: ROI = region of interest, SDNR = signal difference–to-noise ratio, SE = spin echo, STIR = short inversion time inversion recovery
Author contributions: Guarantors of integrity of entire study, M.R.S., J.H., M.Z.; study concepts, M.R.S., C.A.B., M.Z.; study design, M.Z., M.R.S., P.V.; literature research, M.Z., M.R.S.; clinical studies, M.R.S., P.V.; data acquisition, M.R.S., C.A.B., M.Z.; data analysis/interpretation, M.R.S., P.V., J.H., M.Z.; statistical analysis, M.R.S., J.H.; manuscript preparation, M.R.S., M.Z.; manuscript definition of intellectual content, M.R.S., P.V., M.Z.; manuscript editing and revision/review, all authors; manuscript final version approval, M.R.S., J.H., M.Z.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Schoellnast, H. A. Deutschmann, J. Hermann, G. J. Schaffler, P. Reittner, F. Kammerhuber, D. H. Szolar, and K. W. Preidler Psoriatic arthritis and rheumatoid arthritis: findings in contrast-enhanced MRI. Am. J. Roentgenol., August 1, 2006; 187(2): 351 - 357. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. Fleming II, L. Dodd, and C. A. Helms Prominent Vascular Remnants in the Calcaneus Simulating a Lesion on MRI of the Ankle: Findings in 67 Patients with Cadaveric Correlation Am. J. Roentgenol., December 1, 2005; 185(6): 1449 - 1452. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. S. Collins, M. M. Schaar, D. E. Wenger, and J. N. Mandrekar T1-Weighted MRI Characteristics of Pedal Osteomyelitis Am. J. Roentgenol., August 1, 2005; 185(2): 386 - 393. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. Mayerhoefer, M. Breitenseher, S. Hofmann, N. Aigner, R. Meizer, H. Siedentop, and J. Kramer Computer-Assisted Quantitative Analysis of Bone Marrow Edema of the Knee: Initial Experience with a New Method Am. J. Roentgenol., June 1, 2004; 182(6): 1399 - 1403. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Bloem, T. J. Mosher, M. R. Schmid, M. Zanetti, H. P. Ledermann, W. B. Morrison, and M. E. Schweitzer Dynamic Gadolinium-enhanced MR Imaging in Bone Marrow Disorders [letter] * Dr Mosher responds: * Drs Schmid and Zanetti respond: * Dr Ledermann and colleagues respond: Radiology, April 1, 2003; 227(1): 303 - 305. [Full Text] [PDF]
|
|
|
|
|
|
T. J. Mosher Diagnostic Effectiveness of Gadolinium-enhanced MR Imaging in Evaluation of Abnormal Bone Marrow Signal Radiology, August 1, 2002; 224(2): 320 - 322. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
